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

== 2.5 Projected Impacts and Risk for Species, Communities, Biomes, Key Ecosystems and Their Services ==

<div id="h1-6-siblings" class="h1-siblings"></div>

Under the risk assessment framework that was introduced in AR5 ( [[#IPCC--2014b|IPCC, 2014b]] ), risk means the probability of harmful consequences resulting from climate change. It results from the interaction of vulnerability, exposure and hazard and can be represented as the probability of occurrence of hazardous events or trends multiplied by the impacts if these events or trends occur (see Chapter 1, this report). The framework defines vulnerability as a pre-existing condition, incorporating the extent to which species or ecosystems are susceptible to, or unable to cope with, the adverse effects of climate change. Vulnerable species have limited adaptive capacity, stemming from physiological and behavioural constraints, limited dispersal abilities and restricted resource requirements or capacities for distributional and genetic changes ( [[#Foden--2013|Foden et al., 2013]] ; [[#Cizauskas--2017|Cizauskas et al., 2017]] ; [[#Foden--2019|Foden et al., 2019]] ). Traits that render entire ecosystems vulnerable are harder to define, but it is clear that vulnerabilities are high in the coldest habitats, in those with limited geographic ranges such as low-lying islands and in specialised, restricted habitats such as serpentine outcrops in California ( [[#Anacker--2012|Anacker and Harrison, 2012]] ) and dry meadows in Fennoscandia and Tibet ( [[#Yang--2018a|Yang et al., 2018a]] ). Ecosystem vulnerability can depend critically on the fates of plants that function as ‘foundation species’, providing community biomass above and below the ground, structuring habitat for fauna and providing ecosystem services such as erosion control ( [[#Camac--2021|Camac et al., 2021]] ).

<div id="2.5.1" class="h2-container"></div>

<span id="projected-changes-at-species-and-community-levels"></span>
=== 2.5.1 Projected Changes at Species and Community Levels ===

<div id="h2-12-siblings" class="h2-siblings"></div>

<div id="2.5.1.1" class="h3-container"></div>

<span id="assessment-of-models-and-sources-of-uncertainties"></span>
==== 2.5.1.1 Assessment of Models and Sources of Uncertainties ====

<div id="h3-29-siblings" class="h3-siblings"></div>

Methods for projecting the impacts of climate change on biodiversity can be classified into three types: (1) statistical models such as SDMs ( [[#Elith--2009|Elith and Leathwick, 2009]] ); (2) mechanistic or process-based models ( [[#Chuine--2017|Chuine and Régnière, 2017]] ) and (3) trait-based models ( [[#Pacifici--2015|Pacifici et al., 2015]] ). It is only recently that models have been developed looking at lower levels of warming like 1.5°C ( [[#Hoegh-Guldberg--2018|Hoegh-Guldberg et al., 2018]] ; [[#Warren--2018|Warren et al., 2018]] ).

SDMs or niche-based models assess potential geographic areas of suitable climate for the species in current conditions and then project them into future conditions ( [[#Trisurat--2018|Trisurat, 2018]] ; [[#Vieira--2018|Vieira et al., 2018]] ). There are limitations in all models and it is critical that modellers understand the assumptions, proper parameterization and limitations of each model technique, including differences between climate models, emission scenarios or RCPs and baselines ( [[#Araujo--2019|Araujo et al., 2019]] ). Several systems automate the development of SDMs, including R-packages ( [[#Beaumont--2016|Beaumont et al., 2016]] ; [[#Hallgren--2016|Hallgren et al., 2016]] ), and other model types ( [[#Foden--2019|Foden et al., 2019]] ) and aid in the use of climate model data ( [[#Suggitt--2017|Suggitt et al., 2017]] ), including allowing for connectivity constraints ( [[#Peterson--2013|Peterson et al., 2013]] ). [[#Buisson--2010|Buisson et al. (2010)]] found most variation in model outputs stems from differences in design, followed by general circulation models (GCMs).

Mechanistic approaches, also known as process-based models, project the responses of species to climate changes by explicitly incorporating known biological processes, thresholds and interactions ( [[#Morin--2009|Morin and Thuiller, 2009]] ; [[#Maino--2016|Maino et al., 2016]] ). Mechanistic models are able to accommodate a broad range of mechanisms of climate change impacts and include species-specific characteristics such as dispersal distances, longevity, fecundity, genetic evolution and phenotypic plasticity. However, sufficient knowledge is available for only a few well-studied species. Species’ traits have been used to more broadly estimate potential climate change impacts ( [[#Foden--2013|Foden et al., 2013]] ; [[#Cizauskas--2017|Cizauskas et al., 2017]] ).

Most models are on a large scale (20–50 km), and so cannot capture micro-climatic refugia generated by diversities of slope aspect, elevation or shade ( [[#Suggitt--2015|Suggitt et al., 2015]] ; [[#Suggitt--2018|Suggitt et al., 2018]] ). In analysing records of 430 climate-threatened and range-declining species in England, (Suggett et al., 2015; [[#Suggitt--2018|Suggitt et al., 2018]] ) showed that topographic diversity reduced population declines most strongly in areas experiencing the most local warming and in the species most sensitive to warming. Under these circumstances, topographic diversity reduced the risk of population extinction by 22% for plants and 9% for insects.

None of the modelling techniques are predictions of the future, they are rather projections of possible futures. To date, only a few studies have validated model performance against observations, but the studies that have been conducted do generally validate models using either SDMs or process-based models ( [[#Johnston--2013|Johnston et al., 2013]] ; [[#Fordham--2018|Fordham et al., 2018]] ). SDMs should be considered as hypotheses of what a future world might look like if the climate projections came to pass. Suggestions have been made on how to start bringing more biotic interactions into SDMs ( [[#Early--2019|Early and Keith, 2019]] ), but limited basic ecological understanding of interactions, along with limits on computation and funding, constrains how far and how fast these modelling techniques can advance.

'''Table 2.3 |''' Assessing uncertainty in detection and attribution of observed changes in terrestrial and freshwater species and ecosystems to climate change. The lines of evidence used to support given uncertainty statements, including confidence statements, of the attribution of key conclusions on observed biological changes to climate change and increased atmospheric CO2. Icons represent lines of evidence. This is a summary table that is fully detailed in Table SM2.1.

[[File:8059cc2fe5c8d787cd29275db8706dcf IPCC_AR6_WG2_Chapter2_Table_2_3_1.png]] [[File:4e685b2dc6ec250ec778d343ef8fd8ba IPCC_AR6_WG2_Chapter2_Table_2_3_2.png]] [[File:acf6972c018951e2b0845b7709bb2220 IPCC_AR6_WG2_Chapter2_Table_2_3_3.png]] [[File:1222a6aebf41ab38cd303336fab9b90f IPCC_AR6_WG2_Chapter2_Table_2_3_4.png]]

<div id="2.5.1.2 " class="h3-container"></div>

<span id="risk-assessment-and-non-modelling-approaches"></span>
==== 2.5.1.2 Risk Assessment and Non-Modelling Approaches ====

<div id="h3-30-siblings" class="h3-siblings"></div>

In order to add realism and reliability to risk assessments at the species and community levels, non-modelling approaches, based on known biological traits or processes as well as on expert opinion ( [[#Camac--2021|Camac et al., 2021]] ), are used to temper model outputs with ground-based validation. Trait-based assessment approaches use species’ biological characteristics as predictors of sensitivity, adaptive capacity and extinction risk due to climate change. Climate exposure can be estimated using GIS-based modelling, statistical programs or expert judgment ( [[#Chin--2010|Chin et al., 2010]] ). These trait-based approaches are widely applied to predict responses of biodiversity to climate change because they do not require modelling expertise or detailed distibutional data ( [[#Pacifici--2015|Pacifici et al., 2015]] ; [[#Willis--2015|Willis et al., 2015]] ). Most of these methods have not been independently validated and do not allow direct comparison of vulnerability and risk across taxonomic groups.

Some studies have combined two or three approaches for the assessment of climate change risk to biodiversity, in order to capture the advantages of each and avoid their limitations. [[#Warren--2013|Warren et al. (2013)]] used combinations of SDMs and trait-based approaches to estimate the proportions of species losing their climatically suitable ranges under the various future scenarios of climate and dispersal rate. Similarly, spatial projections of exposure to climate change were combined with traits to assess the vulnerability of sub-Saharan amphibians ( [[#Garcia--2014|Garcia et al., 2014]] ). [[#Laurance--2012|Laurance et al. (2012)]] combined 31 functional groups of species and 21 potential drivers of environmental change, in order to assess both the ecological integrity and threats to protected tropical areas on a global scale. [[#Keith--2014|Keith et al. (2014)]] used a combination of three approaches (SDMs–trait–mechanistic) to determine how long before extinction a species would become eligible for listing as threatened, based on the IUCN Red List criteria.

<div id="2.5.1.3" class="h3-container"></div>

<span id="risk-of-species-extinctions"></span>
==== 2.5.1.3 Risk of Species’ Extinctions ====

<div id="h3-31-siblings" class="h3-siblings"></div>

<div id="2.5.1.3.1" class="h4-container"></div>

<span id="overview-2"></span>
===== 2.5.1.3.1 Overview =====

<div id="h4-32-siblings" class="h4-siblings"></div>

This assessment of current findings is of studies across a range of taxa and modelling techniques. Extinction risk estimates whether or not a particular species may be at risk of extinction over the coming decades if climatic trends continue, and usually does not take into account other human-induced stressors (e.g., invasive species or pollution). It is not a prediction that a species will definitely become extinct because, even when complete loss of a species’ range is projected, the scale of the model cannot estimate persistence in very small-scale micro-climatic refugia (that can be on the order of metres in size) ( [[#Suggitt--2015|Suggitt et al., 2015]] ; [[#Suggitt--2018|Suggitt et al., 2018]] ). Individuals and populations can survive after the conditions for successful reproduction are gone, leading to a lagged decline, called ‘extinction debt’ (see section 2.4.2.8) ( [[#Alexander--2018|Alexander et al., 2018]] ). Therefore, range loss is an established criterion for assessing endangerment status and risk of extinction. As a species range becomes smaller and occupied habitats become more isolated, the likelihood of a single stochastic event causing extinction increases. It is this combination of projected loss of climatically suitable space and additional stressors (especially LULCC of critical habitat) that is expected to drive future extinctions.

The IUCN Red List Criteria ( [[#IUCN--2019|IUCN, 2019]] ) classifies a species as ‘critically endangered’ if it has suffered a range loss of ≥80%, with a resulting likelihood of extinction of &gt;50% in the near term (10–100 yrs, depending upon generation length). A species is classified as ‘endangered’ if it has suffered a range loss of ≥50%, with a resulting likelihood of extinction of &gt;20% in the near term (10–100 years). In this assessment, a species that is projected to become classified as ‘endangered’ is deemed to be at ''‘high risk’'' of extinction, and becoming classified as ‘critically endangered’ is deemed at ''‘very high risk’'' of extinction.

<div id="2.5.1.3.2" class="h4-container"></div>

<span id="projections-for-freshwater-biodiversity"></span>
===== 2.5.1.3.2 Projections for freshwater biodiversity =====

<div id="h4-33-siblings" class="h4-siblings"></div>

Because risk to freshwater species has been limited in past reports, this section provides details of freshwater risk. Lakes, rivers and freshwater wetlands cover approximately 7.7–9.1% of global land surface area; ( [[#Lehner--2008|Lehner et al., 2008]] ; [[#Fluet-Chouinard--2015|Fluet-Chouinard et al., 2015]] ; [[#Allen--2018|Allen and Pavelsky, 2018]] ) and hold 9.5% of the Earth’s described animals ( [[#Balian--2008|Balian et al., 2008]] ), with climate change indicated as a threat to 50–75% of fish ( [[#Xenopoulos--2005|Xenopoulos et al., 2005]] ; [[#Darwall--2015|Darwall and Freyhof, 2015]] ). Climate change is cited as a primary factor in species’ extinction risk due to changes in water temperatures, stream flow, loss of cold water habitat, increased variability of precipitation and increased disease risk from warming temperatures ( ''robust evidence'' , ''high agreement'' , ''high confidence'' ) ( [[#Knouft--2017|Knouft and Ficklin, 2017]] ; [[#Pletterbauer--2018|Pletterbauer et al., 2018]] ; [[#Jaric--2019|Jaric et al., 2019]] ; [[#Reid--2019|Reid et al., 2019]] ) adding to the stress of overexploitation and LULCC ( [[#Craig--2017|Craig et al., 2017]] ; [[#IPBES--2019|IPBES, 2019]] ).

Increased frequency of stream drying events, reducing hydrologic connectivity and limiting access of native fishes to spawning habitats is projected for RCP8.5 in Colorado, USA ( ''medium evidence'' , ''medium agreement'' ) ( [[#Jaeger--2014|Jaeger et al., 2014]] ). Cold-water habitats and associated obligate species are particularly vulnerable, and losses in these habitats have been both documented and projected, for example, in salmonids ( [[#Santiago--2016|Santiago et al., 2016]] ; [[#Fullerton--2017|Fullerton et al., 2017]] ; [[#Merriam--2017|Merriam et al., 2017]] ). River networks are projected to lose connections to cold tributary refugia, that are important thermal refuges for cold water species ( ''robust evidence'' , ''high agreement'' ) ( [[#Isaak--2016|Isaak et al., 2016]] ) during low flows ( [[#Merriam--2017|Merriam et al., 2017]] ).

Community turnovers are expected in freshwaters as cold-adapted species lose and warm-adapted species gain climatically suitable habitat ( [[#Domisch--2011|Domisch et al., 2011]] ; [[#Domisch--2013|Domisch et al., 2013]] ; [[#Shah--2014|Shah et al., 2014]] ). While a number of warm-adapted species may experience range expansions, the majority of species are predicted to lose climatically suitable areas by, on average, 38–44%, depending on the emission scenario (A2a and B2a) ( ''medium evidence'' ) ( [[#Domisch--2013|Domisch et al., 2013]] ).

Molluscs are projected to be the most at-risk group, given their limited dispersal capability ( [[#Woodward--2010|Woodward et al., 2010]] ). Mediterranean freshwater fish are especially susceptible to climate change due to increasing flood and drought events and the risk of surpassing critical temperature thresholds ( [[#Santiago--2016|Santiago et al., 2016]] ; [[#Jaric--2019|Jaric et al., 2019]] ). In southern Europe, aquatic insects (Ephemeroptera, Plecoptera and Trichoptera) are endangered by climate change ( [[#Conti--2014|Conti et al., 2014]] ). European protected areas are not expected to be sufficient under warming to provide habitat for the majority of rare molluscs and fish ( [[#Markovic--2014|Markovic et al., 2014]] ). Observed trends agree with model projections in direction, but magnitude remains uncertain ( ''medium evidence'' , ''medium agreement'' , ''medium confidence'' ) (see Figure 2.8 for extinction risk globally for dragonflies, amphibians and turtles).

Regional threats from climate change have been reported for 40% of amphibians in China, ( [[#Wu--2020|Wu, 2020]] ), 33% of European freshwater fish species ( [[#Janssen--2016|Janssen et al., 2016]] ) and 56–69% of odonates in Australia, ( [[#Bush--2014b|Bush et al., 2014b]] ). Assessment of site-specific extirpation for 88 aquatic insect taxa projected that climate change-induced hydrological alteration would result in a 30–40% loss of taxa in warmer, drier ecoregions and a 10–20% loss in cooler, wetter ecoregions ( ''medium evidence, medium agreement'' ) ( [[#Pyne--2017|Pyne and Poff, 2017]] ). In Africa’s Albertine Rift, 51% ( ''n'' = 551) of fish are expected to be impacted by climate change, with 5.5% at a high risk due to their sensitivity and poor adaptative capability ( ''medium evidence, high agreement'' ) ( [[#Carr--2013|Carr et al., 2013]] ).

The GLOBIO-Aquatic model ( [[#Janse--2015|Janse et al., 2015]] a) links models for demography, economy, LUCs, climate change, nutrient emissions, a global hydrological model and a global map of water bodies. It projects that changes in both water quality (eutrophication) and quantity (flow) will generate negative relations in freshwater ecosystems between the persistence of species originally present in each community and a constellation of stressors, including harmful algal blooms. Under a 4°C rise by 2050, mean abundance of species is projected to decline by 70% in running water and by 80% in standing water ( ''medium evidence'' , ''high agreement'' , ''medium confidence'' ) ( [[#Janse--2015|Janse et al., 2015]] a ).

<div id="2.5.1.3.3" class="h4-container"></div>

<span id="global-projections-of-extinction-risk"></span>
===== 2.5.1.3.3 Global projections of extinction risk =====

<div id="h4-34-siblings" class="h4-siblings"></div>

In previous reports, risk assessed from the literature was generally based on estimates of overall range contractions with climate change. In AR4, extinction risk was carefully quantified: ‘There is ''medium confidence'' that approximately 20–30% of species assessed so far are ''likely'' to be at increased risk of extinction if increases in global average warming exceed 1.5–2.5°C (relative to mean temperatures from 1980–1999). As global average temperature increase exceeds about 3.5°C, model projections suggest significant extinctions (40–70% of species assessed) around the globe.’ These estimates approximately correspond to 50–80% reductions in range size (depending upon study), that this assessment equates with a ''‘high’'' and ''‘very high’'' extinction risk, respectively ( [[#IPCC--2007|IPCC, 2007]] ). AR5 stated: ‘a large fraction of terrestrial and freshwater species face increased extinction risk under projected climate change during and beyond the 21st century, especially as climate change interacts with other pressures ( ''high confidence'' )’ ( [[#Field--2014|Field et al., 2014]] ). A series of multi-species and global analyses have been published since AR5, using both statistical models and trait-based approaches.

In this chapter, risk to species, with implications for ecosystems, is assessed using three different approaches. First is an assessment of the geographic distributions of local species’ losses at different levels of GSAT warming, termed ‘local biodiversity loss’, measured as the proportion of species within a given location becoming classified as “endangered” or worse (sensu IUCN), and so at ''high'' ''risk'' of local population losses (local population extinctions) (Figure 2.6). This measure provides the best estimates of which sites are at most risk of losing substantial numbers of species locally, leading to degradation of that ecosystem’s ability to function.

<div id="_idContainer036" class="Figure"></div>

[[File:e2574f66c47e123775e3a9fbd8d549a1 IPCC_AR6_WGII_Figure_2_006.png]]

'''Figure 2.6 | Biodiversity loss for different areas at increasing levels of climate change.''' The higher the percentage of species projected to lose suitable climate in a given area, the higher the risk to ecosystem integrity, functioning and resilience to climate change. Warming levels are based on global levels (GSAT) above pre-industrial temperatures. Colour shading represents proportion of species for which the climate is projected to become sufficiently unsuitable that the species becomes locally ‘endangered’ and at ''high risk'' of local extinction within a given pixel across their current distributions at a given GSAT warming level, based on underlying data ( [[#Warren--2018|Warren et al., 2018]] ) (modelled ''n'' = 119,813 species globally, with no dispersal, averaged over 21 CMIP5 climate models). Areas shaded in deep orange and red represent a significant risk of biodiversity loss (areas where climates become sufficiently unsuitable that it renders &gt;50% and &gt;75% of species at ''high ris'' k of becoming locally extinct, respectively). The maps of species richness remaining have been overlaid with a landcover layer (2015) from the European Space Agency (ESA) Climate Change Initiative. This landcover layer leaves habitats classified by the ESA as natural as transparent. Areas with a landcover identified as agriculture are 5% transparent, such that the potential species richness remaining if the land had not been converted for agriculture shows as pale shading of the legend colours (very pale yellow to very pale red). These paler areas represent biodiversity loss due to habitat destruction, but with a potential to be restored, with yellow shading having the potential for restoration to greater species richness than orange or red shading.

Second is assessment of the proportions of species becoming endangered globally (not just locally), so at ''high'' ''risk'' of global extinction of the species, termed ‘global biodiversity loss’ (Figure 2.8b). This metric (losing &gt; 50% of suitable climate space across the species’ entire range) also serves to estimate a species’ becoming sufficiently rare that the species no longer fully contributes to ecosystem functioning, a state that often occurs decades before complete extinction (death of the last individual). The proportions of species becoming at ''high'' ''risk'' of global extinction is the foundation for the burning embers diagram on global biodiversity loss in Table 2.5 and Figure 2.11.

Third is an assessment of risk of the proportions of species becoming at ''very high'' ''risk'' of extinction globally at different levels of GSAT warming, measured using the IUCN criteria for ‘critically endangered’, and termed ‘species’ extinction risk’ (Figure 2.7 and Figure 2.8a). This measure is closest to assessing the complete loss of a species in the wild and can be used to compare to past (palaeo) extinction rates. These three approaches provide complementary information of the overall risks to individual species, to biodiversity at the community scale, and to ecosystem integrity and functioning at different levels of warming.

<div id="_idContainer039" class="Figure"></div>

[[File:78fbe924ed17f551321ff6e54460af6e IPCC_AR6_WGII_Figure_2_007.png]]

'''Figure 2.7 | Global assessment of species’ extinction risks under different levels of warming.''' Graph shows a synthesis of climate-driven models of individual species projected to become at ''very high'' ''risk'' of extinction globally (i.e, becoming “critically endandered” sensu IUCN by losing &gt;80% of their suitable climate space or through estimates of extinction risk from process-based models). The relationship between modelled projections of extinction (expressed as a proportion of species at a risk of extinction assessed in individual studies) and GSAT increase above the pre-industrial average. Data (global sample size ''n'' = 178 modelled estimates) were taken from a number of sources, including digitization of data points in Figure 2 in the synthetic analysis of ( [[#Urban--2015|Urban, 2015]] ) (note: unweighted for sample size) ''n'' = 126; Table 4.1 of AR4 WGII [https://www.ipcc.ch/report/ar6/wg2/chapter/chapter-2 Chapter 2] ( [[#Fischlin--2007|Fischlin et al., 2007]] ), ''n'' = 40; ( [[#Hannah--2020|Hannah et al., 2020]] ) ''n'' = 6; and ( [[#Warren--2018|Warren et al., 2018]] ) ''n'' = 6. The quantile regression (which is robust to the non-normal distribution of the response variable, and less sensitive to data outliers) was chosen as a descriptive statistic to fit quantile estimates for levels relevant to informing ''likely'' (between the 0.17 and 0.83 quantiles, shaded in orange) and ''very likely'' ranges (between the 0.05 and 0.95 quantiles, shaded in green) relating extinction risk to GSAT increase (quantile regression implemented using the Barrodale and Roberts algorithm in XLSTAT). The roughly equivalent estimate of this risk as expressed in AR4 ( [[#Fischlin--2007|Fischlin et al., 2007]] ) is indicated by the dotted block indicating the ''medium confidence'' statement ‘Approximately 20–30% of plant and animal species assessed so far (in an unbiased sample) are likely to be at increasingly high risk of extinction as global mean temperatures exceed a warming of 2–3°C above pre-industrial levels ( ''medium confidence'' ) ''.'' ’ This box is open on the right side because AR4 estimates stipulated temperatures at or exceeding the given levels. Thick dark horizontal bars show the median values of percent of species at ''very high risk'' of extinction at 1.5°C, 2°C, 3°C, 4°C and 5°C, indicating that half of the data points lie above the bar and half below for a given level of global warming.

Risk of local biodiversity loss, estimated as the proportion of species in a given area projected to become endangered (sensu IUCN), and therefore at ''high'' ''risk'' of extinction, is projected to affect a greater number of regions experiencing increasing warming. About one-third of land area risks more than 50% of species becoming “endangered” by 4.0° GSAT warming (Figure 2.6). That is, the deep orange and red areas in Figure 2.6 are those areas for which &gt;50% of species currently inhabiting those ecosystems are projected to lose &gt;50% of their climatically suitable habitat. Species’ losses are projected to be worst in northern South America, southern Africa, most of Australia and at northern high latitudes ( ''medium confidence'' ) (Figure 2.6).

For risk of global biodiversity loss, at 1.58°C global warming (median estimate), &gt;10% of species are projected to become “endangered”, and so at ''high'' ''risk'' of extinction (sensu IUCN). At 2.07°C (median) &gt;20% of species are projected to become endangered. Ten-twenty percent losses represent ''high'' and ''very high'' ''risk'' of biodiversity losses, respectively, substantial enough to reduce ecosystem integrity and functioning ( ''medium confidence'' ) (Figure 2.8b) (see [[#2.5.4|Section 2.5.4]] ; Figure 2.11; Table 2.5, Table SM2.5).

Risk of global biodiversity loss differs among taxonomic groups. The percent of species projected at ''high risk'' of extinction was 49% for all insects, 44% for all plants and 26% for all vertebrates at ~3°C global rise in temperature (Figure 2.8b) ( [[#Warren--2018|Warren et al., 2018]] ). These estimates dropped considerably at lower levels of warming, down to 18%, 16% and 8% at 2°C; and 6%, 8% and 4% at 1.5°C (Figure 2.8b) ( [[#Warren--2018|Warren et al., 2018]] ), so not entirely dissimilar to the numbers in AR4 (Figure 2.7).

<div id="_idContainer041" class="Figure"></div>

[[File:d08b5d24cdb99240572070d277b5660f IPCC_AR6_WGII_Figure_2_008.png]]

'''Figure 2.8 | Percent of species of different groups classified as being at risk of extinction.'''

'''(a)''' Species groups listed projected to be at a ''very high'' ''risk'' of extinction, corresponding to the IUCN Red List criteria for a species classified as ‘critically endangered’ (v3.1) by losing &gt;80% of its climatically suitable range area.

'''(b)''' Species groups listed projected to be at a ''high'' ''risk'' of extinction, corresponding to the IUCN Red List criteria for a species classified as ‘endangered’ (v3.1) by losing &gt;50% of its climatically suitable range area. For (a) and (b), values were calculated from the underlying data in ( [[#Warren--2018|Warren et al., 2018]] ). Values for each temperature are the mean values across 21 CMIP5 models. The grey band represents the high end of extinction risk from the 10th percentile of the climate models to show the maximum range of values, while the low end (90th percentile, 1.5°C) is not shown as it is too small to appear on the plots. Taxa marked with * represent potential benefits from adaptation, specifically dispersal at realistic rates ( [[#Warren--2018|Warren et al., 2018]] ); those with no * have dispersal rates that are essentially not detected in the spatial resolution of the models (20 km). See ( [[#Warren--2018|Warren et al., 2018]] ) for caveats and more details. Sample size for each group is as follows: 1) fungi (16187 species); 2) all plants (72399 species), broken down into sub-groups of plants: flowering plants (52310 species), timber species (1328 species), grasses (3389 species) and pines (340 species); 3) all invertebrates (33,949 species), broken down into sub-groups of invertebrates: annelid worms (155 species), flies (4809 species), beetles (7630 species), moths (6910 species), true bugs (1728 species), spiders (2212 species), all pollinators (1755 species), butterflies (1684 species), ants/bees/wasps (5914 species), dragonflies (599 species); 4) Chordates (12642 species), broken down into major groups: 4i) all amphibians (1055 species), broken down into sub-groups of amphibians: frogs (887 species) and salamanders (163 species); 4ii) reptiles (1850 species), snakes (1741 species) and turtles (94 species); 4iii) all mammals (1769 species), broken down into sub-groups of mammals: ungulates (80 species), bats (500 species), carnivores (107 species), 4iv) all birds (7968 species), broken down into sub-groups of birds: passeriforme birds (4744 species), and non-passeriforme birds (3224 species).

‘Species’ extinction risk’, estimated as at ''very high'' ''risk'' of extinction globally, i.e. becoming “critically endangered” (sensu IUCN) is shown in Figures 2.7 across 178 studies and in Figure 2.8a split by taxonomic group. The percentage of species at ''very high'' ''risk'' of extinction (median estimates and maximum ''likely'' range) will be 9% (max. 14%) at 1.5°C, 10% (max. 18%) at 2°C, 12% (max. 29%) at 3.0°C, 13% (max. 39%) at 4°C and 15% (max. 48%) at 5°C (Figure 2.7). Maximum estimates of species at ''very high'' ''risk'' of extinction reach 60% within the 95% quartiles, ie the ''very likely'' range, for 5°C GSAT warming. Among the groups containing the largest numbers of species at a ''very high risk'' of extinction for mid-levels of projected warming (3.2°C rise in GSAT) are: invertebrates (15%), specifically pollinators (12%), amphibians (11%, but 24% for salamanders) and flowering plants (10%) (Figure 2.8a). All groups fare substantially better at 2°C, with extinction projections reducing to &lt;3% for all groups, except salamanders at 7% ( ''medium confidence'' ) (Figure 2.8a) ''.''

Figure 2.8 also shows the benefits of dispersal in reducing extinction risk in birds, mammals, butterflies, moths and dragonflies (depicted with an asterix). While dispersal may benefit individual species, it poses additional risks to communities and ecosystems that species are moving into, as interactions between species are changed or eliminated.

Projected species extinctions at future global warming levels are in accord with projections from AR4, assessed on much larger numbers of species with much greater geographic coverage and a broader range of climate models. (Figure 2.7; Figure 2.8a). Even the lowest estimates of species extinction (median of 9% at 1.5°C warming, Figure 2.7) are 1000 times the natural background rates ( [[#De%20Vos--2015|De Vos et al., 2015]] ).

Using data from geological timescales, ( [[#Song--2021|Song et al., 2021]] ) predicted that a warming of 5.2°C above pre-industrial levels would result in a mass extinction comparable to that of the five mass extinctions over the past 540 Myr, on the order of 70–85% of species becoming extinct, in the absence of non-climatic stressor. ( [[#Mathes--2021|Mathes et al., 2021]] ) found evidence in the geological record that short-term rapid warming, on top of long-term warming trends, increases extinction risk by up to 40% over that expected from the long-term trend alone, with a biodiversity ‘memory’ of up to 60 Myr, indicating an additonal risk of multi-decadal overshoot.

Most of the large-scale studies that have been performed are for losses based on climate alone (Figures 2.6, 2.7, 2.8). However, climate is rarely the only stressor affecting species survival. Habitat loss is currently the largest driver of range loss and extinction risk for most species ( [[#IUCN--2019|IUCN, 2019]] ). Communities in different regions are becoming more similar to each other as species tolerant of human activities prosper and spread, with many rare and endemic species already having been driven to extinction, primarily by LULCC ( [[#Pimm--2006|Pimm et al., 2006]] ). Thus, it will likely be the interaction of climate change and habitat conversion (often driven by climate change) that will ultimately determine the risk and ability of many species to survive over the next century.

<div id="2.5.1.4" class="h3-container"></div>

<span id="changing-risks-of-diseases"></span>
==== 2.5.1.4 Changing Risks of Diseases ====

<div id="h3-32-siblings" class="h3-siblings"></div>

Multiple studies predict increases in disease incidence or geographic and phenological changes of pathogens, vectors and reservoir host species due to climate change with or without other non-climatic variables ( [[#González--2010|González et al., 2010]] ; [[#Moo-Llanes--2013|Moo-Llanes et al., 2013]] ; [[#Roy-Dufresne--2013|Roy-Dufresne et al., 2013]] ; [[#Liu-Helmersson--2014|Liu-Helmersson et al., 2014]] ; [[#Laporta--2015|Laporta et al., 2015]] ; [[#Ryan--2015|Ryan et al., 2015]] ; [[#Haydock--2016|Haydock et al., 2016]] ; [[#Hoover--2016|Hoover and Barker, 2016]] ; [[#Prist--2017|Prist et al., 2017]] ; [[#Blum--2018|Blum and Hotez, 2018]] ; [[#Dumic--2018|Dumic and Severnini, 2018]] ; [[#Hundessa--2018|Hundessa et al., 2018]] ; [[#Ryan--2019|Ryan et al., 2019]] ; [[#Ryan--2021|Ryan et al., 2021]] ). However, models predicting changes in infectious disease risk are complex and sometimes produce conflicting results and lack consensus ( [[#Caminade--2014|Caminade et al., 2014]] ; [[#Giesen--2020|Giesen et al., 2020]] ). For example, malaria is projected to increase in some regions of Africa, Asia and South America by the end of the 21st century if public health interventions are not sufficient, but is also forecasted to decrease in some higher-risk areas (Cross-Chapter Box Illness in this chapter) ( [[#Peterson--2009|Peterson, 2009]] ; [[#Caminade--2014|Caminade et al., 2014]] ; [[#Ryan--2015|Ryan et al., 2015]] ; [[#Khormi--2016|Khormi and Kumar, 2016]] ; [[#Leedale--2016|Leedale et al., 2016]] ; [[#Murdock--2016|Murdock et al., 2016]] ; [[#Endo--2020|Endo and Eltahir, 2020]] ; [[#Mordecai--2020|Mordecai et al., 2020]] ).

While malaria risk is predicted to decrease in some lowland tropical areas as temperatures become too hot for vector or parasite development, other warm-adapted diseases, like dengue and Zika, transmitted by ''A. aegypti'' , are predicted to increase (Cross-Chapter Box Illness in this chapter, chapter 7) ( [[#Ryan--2019|Ryan et al., 2019]] ; [[#Ryan--2021|Ryan et al., 2021]] ). In more temperate regions, arboviruses and other VBDs with wider thermal breadths, such as West Nile fever, Ross River fever and Lyme disease, are predicted to increase with climate warming ( [[#Ogden--2008|Ogden et al., 2008]] ; [[#Leighton--2012|Leighton et al., 2012]] ; [[#Shocket--2018|Shocket et al., 2018]] ; [[#Shocket--2020|Shocket et al., 2020]] ; [[#Couper--2021|Couper et al., 2021]] ). Drought can exacerbate these effects of temperature ( [[#Paull--2017|Paull et al., 2017]] ).

A global analysis of 7346 wildlife populations and 2021 host–parasite combinations found that organisms adapted to cool and mild climates are likely to experience increased risks of outbreaks along with climate warming, while warm-adapted organisms may experience a lower disease risk, providing further support for predictions that climate change will increase the transmission of infectious diseases at higher latitudes across a taxonomically diverse array of pathogens ( ''robust evidence'' , ''high agreement'' ) ( [[#Cohen--2020|Cohen et al., 2020]] ). A study examining the future risk of arboviruses (chikungunya, dengue, yellow fever and Zika viruses) spread by ''A. aegypti'' and ''A. albopictus'' projected increased disease risk due to interactions of multiple variables, including increased human connectivity, urbanisation and climate change ( [[#Kraemer--2019|Kraemer et al., 2019]] ), although vector species’ ranges will broaden only slightly ( [[#Campbell--2015|Campbell et al., 2015]] ).

In sum, climate change is expected to expand and redistribute the burden of vector-borne and other environmentally transmitted diseases of wild animals, domesticated animals and humans, by shifting many regions toward the thermal optima of VBD transmission for multiple parasites, thereby increasing risk of transmission, while pushing temperatures above optimal and towards upper thermal limits for other vectors and pathogens, thus decreasing their transmission ''(high confidence)'' (see also chapter 7) ( [[#Mordecai--2019|Mordecai et al., 2019]] ; [[#Mordecai--2020|Mordecai et al., 2020]] ). These effects are mediated by other human impacts such as LUC, mobility, socioeconomic conditions and vector and pathogen control measures ( [[#Parham--2015|Parham et al., 2015]] ; [[#Tjaden--2018|Tjaden et al., 2018]] ).

<div id="2.5.2" class="h2-container"></div>

<span id="projected-changes-at-level-of-biomes-and-whole-ecosystems"></span>
=== 2.5.2 Projected Changes at Level of Biomes and Whole Ecosystems ===

<div id="h2-13-siblings" class="h2-siblings"></div>

<div id="2.5.2.1" class="h3-container"></div>

<span id="global-overview-assessment-of-ecosystem-level-models-and-sources-of-uncertainties"></span>
==== 2.5.2.1 Global Overview, Assessment of Ecosystem-Level Models and Sources of Uncertainties ====

<div id="h3-33-siblings" class="h3-siblings"></div>

Shifts in terrestrial biome and changes in ecosystem processes in response to climate change are most frequently projected with dynamic global vegetation models (DGVMs) or land-surface models that form part of ESMs, which use gridded climate variables, atmospheric CO 2 concentration and information on soil properties as input variables. Since AR5, most DGVMs have been upgraded to capture carbon–nitrogen cycle interactions (e.g., ( [[#Le%20Quéré--2018|Le Quéré et al., 2018]] ), many also include a representation of wildfire and fire–vegetation interactions ( [[#Rabin--2017|Rabin et al., 2017]] ) and a small number now also account for land management (e.g., wood removal from forests and crop fertilisation harvest of irrigation ( [[#Arneth--2017|Arneth et al., 2017]] ). Other forms of disturbance, such as tree mortality, in response to, for example, episodic weather extremes or insect pest outbreaks, are relatively poorly represented or not at all, although they demonstrably impact calculated carbon cycling ( [[#Pugh--2019a|Pugh et al., 2019a]] ). Simulated biome shifts are generally in agreement in projecting broad patterns on a global scale but vary greatly regarding the simulated trends in historical and future carbon uptake or losses, both regionally and globally ( [[#Chang--2017|Chang et al., 2017]] ; [[#Canadell--2021|Canadell et al., 2021]] ).

Similar to other models, models to project large-scale changes in vegetation and ecosystem processes have to deal with structural uncertainty (associated with the choice and the representation of processes in models), input-data uncertainty (associated with variability in initial conditions and parameter values) and error propagation (associated with coupling models) ( [[#Rounsevell--2019|Rounsevell et al., 2019]] ). The IPBES methodological assessment report on scenarios and models of biodiversity and ecosystem services provides a comprehensive overview over the relevant issues ( [[#Ferrier--2016|Ferrier et al., 2016]] ).

In order to assess performance, most models have been individually evaluated against a range of observations. Moreover, in the annual updates of the global carbon budgets, a model has to meet a small set of basic criteria to have its output included ( [[#Le%20Quéré--2018|Le Quéré et al., 2018]] ). More systematic benchmarking approaches have also been proposed that utilise a range of different datasets ( [[#Kelley--2013|Kelley et al., 2013]] ; [[#Chang--2017|Chang et al., 2017]] ) to assess multiple simulated processes. These methods, in principle, facilitate assigning quality scores to models based on their overall performance ( [[#Kelley--2013|Kelley et al., 2013]] ). So far, this scoring does not yet allow a clear quality ranking of models, since individual DGVMs tend to score well for some variables and badly for others. A recent comparison of global fire–vegetation model outputs was also able to clearly identify outliers when using a formalised benchmarking and scoring approach ( [[#Hantson--2020|Hantson et al., 2020]] ). However, benchmarking does not address sources of uncertainty and it would be advisable to perform ‘perturbed-physics’ experiments, in which multiple model parameters are varied in parallel more frequently as a means to test parameter-value uncertainty ( [[#Wramneby--2008|Wramneby et al., 2008]] ; [[#Booth--2012|Booth et al., 2012]] ; [[#Lienert--2018|Lienert and Joos, 2018]] ).

Species diversity impacts ecosystem functioning and hence ecosystem services ( [[#Hooper--2012|Hooper et al., 2012]] ; [[#Mokany--2016|Mokany et al., 2016]] ). So far, however, integrated modelling of ecosystem processes and biodiversity across multiple trophic levels and food webs is in its infancy ( [[#Harfoot--2014|Harfoot et al., 2014]] ). Whether or not the enhanced integration of state, function and functional diversity across multiple trophic levels in models will markedly alter projections of how ecosystems respond to climate change thus remains an open research question.

Beyond dynamic simulation of biome shifts and carbon cycling, which are important aspects of climate regulation, DGVMs can also provide information on a number of variables closely linked to other ecosystem services such as water availability, air quality or food provisioning ( [[#Krause--2017|Krause et al., 2017]] ; [[#Rabin--2020|Rabin et al., 2020]] ). However, they are not intended to provide a comprehensive assessment of ecosystem services. For these, other approaches applied but, to date, these are mostly applied on regional scales and are only weakly dynamic ( [[#Ferrier--2016|Ferrier et al., 2016]] ).

<div id="2.5.2.2" class="h3-container"></div>

<span id="projected-changes-globally-at-the-biome-level"></span>
==== 2.5.2.2 Projected Changes Globally at the Biome Level ====

<div id="h3-34-siblings" class="h3-siblings"></div>

Climate change and the associated change in atmospheric CO 2 levels already exacerbate other human-caused impacts on the structure and composition of land and freshwater ecosystems, such as LULCC, nitrogen deposition and pollution. The relative importance of these drivers for ecosystems over the coming decades will likely differ between biomes, but climate change and atmospheric CO 2 will be pervasive unless there is a rapid lowering of fossil-fuel emissions and warming ( ''high confidence'' ) ( [[#Pereira--2010|Pereira et al., 2010]] ; [[#Warren--2011|Warren, 2011]] ; [[#Ostberg--2013|Ostberg et al., 2013]] ; [[#Davies-Barnard--2015|Davies-Barnard et al., 2015]] ; [[#Pecl--2017|Pecl et al., 2017]] ; [[#Ostberg--2018|Ostberg et al., 2018]] ). Global vegetation and ESMs agree on climate change-driven shifts of biome boundaries of potentially hundreds of kilometres over this century, combined with several substantial alterations that take place within biomes (e.g., changes in phenology, canopy structure and functional diversity, etc.). Large discrepancies exist between models and between scenarios regarding the region and the speed of change ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Pereira--2010|Pereira et al., 2010]] ; [[#Pecl--2017|Pecl et al., 2017]] ), but robust understanding is emerging in that the degree of impact increases in high-emission and high-warming scenarios ( ''high confidence'' ) (Figure 2.9).

<div id="_idContainer043" class="Figure"></div>

[[File:f307bbaf4831631b9ffa621bd753e09d IPCC_AR6_WGII_Figure_2_009.png]]

'''Figure 2.9 | Projected fraction of global terrestrial area that could experience a biome shift by 2100.''' Shifts due to climate change (filled symbols) or a combination of climate change and LUC (outline symbols), from publications in Supplementary Table SM2 '''.''' 3 (projected vulnerabilities and risks of ecosystems to biome shifts). Filled circles ( [[#Bergengren--2011|Bergengren et al., 2011]] ), filled squares ( [[#Alo--2008|Alo and Wang, 2008]] ), filled diamonds ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ), filled triangle pointing up ( [[#Scholze--2006|Scholze et al., 2006]] ), filled triangle pointing down ( [[#Sitch--2008|Sitch et al., 2008]] ), filled triangle on its side ( [[#Li--2018|Li et al., 2018]] ), filled cross ( [[#Warszawski--2013|Warszawski et al., 2013]] ), outlined circle ( [[#Ostberg--2018|Ostberg et al., 2018]] )and outlined diamond ( [[#Eigenbrod--2015|Eigenbrod et al., 2015]] ).

Substantial changes in vegetation structure and ecosystem processes are already happening (see [[#2.4|Section 2.4]] ). Many of these observations have already been projected to take place as early as at least IPCC AR3 ( [[#Rosenzweig--2007|Rosenzweig et al., 2007]] ), and can they now be increasingly tested for their robustness with observational evidence. These multiple changes in response to warming (and changes in precipitation and increasing atmospheric CO 2 levels that go hand-in-hand with warming) are further expected for already relatively small additional temperature increases. In particular, in cold (boreal and tundra) regions, as well as in dry regions ( ''high confidence'' ), alterations of 2–47% of the areal extent of terrestrial ecosystems in scenarios of &lt;2°C warming above pre-industrial levels have been projected, increasing drastically with higher-warming scenarios ( [[#Warren--2011|Warren, 2011]] ; [[#Wårlind--2014|Wårlind et al., 2014]] ). More recent work, applying also probabilistic methods, confirm the risk of drastic changes in vegetation cover (e.g., forest to non-forest or vice versa) at the end of the 21st century even for approximately 2°C warming scenarios, especially in tundra, and also in tropical forest and savannah regions, with more subtle changes (within a given biome type) likely to occur in all regions ( [[#Ostberg--2013|Ostberg et al., 2013]] ; [[#Ostberg--2018|Ostberg et al., 2018]] ). Model studies have found 5–20% of terrestrial ecosystems affected by warming of around 2°C–3°C, increasing to above one-third at a warming of 4°C–5°C ( [[#Ostberg--2013|Ostberg et al., 2013]] ; [[#Warszawski--2013|Warszawski et al., 2013]] ).

In general, vegetation types are projected to be moving into their ‘neighbouring’ climates, depending on whether temperature or precipitation is expected to be the predominant factor and how vegetation interacts with the increasing CO 2 levels in the atmosphere ( [[#Wårlind--2014|Wårlind et al., 2014]] ; [[#Scheiter--2015|Scheiter et al., 2015]] ; [[#Schimel--2015|Schimel et al., 2015]] ; [[#Huntzinger--2017|Huntzinger et al., 2017]] ). For instance, boreal or temperate forest vegetation is simulated to migrate polewards, closed tropical (moist) forest is expected to transition towards dry tropical forest types, while climate-driven degradation might expand arid vegetation cover (Sections 2.5.2.2–2.5.2.9). However, ‘novel ecosystems’, that is, communities with no current or historical equivalent because of the novel combinations of abiotic conditions under climate change, are expected to be increasingly common in the future ( ''medium confidence'' ), although the regions where these novel ecosystems might emerge are still disputed ( [[#Reu--2014|Reu et al., 2014]] ; [[#Radeloff--2015|Radeloff et al., 2015]] ; [[#Ordonez--2016|Ordonez et al., 2016]] ). The possibility of these novel ecosystems and the communities that live within them are a challenge for current modelling of ecosystem shifts, and new approaches to conservation will be required that are designed to adapt to rapid changes in species composition and the ensuing challenges.

<div id="2.5.2.3" class="h3-container"></div>

<span id="risk-to-arid-regions"></span>
==== 2.5.2.3 Risk to Arid Regions ====

<div id="h3-35-siblings" class="h3-siblings"></div>

Shifts in arid system structure and functioning that have been observed to date ( [[#2.4.3.3|Section 2.4.3.3]] ) are projected to continue ''(medium confidence)'' . These include widespread woody plant encroachment, notably in savanna systems in Africa, Australia and South America, and are attributed to interactions of LULCC, climate change and CO 2 fertilisation effects ( [[#Fensholt--2012|Fensholt et al., 2012]] ; [[#Fang--2017|Fang et al., 2017]] ; [[#Stevens--2017|Stevens et al., 2017]] ). Arid Mongolian steppe grassland did not respond to experimentally elevated CO 2 ( [[#Song--2019|Song et al., 2019]] ). Woody encroachment is projected to continue or not reverse in North American drylands ( [[#Caracciolo--2016|Caracciolo et al., 2016]] ) and southern African arid ecosystems ( [[#Moncrieff--2014b|Moncrieff et al., 2014b]] ). Dryland woody encroachment may increase carbon stocks, depending on emissions scenario ( [[#Martens--2021|Martens et al., 2021]] ), but reduce soil water and biodiversity of grassland-dependent species diversity ( [[#Archer--2017|Archer et al., 2017]] ). Warm season (C4) grass expansion into arid shrublands risks sudden ecosystem transformation due to introduced wildfire ( [[#Bradley--2016|Bradley et al., 2016]] ), a risk anticipated for grass-invaded desert ecosystems of Australia and the southwestern USA (Horn and St. Clair, 2017). Novel fire regimes in grassy shrublands have enhanced grass cover locally in the southern African Nama-Karoo ( [[#du%20Toit--2015|du Toit et al., 2015]] ).

Range retractions are projected for endemic plants in southern Africa ( [[#Young--2016|Young et al., 2016]] ) and dry woodlands in Morocco ( [[#Alba-Sánchez--2015|Alba-Sánchez et al., 2015]] ). Increasing thermal stress is projected to increase woody plant mortality in the Sonoran Desert ecosystems ( [[#Munson--2016|Munson et al., 2016]] ) and facilitate perennial grass replacement by xeric shrubs in the southwestern USA ( [[#Bestelmeyer--2018|Bestelmeyer et al., 2018]] ). Ecological effects may occur rapidly when extreme events compound long-term trends ( [[#Hoover--2015|Hoover et al., 2015]] ), but evolve more slowly as opportunity costs accumulate due to warming (Cross-Chapter Paper 3) ( [[#Cunningham--2021|Cunningham et al., 2021]] ).

<div id="2.5.2.4 " class="h3-container"></div>

<span id="risk-to-mediterranean-type-ecosystems-mtes"></span>
==== 2.5.2.4 Risk to Mediterranean-Type Ecosystems (MTEs) ====

<div id="h3-36-siblings" class="h3-siblings"></div>

The regions containing MTEs all show ''high confidence'' in projected increases in the intensity and frequency of hot extremes and decreases in the intensity and frequency of cold extremes, and ''medium confidence'' in increasing ecological drought due to increased evapotranspiration (in all regions) and reduced rainfall (excluding California, USA, where model agreement is low) (see WGI Chapter 11). Projections also show a ''robust'' increase in the intensity and frequency of heavy precipitation in the event of ≥2°C warming for MTEs in South Africa, the Mediterranean Basin and California, USA, but are less clear for Australia and Chile ( [[#Seneviratne--2021|Seneviratne et al., 2021]] ).

MTEs are characterised by the distinctive seasonal timing of precipitation and temperature, and the disruption of this regime is likely to be critical for their maintenance. Unfortunately, projections of changes in rainfall seasonality have received less attention and are far more uncertain than many other aspects of climate change ( [[#Pascale--2016|Pascale et al., 2016]] ; [[#Breinl--2020|Breinl et al., 2020]] ), thus limiting our ability to predict the ecological consequences of climate change in MTEs. Responses to experimental manipulation of rainfall seasonality show the potential for shifts in plant functional composition and diversity loss, but results vary with soil type ( [[#van%20Blerk--2021|van Blerk et al., 2021]] ).

Unfortunately, global- and regional-scale dynamic vegetation models show a poor performance for large areas of MTEs, because they do not characterise shrub and Crassulacean acid metabolism (CAM)-photosynthetic plant functional types well ( [[#Moncrieff--2015|Moncrieff et al., 2015]] ). Furthermore, the grain of these models is too coarse for quantifying impacts to many vegetation formations which are patchy or of limited extent (e.g., small stands of trees). There is ''high confidence'' that observations of high mortality in trees and other growth forms, reduced reproductive and recruitment success, range shifts, community shifts towards more thermophilic species and type conversions are set to continue, due to either direct climate impacts through drought and other extreme weather events or to their interaction with factors like fire and pathogens (Sections 2.4.3.5, 2.4.3.6; 2.4.3.7; 2.4.4.2; 2.4.4.3; 2.5.2.5, 2.5.2.6, 2.5.2.7, 2.5.4).

Fire is a key driver across most MTEs due to summer-dry conditions. Climate projections for the MTEs translate into high confidence that periods of low fuel moisture will become more severe and prolonged, and that episodes of extreme fire weather will become more frequent and severe (see ( [[#Douville--2021|Douville et al., 2021]] ; [[#Seneviratne--2021|Seneviratne et al., 2021]] )). This will lead to the birth of novel fire regimes in MTEs, characterised by an increase in the probability of greater burned area and extreme wildfire events (e.g., megafires), with associated loss of human life and property, long-term impacts on ecosystems and acceleration of the possible loss of resilience and capacity to recover ( [[#Abatzoglou--2016|Abatzoglou and Williams, 2016]] ; [[#González--2018|González et al., 2018]] ; [[#Boer--2020|Boer et al., 2020]] ; [[#Moreira--2020|Moreira et al., 2020]] ; [[#Nolan--2020|Nolan et al., 2020]] ; [[#Duane--2021|Duane et al., 2021]] ; [[#Gallagher--2021|Gallagher et al., 2021]] ).

Fire is virtually certain to have additional impacts through compound events (see Section 11.8 in ( [[#Seneviratne--2021|Seneviratne et al., 2021]] )). Extreme post-fire weather is extremely likely to continue to impact diversity ( [[#Slingsby--2017|Slingsby et al., 2017]] ), retard vegetation regrowth ( [[#Slingsby--2020a|Slingsby et al., 2020a]] ) and accelerate vegetation shifts ( [[#Batllori--2019|Batllori et al., 2019]] ). Any increases in the intensity and frequency of heavy precipitation are highly likely to compromise soil stability in recently burnt areas ( [[#Morán-Ordóñez--2020|Morán-Ordóñez et al., 2020]] ). The impacts of fire often depend on interactions with non-climatic factors such as habitat fragmentation ( [[#Slingsby--2020b|Slingsby et al., 2020b]] ) and management ( [[#Steel--2015|Steel et al., 2015]] ) or the spread of flammable exotic plantation forestry and invasive species ( [[#Kraaij--2018|Kraaij et al., 2018]] ; [[#McWethy--2018|McWethy et al., 2018]] ). Managing these factors provides opportunities for adaptation and mitigation ( [[#Moreira--2020|Moreira et al., 2020]] ). (See sections 2.4.4.2 and 2.5.3.2).

Human adaptation and mitigation responses to climate change may create additional threats to MTEs. MTEs have dry summers by definition, posing a challenge for the year-round supply of water to growing human populations and agriculture. With recent major droughts in all MTEs ( [[#2.4.3.6|Section 2.4.3.6]] ), there is increasing reliance on groundwater for the bulk of the water supply ( [[#Kaiser--2018|Kaiser and Macleod, 2018]] ). The majority of groundwater systems have exceeded or are rapidly approaching their environmental flow limits ( [[#de%20Graaf--2019|de Graaf et al., 2019]] ), threatening human populations and ecosystems that depend on these systems for their persistence through unfavourable climatic conditions ( [[#McLaughlin--2017|McLaughlin et al., 2017]] ). Similarly, much of the MTEs are open shrublands and grasslands and proposed extensive tree-planting to sequester atmospheric CO 2 could result in a loss of biodiversity and threaten water security ( [[#Doblas-Miranda--2017|Doblas-Miranda et al., 2017]] ; [[#Bond--2019|Bond et al., 2019]] ).

<div id="2.5.2.5" class="h3-container"></div>

<span id="risk-to-grasslands-and-savannas"></span>
==== 2.5.2.5 Risk to Grasslands and Savannas ====

<div id="h3-37-siblings" class="h3-siblings"></div>

Worldwide, woody cover is increasing in savannas ( [[#Buitenwerf--2012|Buitenwerf et al., 2012]] ; [[#Donohue--2013|Donohue et al., 2013]] ; [[#Stevens--2017|Stevens et al., 2017]] ), as a result of interactions of elevated CO 2 and altered fire and herbivory impacts, some of which stems from LULCC ''(high confidence)'' (see [[#2.4.3.5|Section 2.4.3.5]] ; Cross-Chapter Paper 3.2) ( [[#Venter--2018|Venter et al., 2018]] ; [[#Wu--2021|Wu et al., 2021]] ). In some regions, altered climate may also contribute (Cross-Chapter Paper 3.2). Elevated CO 2 benefits plants with C3 photosynthesis (often woody plants), more than C4 species ( [[#Moncrieff--2014a|Moncrieff et al., 2014a]] ; [[#Scheiter--2015|Scheiter et al., 2015]] ; [[#Knorr--2016a|Knorr et al., 2016a]] ). Increases in woody vegetation in grassy ecosystems could provide some carbon increase ( ''medium confidence'' ) ( [[#Zhou--2017|Zhou et al., 2017]] ; [[#Mureva--2018|Mureva et al., 2018]] ), but is expected to decrease biodiversity ( [[#Smit--2015|Smit and Prins, 2015]] ; [[#Abreu--2017|Abreu et al., 2017]] ; [[#Andersen--2019|Andersen and Steidl, 2019]] ) and water availability ( [[#Honda--2016|Honda and Durigan, 2016]] ; [[#Stafford--2017|Stafford et al., 2017]] ) and alter ecosystem services like grazing and wood provision ( ''high confidence'' ) ( [[#Anadón--2014b|Anadón et al., 2014b]] ).

The relative importance of climate, disturbance (e.g., fire/herbivory) and plant feedbacks in shaping present and future savanna distribution varies between continents ( [[#Lehmann--2014|Lehmann et al., 2014]] ), which makes projections of changing the biome extent challenging ( [[#Moncrieff--2016|Moncrieff et al., 2016]] ). It has been shown that simulation studies that do not account for CO 2 interactions but only consider climate change impacts do not realistically capture the future distribution of savannas ( ''high confidence'' ) ( [[#Higgins--2012|Higgins and Scheiter, 2012]] ; [[#Moncrieff--2016|Moncrieff et al., 2016]] ; [[#Scheiter--2020|Scheiter et al., 2020]] ). Due to the continued strong effect of CO 2 on tree and shrub-to-grass ratios in future, models suggest a loss of savanna extent and conversion into closed canopy forest/thicket and an expansion of savanna-type vegetation into arid grasslands ( [[#Wårlind--2014|Wårlind et al., 2014]] ; [[#Moncrieff--2016|Moncrieff et al., 2016]] ). In arid savannas and their interface to grasslands, survival of woody vegetation (which may be stimulated to grow by increasing CO 2 ) will depend on their capacity to survive potentially more severe and frequent droughts ( [[#Sankaran--2019|Sankaran and Staver, 2019]] ). Across a range of models, for RCP4.5 future climate change and CO 2 concentrations, savanna expanse declines by around 50% (converting to closed canopy systems) by 2070 in Africa and South America, 25% in Asia and with small changes in Australia ( [[#Moncrieff--2016|Moncrieff et al., 2016]] ; [[#Kumar--2021|Kumar et al., 2021]] ). Future fire-spread is expected to be reduced with increased woody dominance ( [[#Scheiter--2015|Scheiter et al., 2015]] ; [[#Knorr--2016b|Knorr et al., 2016b]] ; [[#Scheiter--2020|Scheiter et al., 2020]] ), feeding back to further increase tree-to-grass ratios ( ''high confidence'' ).

Like the tropical forest biome, savannas are at a high risk, given the projected climate changes in combination with LULCC (see Cross-Chapter Paper 3). About 50% of the Brazilian Cerrado has been converted to agricultural land and pastures ( [[#Lehman--2016|Lehman and Parr, 2016]] ), and African savannas have been proposed to follow a similar tropical agricultural revolution pathway to enhance agronomic prosperity ( [[#Ryan--2016|Ryan et al., 2016]] ). In fact, indirect climate change impacts arising from mitigation efforts may be particularly perilous to savannas; extensive tree-planting to restore ecosystems and remove CO 2 from the atmosphere, as pledged, for example, under the African Forest Restoration Initiative, could lead to carbon losses and the loss of biodiversity as well as damage the water balance if trees are planted on what was naturally grassland or savanna (Box 2.2; FAQ 2.6) ( [[#Bond--2019|Bond et al., 2019]] ).

<div id="2.5.2.6" class="h3-container"></div>

<span id="risk-to-tropical-forests"></span>
==== 2.5.2.6 Risk to Tropical Forests ====

<div id="h3-38-siblings" class="h3-siblings"></div>

Key factors affecting the future distribution of tropical humid and dry forests are amounts and seasonalities of precipitation, increased temperatures, prolonged droughts and droughted-moderated fires ( ''robust evidence'' , ''high agreement'' ) ( [[#Bonai--2016|Bonai et al., 2016]] ; [[#Corlett--2016|Corlett, 2016]] ; [[#Lyra--2017|Lyra et al., 2017]] ; [[#Anderson--2018|Anderson et al., 2018]] ; [[#da%20Silva--2018|da]] [[#Silva--2018|Silva et al., 2018]] ; [[#Fontes--2018|Fontes et al., 2018]] ; [[#O’Connell--2018|O’Connell et al., 2018]] ; [[#Aguirre-Gutiérrez--2019|Aguirre-Gutiérrez et al., 2019]] ; [[#Bartlett--2019|Bartlett et al., 2019]] ; [[#Brando--2019|Brando et al., 2019]] ; [[#Stan--2019|Stan and Sanchez-Azofeifa, 2019]] ). The probability of severe drought is projected to quadruple in natural areas in Brazil with &gt;2°C warming ( [[#Barbosa--2016|Barbosa and Lakshmi Kumar, 2016]] ; [[#Marengo--2020|Marengo et al., 2020]] ). Most multi-model studies assuming rapid economic growth/business-as-usual scenarios (A2, A1B and RCP8.5) show an increase in future woody biomass and areas of woody cover towards the end of the 21st century in temperate regions ( [[#Boit--2016|Boit et al., 2016]] ; [[#Nabuurs--2017|Nabuurs et al., 2017]] ) and tropical forests in East Africa ( [[#Ross--2021|Ross et al., 2021]] ) but a decrease in the remaining tropical regions ( [[#Anadón--2014a|Anadón et al., 2014a]] ; [[#Boit--2016|Boit et al., 2016]] ; [[#Lyra--2017|Lyra et al., 2017]] ; [[#Nabuurs--2017|Nabuurs et al., 2017]] ; [[#Maia--2020|Maia et al., 2020]] ). Terrestrial species are predicted to shift to cooler temperatures and higher elevations ( [[#Pecl--2017|Pecl et al., 2017]] ). Tropical species are more susceptible to climate warming than temperate species ( [[#Rehm--2016|Rehm and Feeley, 2016]] ; [[#Sentinella--2020|Sentinella et al., 2020]] ). This susceptibility will be exacerbated by road-building increasing the ease of access into forests ( [[#Brinck--2017|Brinck et al., 2017]] ; [[#Taubert--2018|Taubert et al., 2018]] ; [[#Bovendorp--2019|Bovendorp et al., 2019]] ; [[#Senior--2019|Senior et al., 2019]] ). Furthermore, most tropical cloud forest species are unable to invade grasslands and this will increase the risk of extinctions in tropical cloud forests ( [[#Rehm--2015|Rehm and Feeley, 2015]] ).

SLR as the result of climate change is likely to influence mangroves in all regions, with greater impact on North and Central America, Asia, Australia and East Africa than on West Africa and South America ( ''robust evidence'' , ''high agreement'' ) ( [[#Alongi--2015|Alongi, 2015]] ; [[#Ward--2016|Ward et al., 2016]] ). On a small scale, mangroves are potentially moving landward ( [[#Di%20Nitto--2014|Di Nitto et al., 2014]] ), while on a large scale they will continue to expand poleward ( [[#Alongi--2015|Alongi, 2015]] ).

Most simulations predict a significant geographical shift of transition areas between tropical forests and savanna in the tropical and subtropical Americas and Himalayas ( [[#Anadón--2014a|Anadón et al., 2014a]] ; [[#Rashid--2015|Rashid et al., 2015]] ). Forest dieback, as postulated for the Amazon region, does not occur in the majority of simulations ( [[#Malhi--2009|Malhi et al., 2009]] ; [[#Poulter--2010|Poulter et al., 2010]] ; [[#Rammig--2010|Rammig et al., 2010]] ; [[#Higgins--2012|Higgins and Scheiter, 2012]] ; [[#Huntingford--2013|Huntingford et al., 2013]] ; [[#Davies-Barnard--2015|Davies-Barnard et al., 2015]] ; [[#Sakschewski--2016|Sakschewski et al., 2016]] ; [[#Wu--2016a|Wu et al., 2016a]] ). Model projections of future biodiversity in tropical forests are rare. Arguably, species are most vulnerable to climate change effects at higher altitudes or at the dry end of tropical forest occurrence ( ''medium evidence'' , ''medium agreement'' ) ( [[#Krupnick--2013|Krupnick, 2013]] ; [[#Nobre--2016|Nobre et al., 2016]] ; [[#Trisurat--2018|Trisurat, 2018]] ). Tropical lowlands are expected to lose plant species as temperatures rise above species’ heat tolerance, but could also generate novel communities of heat-tolerant species ( ''robust evidence'' , ''high agreement'' ) ( [[#Colwell--2008|Colwell et al., 2008]] ; [[#Trisurat--2009|Trisurat et al., 2009]] ; [[#Trisurat--2011|Trisurat et al., 2011]] ; [[#Krupnick--2013|Krupnick, 2013]] ; [[#Zomer--2014a|Zomer et al., 2014a]] ; [[#Zomer--2014b|Zomer et al., 2014b]] ; [[#Sullivan--2020|Sullivan et al., 2020]] ; [[#Pomoim--2021|Pomoim et al., 2021]] ).

Statistical models that correlate data on species abundance with information on human pressures, such as LUCs ( [[#Srichaichana--2019|Srichaichana et al., 2019]] ), population density ( [[#Leclère--2020|Leclère et al., 2020]] ) and hunting ( [[#Mockrin--2011|Mockrin et al., 2011]] ), found in tropical and subtropical forests that birds, invertebrates, mammals and reptiles show a decline in their probability of presence with declining forest cover, particularly pronounced in forest specialists or species with narrow ranges ( [[#Newbold--2014|Newbold et al., 2014]] ). Different soil fauna groups showed different responses in abundance and diversity to climate change conditions ( [[#Coyle--2017|Coyle et al., 2017]] ; [[#Facey--2017|Facey et al., 2017]] ) but these responses can impact decomposition rates and biogeochemical cycles ( ''medium evidence'' , ''low agreement'' ).

Invasive plant species are predicted to expand upward by 500–1,500 m in the western Himalayas ( [[#Thapa--2018|Thapa et al., 2018]] ), and by 6–35% yr -1 from the current extent in South America ( ''robust evidence'' , ''high agreement'' ) ( [[#Bhattarai--2014|Bhattarai and Cronin, 2014]] ). Global assessment ( [[#Wang--2017|Wang et al., 2017]] ) also revealed that ecoregions of high-elevation tropical forests and subtropical coniferous forests have a high risk of invasive plant expansion in low-CO 2- emission scenarios, with negative impacts on ecosystem functioning and local livelihoods ( [[#Shrestha--2019|Shrestha et al., 2019]] ).

The impact of unsustainable land use on tropical forests continues in all regions (see Cross-Chapter Paper 7). Projected climate changes will not only impact biodiversity but also the livelihoods of affected people ( ''robust evidence'' , ''high agreement'' ). Increased drought drives crop failures that cause local communities to expand their agricultural area by further clearing native forests ( [[#Desbureaux--2018|Desbureaux and Damania, 2018]] ). Climate change is projected to enlarge the area of suitability for booming tree crops such as oil palm, acacia, Eucalyptus and rubber ( [[#Koninck--2011|Koninck et al., 2011]] ; [[#Cramb--2015|Cramb et al., 2015]] ; [[#Nath--2016|Nath, 2016]] ; [[#Hurni--2017|Hurni et al., 2017]] ; [[#Li--2017|Li et al., 2017]] ; [[#Varkkey--2018|Varkkey et al., 2018]] ). An increase of 8% in the area of rubber plantations in Yunnan Province, China, between 2002–2010 to higher altitudes due to decreased environmental limits, has potentially increased pressure on the remaining biodiversity both within and outside of protected areas ( [[#Zomer--2014a|Zomer et al., 2014a]] ). As a consequence, the suitable area for mammals is projected to be reduced by 47.7% (RCP2.6) and 67.7% (RCP8.5) by 2070, with large variability depending on the different species (Cross-Chapter Paper 7) ( [[#Brodie--2016|Brodie, 2016]] ). To minimize these potential threats, the Yunnan provincial government has identified suitable areas for the establishment of national parks, including the Asian Elephant National Park since 2006. And the government of China developed a national park system in 2013 across the country.

<div id="2.5.2.7" class="h3-container"></div>

<span id="risks-to-boreal-and-temperate-forests"></span>
==== 2.5.2.7 Risks to Boreal and Temperate Forests ====

<div id="h3-39-siblings" class="h3-siblings"></div>

As in the Arctic, warming substantially exceeding the global average has already been observed in the northern parts of the temperate and boreal forest zone ( [[#Gauthier--2015|Gauthier et al., 2015]] ), and is projected to continue (see [https://www.ipcc.ch/chapter/cross-chapter-paper-6 Cross-Chapter Paper 6] and ( [[#Lee--2021|Lee et al., 2021]] )). As a consequence, boreal tree species are expected to move northwards (or in mountainous regions, upwards) into regions dominated by tundra, unless constrained by edaphic features, and temperate species are projected to grow in regions currently occupied by southern boreal forest ( ''high confidence'' ). In both biomes, deciduous trees are simulated to grow increasingly in regions currently dominated by conifers ( [[#Wårlind--2014|Wårlind et al., 2014]] ; [[#Boulanger--2017|Boulanger et al., 2017]] ). These simulation results have been supported by observational examples. In eastern Siberia, fire disturbance of larch-dominated forest was followed by recovery to birch-dominated forest ( [[#Stuenzi--2020|Stuenzi and Schaepman-Strub, 2020]] ). In Alberta, lodgepole pine ( ''Pinus contorta'' ) lost its dominant status after attacks by mountain pine beetles ( ''Dendroctonus ponderosae'' ) caused the canopy to switch to non-pine conifers and broadleaf trees ( [[#Axelson--2018|Axelson et al., 2018]] ). In contrast to the examples above, some boreal forests have proven resilient to disturbances including recent unprecedented insect outbreaks ( [[#Campbell--2019a|Campbell et al., 2019a]] ; [[#Prendin--2020|Prendin et al., 2020]] ).

Reforestation, either natural or anthropogenic, leads to summer cooling and winter warming of the ground, while forest thinning or removal by fire has reverse effects, deepening the upper layer that is free of permafrost ( [[#Stuenzi--2021a|Stuenzi et al., 2021a]] ). Interactions between permafrost and vegetation are important. For example, trees in the east Siberian taiga obtained water mostly from rain in wet summers and permafrost melt water in dry summers ( [[#Sugimoto--2002|Sugimoto et al., 2002]] ), suggesting that these forests will be particularly vulnerable to the combination of drought with the retraction of permafrost further underground due to climate warming.

<div id="2.5.2.8 " class="h3-container"></div>

<span id="risk-to-peatland-systems"></span>
==== 2.5.2.8 Risk to Peatland Systems ====

<div id="h3-40-siblings" class="h3-siblings"></div>

The overall effect of climate change on the extent of northern peatlands is still debated ( ''limited evidence'' , ''low agreement'' ). It is expected that climate change will drive the expansion of high-latitude peatlands poleward of their present distribution due to warming, permafrost degradation and glacier retreat, which could provide new land and conditions favourable for peat development ( ''limited evidence'' , ''medium agreement'' ) ( [[#Zhang--2017b|Zhang et al., 2017b]] ), as seen during the last de-glacial warming ( ''robust evidence'' , ''high agreement'' ) ( [[#MacDonald--2006|MacDonald et al., 2006]] ; [[#Jones--2010|Jones and Yu, 2010]] ; [[#Ratcliffe--2018|Ratcliffe et al., 2018]] ). Peatland area loss (shrinking) near the southern limit of their current distribution or in areas where the climate becomes unsuitable is also expected ( ''medium evidence'' , ''medium agreement'' ) ( [[#2.3|Section 2.3.4.3.2]] ) ( [[#Finkelstein--2011|Finkelstein and Cowling, 2011]] Gallego-Sala and Prentice, 2013; [[#Schneider--2016|Schneider et al., 2016]] ; [[#Müller--2020|Müller and Joos, 2020]] ) ( [[#Müller--2021|Müller and Joos, 2021]] ), but they could persist if moisture is maintained via their capacity to self-regulate. In western Canada, a study suggests that peatlands may persist until 2100, even though the climate will be less suitable ( [[#Schneider--2016|Schneider et al., 2016]] ). Simulations suggest that climate change-driven increases in temperature and atmospheric CO 2 could drive reductions in the northern peatland area up to 18% (SSP1–2.6), 41% (SSP2–4.5) and 61% (SSP5–8.5) by 2300 ( [[#Müller--2020|Müller and Joos, 2020]] ). This is in contrast with the findings of northern peatland persistence and expansion under RCP2.6 and RCP6.0 scenarios in 1861–2099 by another modelling study ( [[#Qiu--2020|Qiu et al., 2020]] ). In the Tropics, the only available study suggests peatland area will increase until 2300, mainly due to increases in precipitation and the CO 2 fertilisation effect ( [[#Müller--2020|Müller and Joos, 2020]] ; [[#Müller--2021|Müller and Joos, 2021]] ).

The combination of changes in climate and land use represents a substantial risk to peatland carbon stocks, but full assessment is impeded because peatlands are yet to be included in ESMs ( ''limited evidence'' , ''high agreement'' ) ( [[#Loisel--2021|Loisel et al., 2021]] ). It is expected that the carbon balance of peatlands globally will switch from sink to source in the near future (2020–2100), mainly because tropical peatland emissions, together with those from climate change-driven permafrost thaw, will likely surpass the carbon gain expected from climate change-driven enhanced plant productivity in northern high latitudes ( [[#Gallego-Sala--2018|Gallego-Sala et al., 2018]] ; [[#Chaudhary--2020|Chaudhary et al., 2020]] ; [[#Turetsky--2020|Turetsky et al., 2020]] ; [[#Loisel--2021|Loisel et al., 2021]] ) which are mainly caused by groundwater drawdown ( ''robust evidence'' , ''medium agreement'' ) ( [[#Hirano--2014|Hirano et al., 2014]] ; [[#Brouns--2015|Brouns et al., 2015]] ; [[#Cobb--2017|Cobb et al., 2017]] ; [[#Itoh--2017|Itoh et al., 2017]] ; [[#Evans--2021|Evans et al., 2021]] ) ''.'' The overall northern peatland carbon sink has been simulated to persist for at least 300 years under RCP2.6, but not under RCP8.5 ( [[#Qiu--2020|Qiu et al., 2020]] ).

Increases in the extent, severity and duration of fires are expected in all peatland regions in the future due to temperature increases ( [[IPCC:Wg2:Chapter:Chapter-4#4.3.1|Section 4.3.1.1]] ), changes in precipitation patterns ( [[IPCC:Wg2:Chapter:Chapter-4#4.3.1|Section 4.3.1.2]] ) and increases in ignition sources (e.g., lightning) ( [[IPCC:Wg2:Chapter:Chapter-5#5.4.3.2|Section 5.4.3.2]] ), with associated rapid carbon losses to the atmosphere ( ''medium evidence'' , ''high agreement'' ) ( [[#Dadap--2019|Dadap et al., 2019]] ; [[#Chen--2021a|Chen et al., 2021a]] ; [[#Nelson--2021|Nelson et al., 2021]] ). For example, drought has been linked to fires in Southeast Asian peatlands ( [[#Field--2009|Field et al., 2009]] ) and there are predicted decreases in mean summer precipitation (10–30%) for high and low RCPs, particularly over the Indonesian region, by the mid and late 21st century ( [[IPCC:Wg2:Chapter:Chapter-12#12.4|Section 12.4.2.2]] ) ( [[#Tangang--2020|Tangang et al., 2020]] ; [[#Taufik--2020|Taufik et al., 2020]] ). During wet years, the fire probability in Indonesian peatlands also significantly increases (by 15–40%) when temperatures in July to October surpass 0.5°C anomalies compared to the 1995–2015 baseline ( [[#Fernandes--2017|Fernandes et al., 2017]] ). Overall, current evidence suggests that peat carbon losses via fire have the potential to be equal to, or greater than, losses due to human peatland drainage and disturbance ( ''limited evidence'' , ''high agreement'' ) ( [[#Turetsky--2015|Turetsky et al., 2015]] ).

Regarding permafrost peatlands, studies differ, with some projecting a net loss and others a net gain of carbon ( ''medium evidence'' , ''low agreement'' ) ( [[#Estop-Aragonés--2018|Estop-Aragonés et al., 2018]] ; [[#Hugelius--2020|Hugelius et al., 2020]] ; [[#Loisel--2021|Loisel et al., 2021]] ; [[#Väliranta--2021|Väliranta et al., 2021]] ). In some permafrost peatlands, prolonged and warmer growing seasons due to climate change ( [[#2.3|Section 2.3.4.3.1]] ), along with increases in nitrogen deposition since 1850 ( [[#Lamarque--2013|Lamarque et al., 2013]] ), are promoting plant primary productivity. Other studies indicate that increased nitrogen-mediated sequestration could be exceeded by increased decomposition due to climate change-driven warming and fire ( ''medium evidence'' , ''low agreement'' ) ( [[#Natali--2012|Natali et al., 2012]] ; [[#Vonk--2015|Vonk et al., 2015]] ; [[#Keuper--2017|Keuper et al., 2017]] ; [[#Burd--2018|Burd et al., 2018]] ; [[#Estop-Aragonés--2018|Estop-Aragonés et al., 2018]] ; [[#Gallego-Sala--2018|Gallego-Sala et al., 2018]] ; [[#Serikova--2018|Serikova et al., 2018]] ; [[#Wild--2019|Wild et al., 2019]] ; [[#Chaudhary--2020|Chaudhary et al., 2020]] ; [[#Hugelius--2020|Hugelius et al., 2020]] ).

Any climate change- or human-driven degradation of peatlands will also entail losses in water storage ( ''limited evidence'' , ''high agreement'' ) ( [[#Wooster--2012|Wooster et al., 2012]] ; [[#Hirano--2015|Hirano et al., 2015]] ; [[#Cole--2019|Cole et al., 2019]] ; [[#Taufik--2019|Taufik et al., 2019]] ) and biodiversity ( [[#Harrison--2013|Harrison, 2013]] ; [[#Lampela--2017|Lampela et al., 2017]] ; [[#Renou-Wilson--2019|Renou-Wilson et al., 2019]] ). The environmental archive contained in peat that preserves records of vegetation, hydrology, climate change, pollution and/or human disturbances is also being lost as peatlands degrade ( [[#Kasischke--2006|Kasischke and Turetsky, 2006]] ; [[#MacDonald--2006|MacDonald et al., 2006]] ; [[#Turunen--2008|Turunen, 2008]] ; [[#Field--2009|Field et al., 2009]] ; [[#Flannigan--2009|Flannigan et al., 2009]] ; [[#Jones--2010|Jones and Yu, 2010]] ; [[#Kasischke--2010|Kasischke et al., 2010]] ; [[#Peterson--2010|Peterson et al., 2010]] ; [[#Finkelstein--2011|Finkelstein and Cowling, 2011]] ; [[#Rooney--2012|Rooney et al., 2012]] ; [[#Gallego-Sala--2013|Gallego-Sala and Colin Prentice, 2013]] ; [[#Lamarque--2013|Lamarque et al., 2013]] ; [[#Hirano--2014|Hirano et al., 2014]] ; [[#Brouns--2015|Brouns et al., 2015]] ; [[#Turetsky--2015|Turetsky et al., 2015]] ; [[#Miettinen--2016|Miettinen et al., 2016]] ; [[#Schneider--2016|Schneider et al., 2016]] ; [[#Cobb--2017|Cobb et al., 2017]] ; [[#Fernandes--2017|Fernandes et al., 2017]] ; [[#Itoh--2017|Itoh et al., 2017]] ; [[#Gallego-Sala--2018|Gallego-Sala et al., 2018]] ; [[#Greiser--2018|Greiser and Joosten, 2018]] ; [[#Ratcliffe--2018|Ratcliffe et al., 2018]] ; [[#Dadap--2019|Dadap et al., 2019]] ; [[#Leifeld--2019|Leifeld et al., 2019]] ; [[#Chaudhary--2020|Chaudhary et al., 2020]] ; [[#Hoyt--2020|Hoyt et al., 2020]] ; [[#Müller--2020|Müller and Joos, 2020]] ; [[#Qiu--2020|Qiu et al., 2020]] ; [[#Tangang--2020|Tangang et al., 2020]] ; [[#Taufik--2020|Taufik et al., 2020]] ; [[#Turetsky--2020|Turetsky et al., 2020]] ; [[#Chen--2021a|Chen et al., 2021a]] ; [[#Evans--2021|Evans et al., 2021]] ; [[#Loisel--2021|Loisel et al., 2021]] ; [[#Nelson--2021|Nelson et al., 2021]] ; [[#Qiu--2021|Qiu et al., 2021]] ).

<div id="2.5.2.9" class="h3-container"></div>

<span id="risks-to-polar-tundra-ecosystems"></span>
==== 2.5.2.9 Risks to Polar Tundra Ecosystems ====

<div id="h3-41-siblings" class="h3-siblings"></div>

For boreal–tundra systems, AR5 projected the transformation of species composition, land cover and permafrost extent, decreasing albedo and increasing GHG emissions ( ''medium confidence'' ). SR1.5 classified tundra and boreal forests as particularly vulnerable to degradation and encroachment by woody shrubs ( ''high confidence'' ). The SROCC projected climate-related changes to arctic hydrology, wildfires and abrupt thaw ( ''high confidence'' ) and the broad disappearance of arctic near-surface permafrost this century, with important consequences for global climate ( ''very high confidence'' ) ''.'' [https://www.ipcc.ch/report/ar6/wg2/chapter/chapter-2 Chapter 2] of AR6 has focused on new key findings about observed and projected changes in tundra vegetation and related hydrology, with implications for feedbacks to the climate system.

Due to the rapid warming at high northern latitudes, the Arctic tundra is one of the terrestrial biomes where climate change impacts are already clearly visible ( [[#Settele--2014|Settele et al., 2014]] ; [[#Uboni--2016|Uboni et al., 2016]] ). Climate models project that warming of the Arctic is likely to continue at more than double the global rate. Compared to the period 1995–2014, mean annual surface air temperatures in the Arctic tundra are projected to increase by 7.9°C–10°C by the end of the century in scenarios of high GHG emissions (RCP7.0 and RCP8.5). In scenarios of low GHG emissions (RCP1.9 and RCP2.6), the projected increase is 2.6°C–3.2°C ( [[#Lee--2021|Lee et al., 2021]] ). The Arctic is also projected to have amongst the largest increases in precipitation globally, but with ''high'' uncertainty. In contrast to climate change, LUC is projected to be very low in Arctic tundra systems ( [[#van%20Asselen--2013|van Asselen and Verburg, 2013]] ).

Models of vegetation response to climate project acceleration in the coming decades of observed increases in shrub dominance and boreal forest encroachment that have been driven by recent warming ( [[#Settele--2014|Settele et al., 2014]] ), leading to a shrinking of the area of tundra globally ( ''medium confidence'' ) ( [[#Mod--2016|Mod and Luoto, 2016]] ; [[#Gang--2017|Gang et al., 2017]] ). Simulating changes in tundra vegetation is complicated by permafrost dynamics (e.g., the formation of thaw ponds and draining of existing ponds), changes in precipitation and low nutrient availability (which may promote the abundance of graminoids) ( [[#van%20der%20Kolk--2016|van der Kolk et al., 2016]] ). Changes in vegetation, when combined with warming and increased precipitation effects on soil thawing and carbon cycling, are projected to modify GHG emissions and have biophysical feedbacks to regional and global climate. High uncertainty in modelled carbon cycle changes arises from differences between the vegetation models ( [[#Nishina--2015|Nishina et al., 2015]] ; [[#Ito--2016|Ito et al., 2016]] ). In addition, climate change is expected to strongly interact with other factors, such as fire, to further increase uncertainty in projections of tundra ecosystem function ( [[#Jiang--2017|Jiang et al., 2017]] ).

<div id="2.5.2.10" class="h3-container"></div>

<span id="committed-impacts-of-climate-change-on-terrestrial-ecosystems-and-implications-of-overshoot"></span>
==== 2.5.2.10 Committed Impacts of Climate Change on Terrestrial Ecosystems and Implications of Overshoot ====

<div id="h3-42-siblings" class="h3-siblings"></div>

Projections point to potentially large changes of canopy structure and composition within and across the terrestrial biomes in response to climate change and changes in atmospheric CO 2 . These changes will contribute to altered ecosystem carbon uptake and losses, biophysical climate feedbacks (Sections 2.3.2; 2.4.4; 2.5.3.2; 2.5.3.3. 2.5.3.4, 2.5.3.5, Figure 2.10, Table 2.4) and multiple other ecosystem services (Sections 2.5.3, 2.5.4) as well impacts on biodiversity (Sections 2.4.2, 2.4.3, 2.4.4, 2.4.5, 2.5.1.3, 2.5.1.4, 2.5.2, Figure Box 2.1.1, Table Box 2.1.1, Table SM2.4). Until now, most studies project changes over next decades until the end of this century.

However, there is an increasing body of literature that has found continued, longer-term responses of ecosystems to climate change, so-called ‘committed changes’, that arise from lags that exist in many systems. Many processes in ecosystems take more than a few decades to quasi-equilibrate to environmental changes. Therefore, the trends of changing vegetation cover identified in simulations of transient warming continue to show up in simulations that hold climate change at low levels of warming ( ''medium confidence'' ) ( [[#Boulton--2017|Boulton et al., 2017]] ; [[#Pugh--2018|Pugh et al., 2018]] ; [[#Scheiter--2020|Scheiter et al., 2020]] ). Such changes, which could tip ecosystems into an alternative state, could also be triggered by a ‘warming overshoot’ if global warming were to exceed a certain threshold, even if mean temperatures afterwards decline again ( [[#Albrich--2020a|Albrich et al., 2020a]] ).

For instance, even if warming achieved by 2100 remained constant after 2100, such committed responses continue to occur. These include: (1) continued Amazon forest loss ( [[#Boulton--2017|Boulton et al., 2017]] ), consistent with results in [[#Pugh--2018|Pugh et al. (2018)]] that found continued tropical forest cover loss across a range of models and simulation setups, and (2) across Africa, an increased shift towards woody C3 vegetation was found in equilibrium state, the overall response depending on the atmospheric CO 2 concentration ( [[#Scheiter--2020|Scheiter et al., 2020]] ). In [[#Pugh--2018|Pugh et al. (2018)]] , the opposite was found for boreal forest cover, which showed a strong committed increase. The committed changes in vegetation composition correspond to large committed changes in terrestrial carbon uptake and losses ( [[#Boulton--2017|Boulton et al., 2017]] ; [[#Pugh--2018|Pugh et al., 2018]] ; [[#Scheiter--2020|Scheiter et al., 2020]] ), and would plausibly also appear in other ecosystem functioning and services. These studies point to the importance of having not only a multi-decadal but also a multi-century perspective when exploring the impacts of political decisions on climate change mitigation taken now. Even if climate-warming targets are met, published evidence so far suggests that fundamental changes in some ecosystems are ''likely'' as these correspond to well-understood ecosystem physiological responses that trigger long-term changes in composition.

<div id="box-2.1" class="h2-container box-container"></div>

'''Box 2.1 | Assessing Past Projections of Ecosystem Change against Observations'''
<div id="h2-30-siblings" class="h2-siblings"></div>

To assess future climate change impacts on ecosystems, we use models to project their future distribution. Comparing the trends in the observed changes against the projections can help assess the strength of the model projections. In this box, we compare observed trends of changes in ecosystem structure to projections highlighted in previous IPCC reports, specifically AR3 ( [[#IPCC--2001|IPCC, 2001]] ), AR4 ( [[#Fischlin--2007|Fischlin et al., 2007]] ) and AR5 ( [[#Settele--2014|Settele et al., 2014]] ). We use this to assess how well the projections are matching up with observed changes. The map represents studies documenting observed changes in common plant functional groups (e.g., trees, grasses and shrubs). Studies documenting changes in plant functional groups were collated from published papers in natural and semi-natural areas. Studies were included if climate change or interactions between climate change and land use showed a causal link to the observed change (Table SM2.4). Studies were excluded if the changes were only from landscape/land use transformation (e.g., deforestation). In each paper, we recorded the geographical location and type of functional change, and noted the causes. Observed changes are plotted onto a biome map derived from the WWF ecoregions database ( [[#Olson--2001|Olson et al., 2001]] ). Trends in changing plant functional types are good indicators of potential biome shifts and are used to assess how observations match up with projections.

[[File:b08a5740a487993ed517554892a93800 IPCC_AR6_WGII_Figure_2_Box_2_1_1.png]]

'''Figure Box 2.1.1 |''' '''Observed changes in the distribution of plant functional types that are caused by climate change or a combination of land use and climate change.''' Shifts in plant functional types are indicative of shift in biome function and structure. Based upon studies listed in Table SM2.4 and section 2.4.

'''Table Box 2.1.1 |''' Comparison of projections on biome change from AR3, AR4 and AR5 ( [[#IPCC--2001|IPCC, 2001]] ; [[#Fischlin--2007|Fischlin et al., 2007]] ; [[#Settele--2014|Settele et al., 2014]] ), with observed changes in ecosystems as assessed in this report (see [[#2.4|Section 2.4]] , Figure Box 2.1.1, Table SM2.4). Observed changes marked in bold show good agreement with past projections; those in red show mismatch with observations and projections.

{| class="wikitable"
|-
! '''Biome'''

! '''AR3'''

! '''AR4'''

! '''AR5'''

! '''Observed trends 1990–2021'''

|-
| ''MTEs''

| Increased disturbance by fire and warming will cause a loss of unique habitats

| Loss of 65% of area due to warming. Increased fire frequencies will favour resprouting plants. An increase in grass dominance. Forest expansion within MTEs due to elevated CO 2 .

| Range contractions of all species

| Increase in water deficit '''and fire activity''' (Sections 2.4.3.6, 2.4.4.2) '''causing a decline in diversity;''' tree mortality (Fig. Box 2.1.1) with resprouting trees worst affected.

'''Increasing dominance of grasses (often alien).''' Increasing dominance of deciduous over evergreen species (Fig. Box 2.1.1).

|-
| ''Tundra''

| Tree and shrub encroachment into tundra

| Increased woody plant growth due to longer and warmer growing seasons and shrub tundra replacing dwarf tundra

Poleward expansion of tundra into polar desert and encroachment of coniferous trees into tundra

| Continued woody expansion in tundra regions with reduced surface albedo due to less snow and more woody cover

| '''Increase in woody shrub cover in tundra and expansion of boreal forest into tundra''' (Fig. Box 2.1.1, 2.4.3.4).

|-
| ''Boreal forest''

| Reduced productivity due to weather-related disturbances (e.g., increased fire risk).

Deciduous broadleaf tree encroachment into boreal forest.

| Extensive boreal tree spread into tundra.

Boreal forest dieback within boreal zone and contraction of boreal forest at southern ecotone with continental grasslands

|
| '''Expansion into tundra and upslope treeline advance''' ( [[#2.4.3.8|Section 2.4.3.8]] and Fig Box 2.1.1).

'''Increased mortality due to drought, fire, beetle infestations''' (Sections 2.4.3.8, 2.4.42.1, 2.4.4.3.1).

|-
| ''Tropical forest''

| Increasing CO 2 concentration would increase NPP

| Increases in forest productivity and biomass through increased CO 2 with localised decreases in the Amazon. Shift in forest species composition.

Expansion of forest area into mesic savanna.

| Shift in the climate envelope of moist tropical forests but forests are less likely to undergo major retractions or expansions than suggested in AR4

| '''Expansion of tropical forest into savannas in Africa, Asia, South America''' ( [[#2.4.3.7|Section 2.4.3.7]] , Fig. Box 2.1.1).

'''Forest biomass increases (though slowing)''' ( [[#2.4.4.4|Section 2.4.4.4]] ).

Forest degradation from drought, warming, fire and shorter residence time of trees ( [[#2.4.3.7|Section 2.4.3.7]] )

'''Shift in species composition''' towards species with more aridity-adapted traits ( [[#2.4.3.7|Section 2.4.3.7]] ).

|-
| ''Temperate forest''

| Forest decline and increased mortality

| Increase in tree mortality from drought-related declines.

A general increase of deciduous vegetation at the expense of evergreen vegetation is predicted at all latitudes.

|
| Map indicates a shift towards deciduous species in western North America (Fig. Box 2.1.1).

'''Tree death due to interactions of drought, pest outbreaks and fire''' (2.4.3.8, 2.4.4.2.1., 2.4.4.3.1)

|-
| ''Grasslands and savannas''

| Increasing CO 2 concentration will increase NPP

| Increased tree dominance in savannas and grasslands (from elevated CO 2 ), with C3 plants benefitting more than C4 plants

| Rising CO 2 will increase the likelihood of woodier states (but the transition will vary in different environments)

| '''Greening and encroachment across tropical and temperate savannas in Africa, Asia, Australia and America''' ( [[#2.4.3.5|Section 2.4.3.5]] ).

'''Expansion of trees into grasslands and advancement of tree lines.'''

Signs of increased C4 grass productivity in drought conditions. Increased C3 grass productivity ( [[#2.4.3.5|Section 2.4.3.5]] ).

|-
| ''Desert/arid shrublands''

|
| An increase in desert vegetation productivity was projected in southern Africa, the Sahel, central Australia, the Arabian Peninsula and parts of central Asia due to a positive impact of rising atmospheric CO 2

|
| '''Greening''' (increased leaf area index [LAI] and woody cover) and increased herbaceous production are occurring at desert–grassland interfaces (Cross-Chapter Paper 3)

|}

Assessment: There is high agreement between observations and projections of tree death in temperate and boreal forests, with current projections (AR6) indicating this trend will continue (Sections 2.4.4.3, 2.5.3.3, 2.5.4). Forest death is most widely recorded in central Europe and western North America (Fig. Box 2.1.1). There is also very high agreement between observations and projections of woody encroachment in savannas, grasslands and tundra, with projections also indicating that this trend is likely to continue (Sections 2.4.3.5, 2.4.3.9, 2.5.2.5, 2.5.2.9, 2.5.4). Observations of desert-greening show good agreement with earlier projections. Patterns of desertification are also occurring, although the geographical match between projections and observations shows moderate agreement, likely due to the strong role of land use in this process. Projections of tropical forest expansion into mesic savannas and boreal forest expansion into tundra also show agreement with the observations.

Projections of the future of Mediterranean shrublands, deserts, xeric shrublands and temperate grassy systems are limited, making assessment of this relationship less clear. It is also unclear, due to limited observations, how widespread a shift there is from deciduous forest species to evergreen forest species. Some observations suggest this is occurring, but it is not clear how widespread this change is and if the geographical pattern is as projected.

<div id="2.5.3" class="h2-container"></div>

<span id="risk-assessment-of-ecosystems-and-related-services"></span>
=== 2.5.3 Risk Assessment of Ecosystems and Related Services ===

<div id="h2-14-siblings" class="h2-siblings"></div>

<div id="2.5.3.1" class="h3-container"></div>

<span id="risks-in-protected-areas"></span>
==== 2.5.3.1 Risks in Protected Areas ====

<div id="h3-43-siblings" class="h3-siblings"></div>

National parks and other protected areas which, in June 2021, covered 15.7% of the global terrestrial area (UNEP-WCMC et al., 2021), conserve greater biodiversity than adjacent unprotected areas ( [[#Gray--2016|Gray et al., 2016]] ), and protect one-fifth of global vegetation carbon stocks and one-tenth of global soil carbon stocks ( [[#2.4.4.4|Section 2.4.4.4]] ). This section assesses climate change specifically in protected areas. Even though it is included in a part of the chapter on projected risks, it includes both observed exposure and projected risks to gather the information on protected areas into one place.

<div id="2.5.3.1.1" class="h4-container"></div>

<span id="observed-exposure-of-protected-areas"></span>
===== 2.5.3.1.1 Observed exposure of protected areas =====

<div id="h4-35-siblings" class="h4-siblings"></div>

In 2009, deforestation, agricultural expansion, overgrazing and urbanisation exposed one-third of the global protected area (6 million km 2 ) to intense human pressure, a 6% increase from 1993 ( [[#Venter--2016|Venter et al., 2016]] ; [[#Jones--2018|Jones et al., 2018]] ). The exposure to observed climate change has not yet been quantified for protected areas globally, but research has analysed the spatial patterns and magnitudes of observed changes for the 360,000 km 2 system of US national parks ( [[#Gonzalez--2018|Gonzalez et al., 2018]] ) including the first national park in the world. From 1895 to 2010, mean annual temperature of the US national park area increased at a rate of 1°C ± 0.2°C per century, double the rate of the whole USA, and precipitation decreased in 12% of the national park area, compared with 4% for the whole USA, due to a high fraction of US national park area being in the Arctic, at high elevations, and in the arid southwestern USA ( [[#Gonzalez--2018|Gonzalez et al., 2018]] ). In addition, analyses of weather-station measurements in and near six South African national parks found that the maximum temperature increased at a rate of 0.024°C ± 0.003°C yr -1 from 1960 to 2010 ( [[#Van%20Wilgen--2016|Van Wilgen et al., 2016]] ). While a substantial fraction of global protected area has been exposed to observed changes in human land cover, the global exposure to observed climate change is unquantified.

<div id="2.5.3.1.2" class="h4-container"></div>

<span id="projected-risks-in-protected-areas"></span>
===== 2.5.3.1.2 Projected risks in protected areas =====

<div id="h4-36-siblings" class="h4-siblings"></div>

Under a climate change scenario of ~3.5°C temperature increase by 2070, current climate could disappear from individual protected areas that comprise half the global protected area, and novel climates (climate conditions that are currently not present in an individual protected area) could emerge in half the global protected area ( [[#Hoffmann--2019b|Hoffmann et al., 2019b]] ). A lower-emissions scenario of ~1.5°C could reduce the disappearance of current climate conditions to 40% and the exposure to novel climates to 41% ( [[#Hoffmann--2019b|Hoffmann et al., 2019b]] ). Models project the highest exposure to novel climates in subtropical projected areas ( [[#Hoffmann--2020|Hoffmann and Beierkuhnlein, 2020]] ). Projected disappearance of current climate conditions in protected areas is most extensive in Africa, Oceania, and North and South America ( [[#Elsen--2020|Elsen et al., 2020]] ).

Projections indicate greater exposure of tropical rainforests, shrublands and grasslands, temperate conifer forests and grasslands, and tundra to novel climates ( [[#Hoffmann--2019b|Hoffmann et al., 2019b]] ; [[#Elsen--2020|Elsen et al., 2020]] ). A climate change scenario of ~3.5°C temperature increase by 2100 could expose 32% of the protected area in humid tropical forests (1.6 million km 2 in 2000) to climate that would be novel to humid tropical-forest protected areas; by 2050, the climate currently present in humid tropical-forest protected areas could disappear from 0.6 million km 2 (12% of the current total area) ( [[#Tabor--2018|Tabor et al., 2018]] ). High rates of deforestation and climate change combined could expose 2% of the humid tropical-forest protected area ( [[#Tabor--2018|Tabor et al., 2018]] ). Regional analyses under RCP8.5 also project the substantial disappearance of the current climate in protected areas in Bolivia, Chile and Peru ( [[#Fuentes-Castillo--2020|Fuentes-Castillo et al., 2020]] ), Canada, Mexico and the USA ( [[#Batllori--2017|Batllori et al., 2017]] ; [[#Holsinger--2019|Holsinger et al., 2019]] ), China ( [[#Zomer--2015|Zomer et al., 2015]] ), Europe ( [[#Nila--2019|Nila et al., 2019]] ) and Indonesia ( [[#Scriven--2015|Scriven et al., 2015]] ). Projected climate change could expose an extensive part of the global protected area to disappearing and novel climate conditions ( ''high confidence'' ) (Cross-Chapter Paper 1).

Continued climate change increases the risks to individual species and vegetation types in protected areas. Under a climate change scenario of 4°C temperature increase by 2100, the suitable climate for two species of baobab trees ( ''Adansonia perrieri'' and ''A. suarezensis'' ) in Madagascar could shift entirely out of the protected areas network ( [[#Vieilledent--2013|Vieilledent et al., 2013]] ). Other species and vegetation types at risk from the partial disappearance of suitable climate in protected areas include Atlantic Forest amphibians in Brazil ( [[#Lemes--2014|Lemes et al., 2014]] ), birds in Finland ( [[#Virkkala--2013|Virkkala et al., 2013]] ), birds and trees in Canada and Mexico ( [[#Stralberg--2020|Stralberg et al., 2020]] ), bog woodlands in Germany ( [[#Steinacker--2019|Steinacker et al., 2019]] ), butterflies and mammals in Egypt ( [[#Leach--2013|Leach et al., 2013]] ) and tropical dry forests in Mexico ( [[#Prieto-Torres--2016|Prieto-Torres et al., 2016]] ). Projected disappearance of suitable climate conditions in protected areas increase risks to the survival of species and vegetation types of conservation concern in tropical, temperate and boreal ecosystems ( ''high confidence'' ) (Cross-Chapter Paper 1).

Protected rivers, lakes and other freshwater protected areas require inter-catchment connectivity to maintain species and population movements ( [[#Bush--2014a|Bush et al., 2014a]] ; [[#Hermoso--2016|Hermoso et al., 2016]] ; [[#Thieme--2016|Thieme et al., 2016]] ), but dams and other barriers interrupt connectivity ( [[#Grill--2019|Grill et al., 2019]] ). Climate change could also reduce freshwater connectivity ( [[#2.3.3.3|Section 2.3.3.3]] ). Globally, over two-thirds of river reaches (by length) lack protected areas in their upstream catchments and nine-tenths of river reaches (by length) do not achieve full, integrated protection ( [[#Abell--2017|Abell et al., 2017]] ).

Terrestrial and freshwater protected areas can also serve as climate change refugia, that is, locations where suitable conditions may persist for the species into the future (e.g., [[#2.6.5.6|Section 2.6.5.6]] ). In Canada, Mexico and the USA, only a fraction of the protected area is located in potential climate change refugia under a 4°C temperature increase, estimated at 4% ( [[#Michalak--2018|Michalak et al., 2018]] ) to 7% ( [[#Batllori--2017|Batllori et al., 2017]] ). Potential refugia from biome shifts due to climate change under temperature increases of 1.8°C–3.4°C cover &lt;1% of the area of US national parks ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ), a fraction that diminishes to near zero when climate change is combined with habitat fragmentation due to LUC ( [[#Eigenbrod--2015|Eigenbrod et al., 2015]] ). Protected areas in boreal ecosystems could serve as refugia for species shifting north in Canada ( [[#Berteaux--2018|Berteaux et al., 2018]] ) and Finland ( [[#Lehikoinen--2019|Lehikoinen et al., 2019]] ). Invasive species, habitat loss and other disturbances in protected areas could be lower than in unprotected areas across Europe ( [[#Gallardo--2017|Gallardo et al., 2017]] ), specifically in Spain ( [[#Regos--2016|Regos et al., 2016]] ), and also in Sri Lanka ( [[#Kariyawasam--2020|Kariyawasam et al., 2020]] ). Protected areas conserve refugia from climate change under a temperature increase of 4°C, which is important for biodiversity conservation but is limited to &lt;10% of the current protected area ( ''medium confidence'' ).

<div id="2.5.3.2" class="h3-container"></div>

<span id="risks-to-ecosystems-and-services-from-wildfire"></span>
==== 2.5.3.2 Risks to Ecosystems and Services from Wildfire ====

<div id="h3-44-siblings" class="h3-siblings"></div>

<div id="2.5.3.2.1" class="h4-container"></div>

<span id="future-projections-of-wildfire-globally"></span>
===== 2.5.3.2.1 Future projections of wildfire globally =====

<div id="h4-37-siblings" class="h4-siblings"></div>

Continued climate change under high-emission scenarios that increase global temperature ~4°C by 2100 could increase global burned area by 50% ( [[#Knorr--2016b|Knorr et al., 2016b]] ) to 70% ( [[#Kloster--2017|Kloster and Lasslop, 2017]] ) and global mean fire frequency by ~30% ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ), with increases on one-third ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ) to two-thirds ( [[#Moritz--2012|Moritz et al., 2012]] ) and decreases on one-fifth ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Moritz--2012|Moritz et al., 2012]] ) of land globally. Lower emissions that would limit the global temperature increase to &lt;2°C would reduce projected increases of global burned area to 30% ( [[#Lange--2020|Lange et al., 2020]] ) to 35% ( [[#Kloster--2017|Kloster and Lasslop, 2017]] ) and projected increases of fire frequency to ~20% ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Huang--2015|Huang et al., 2015]] ). Continued climate change could further lengthen fire weather seasons ( [[#Ranasinghe--2021|Ranasinghe et al., 2021]] ). Models combining projected climate change with potential agricultural expansion project decreases in total burned area ( [[#Huang--2015|Huang et al., 2015]] ; [[#Knorr--2016b|Knorr et al., 2016b]] ; [[#Park--2021|Park et al., 2021]] ). The area of projected increases in burned area and fire frequency due solely to continued climate change is higher for the world as a whole than the area of projected decreases ( ''medium evidence'' , ''medium agreement'' ).

Increased wildfire due to continued climate change increases risks of tree mortality (Sections 2.5.2.6, 2.5.2.7, 2.5.3.2), biome shifts ( [[#2.5.2.2|Section 2.5.2.2]] ) and carbon emissions (Sections 2.5.2.10, 2.5.3.4). Wildfire and biome shifts under a projected climate change of 4°C above the pre-industrial period, combined with international trade and transport, cause high risks from invasive species across one-sixth of the global area including extensive high-biodiversity regions ( [[#Early--2016|Early et al., 2016]] ).

Wildfire risks to people include death and destruction of their homes, respiratory illnesses from smoke ( [[#Ford--2018|Ford et al., 2018]] ; [[#Machado-Silva--2020|Machado-Silva et al., 2020]] ), post-fire flooding from areas exposed by vegetation loss and degraded water quality due to increased sediment flow ( [[#Dahm--2015|Dahm et al., 2015]] ) and the chemical precursors of carcinogenic trihalomethanes when water is later chlorinated for drinking ( [[#2.5.3|Section 2.5.3.7]] ) ( [[#Uzun--2020|Uzun et al., 2020]] ). Under RCP8.5 and shared socioeconomic pathway SSP3 (high population growth, slow urbanisation), the number of people living in fire-prone areas could increase by three-quarters to 720 million in 2100, in a projected global population of 12.4 billion people ( [[#Knorr--2016b|Knorr et al., 2016b]] ). Lower emissions under RCP4.5 could reduce the number of people at risk by 70 million. In these projections, human population growth increases human exposure to wildfires more than increases in burned area ( [[#Knorr--2016b|Knorr et al., 2016b]] ). A global temperature increase &lt;2°C could increase global population exposure to wildfire by ~30% ( [[#Lange--2020|Lange et al., 2020]] ). Increased wildfire under continued climate change increases the probability of human exposure to fire and risks to public health ( ''medium evidence'' , ''high agreement'' ).

<div id="2.5.3.2.2" class="h4-container"></div>

<span id="future-projections-of-wildfire-in-high-risk-areas"></span>
===== 2.5.3.2.2 Future projections of wildfire in high-risk areas =====

<div id="h4-38-siblings" class="h4-siblings"></div>

Regions identified by multiple global analyses as being at a high risk of increased burned area, fire frequency and fire weather include: the Amazon ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Huang--2015|Huang et al., 2015]] ; [[#Knorr--2016b|Knorr et al., 2016b]] ; [[#Burton--2018|Burton et al., 2018]] ; [[#Abatzoglou--2019|Abatzoglou et al., 2019]] ), Mediterranean Europe ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Burton--2018|Burton et al., 2018]] ; [[#Abatzoglou--2019|Abatzoglou et al., 2019]] ), the Arctic tundra ( [[#Moritz--2012|Moritz et al., 2012]] ; [[#Flannigan--2013|Flannigan et al., 2013]] ), Western Australia ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Burton--2018|Burton et al., 2018]] ; [[#Abatzoglou--2019|Abatzoglou et al., 2019]] ) and the western USA ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Moritz--2012|Moritz et al., 2012]] ; [[#Knorr--2016b|Knorr et al., 2016b]] ). Higher-resolution spatial projections indicate high risks of increased wildfire in the Amazon, Australia, boreal ecosystems, Mediterranean Europe and the USA with climate change ( ''medium evidence'' , ''medium agreement'' ).

In the Amazon, climate change under RCP8.5, combined with high deforestation, could double the area of high fire probability ( [[#Fonseca--2019|Fonseca et al., 2019]] ), double the burned area by 2050 ( [[#Brando--2020|Brando et al., 2020]] ), increase the burned area by 400–2800% by 2100 ( [[#Le%20Page--2017|Le Page et al., 2017]] ) and increase fire intensity by 90% ( [[#De%20Faria--2017|De Faria et al., 2017]] ). Lower GHG emissions (RCP4.5) and reduced deforestation could reduce the risk of fires to a one-fifth increase in the area of high fire probability ( [[#Fonseca--2019|Fonseca et al., 2019]] ) and a 100–500% increase in burned area by 2100 ( [[#Le%20Page--2017|Le Page et al., 2017]] ). Moreover, increased fire, deforestation and drought, acting via vegetation–atmosphere feedbacks, increase the risk of extensive forest dieback and potential biome shifts of up to half of the Amazon rainforest to grassland, a tipping point that could release an amount of carbon that would substantially increase global emissions ( [[#Oyama--2003|Oyama and Nobre, 2003]] ; [[#Sampaio--2007|Sampaio et al., 2007]] ; [[#Lenton--2008|Lenton et al., 2008]] ; [[#Nepstad--2008|Nepstad et al., 2008]] ; [[#Malhi--2009|Malhi et al., 2009]] ; [[#Settele--2014|Settele et al., 2014]] ; [[#Lyra--2016|Lyra et al., 2016]] ; [[#Zemp--2017a|Zemp et al., 2017a]] ; [[#Zemp--2017b|Zemp et al., 2017b]] ; [[#Brando--2020|Brando et al., 2020]] ). Continued climate change, combined with deforestation, increases risks of wildfire and extensive forest dieback in the Amazon rainforest ( ''robust evidence'' , ''high agreement'' ).

In Australia, climate change under RCP8.5 increases the risk of pyro-convective fire by 20–40 days in rangelands of Western Australia, South Australia and the Northern Territory ( [[#Dowdy--2019|Dowdy et al., 2019]] ). Pyro-convective fire conditions could reach more frequently into the more populated areas of New South Wales, particularly at the start of the austral summer ( [[#Di%20Virgilio--2019|Di Virgilio et al., 2019]] ). GCMs do not agree, however, on the areas of projected fire increase in New South Wales ( [[#Clarke--2019|Clarke and Evans, 2019]] ). Increases in heat and potential increases in wildfire threaten the existence of temperature montane rainforest in Tasmania, Australia ( [[#Mariani--2019|Mariani et al., 2019]] ).

In Mediterranean Europe, climate change of 3°C of warming could double or triple the burned area whereas keeping the temperature increase to 1.5°C could limit the increase in burned area to 40–50% ( [[#Turco--2018|Turco et al., 2018]] ). Under RCP8.5, the frequency of heat-induced fire weather could increase by 30% ( [[#Ruffault--2020|Ruffault et al., 2020]] ). Severe fire followed by drought could cause biome shifts of forest to non-forest ( [[#Batllori--2019|Batllori et al., 2019]] ) and tree mortality &gt;50% ( [[#Dupire--2019|Dupire et al., 2019]] ).

In the Arctic tundra, boreal forests and northern peatlands, including permafrost areas, climate change under the scenario of a 4°C temperature increase could triple the burned area in Canada ( [[#Boulanger--2014|Boulanger et al., 2014]] ), double the number of fires in Finland ( [[#Lehtonen--2016|Lehtonen et al., 2016]] ), increase the lightning-driven burned area by 30–250% ( [[#Veraverbeke--2017|Veraverbeke et al., 2017]] ; [[#Chen--2021a|Chen et al., 2021a]] ), push half of the area of tundra and boreal forest in Alaska above the burning threshold temperature and double the burned area in Alaska ( [[#Young--2017a|Young et al., 2017a]] ). Thawing of Arctic permafrost due to a projected temperature of 4°C and the resultant wildfires could release 11–200 GtC which could substantially exacerbate climate change ( [[#2.5.2.9|Section 2.5.2.9]] ).

In the USA, climate change under RCP8.5 could increase the burned area by 60–80% by 2049 ( [[#Buotte--2019|Buotte et al., 2019]] ) and the number of fires with an area &gt;50 km 2 by 300–400% by 2070 ( [[#Barbero--2015|Barbero et al., 2015]] ). In montane forests, climate change under RCP8.5 increases the risk of fire-facilitated conversion of ~7% of forest to non-forest by 2050 ( [[#Parks--2019|Parks et al., 2019]] ). In California, climate change under a scenario of a 4°C temperature increase could double fire frequency in some areas ( [[#Mann--2016|Mann et al., 2016]] ), but emission reductions that limit the temperature increase to ~2°C could keep this from increasing ( [[#Westerling--2011|Westerling et al., 2011]] ). Carbon dioxide fertilisation and increased temperature under climate change could increase invasive grasses and wildfire in desert ecosystems of the southwestern USA where wildfire has historically been absent or infrequent, and increase the mortality of the sparse tree cover (Horn and St. Clair, 2017; [[#Klinger--2017|Klinger and Brooks, 2017]] ; [[#Syphard--2017|Syphard et al., 2017]] ; [[#Moloney--2019|Moloney et al., 2019]] ; [[#Sweet--2019|Sweet et al., 2019]] ).

In summary, under a high-emission scenario that increases global temperature 4°C by 2100, climate change could increase the global burned area by 50–70% and the global mean fire frequency by ~30%, with increases on one- to two-thirds and decreases on one-fifth of global land ( ''medium confidence'' ). Lower emissions that would limit the global temperature increase to &lt;2°C would reduce projected increases of burned area to ~35% and projected increases of fire frequency to ~20% ( ''medium confidence'' ). Increased wildfire, combined with erosion due to deforestation, could degrade water supplies ( ''high confidence'' ). For ecosystems with an historically low fire frequency, a projected 4°C rise in global temperature increases risks of fire, contributing to potential tree mortality and conversion of over half the Amazon rainforest to grassland and thawing of the Arctic permafrost that could release 11–200 GtC that could substantially exacerbate climate change ( ''medium confidence'' ).

<div id="2.5.3.3" class="h3-container"></div>

<span id="risks-to-ecosystems-and-services-from-tree-mortality"></span>
==== 2.5.3.3 Risks to Ecosystems and Services from Tree Mortality ====

<div id="h3-45-siblings" class="h3-siblings"></div>

Under continued climate change, increased temperature, aridity, drought, wildfire ( [[#2.5.3.2|Section 2.5.3.2]] ) and insect infestations ( [[#2.4.4.3.3|Section 2.4.4.3.3]] ) will tend to increase tree mortality across many parts of the world ( [[#McDowell--2020|McDowell et al., 2020]] ). Loss of boreal and temperate forest to fire, wind and bark beetles could cause more negative than positive effects for most ecosystem services, including carbon storage to regulate climate change (Sections 2.4.4.3, 2.5.2.6, 2.5.2.7, 2.5.3.4), water supply for people ( [[#2.5.3.6.1|Section 2.5.3.6.1]] ), timber production and other forest products (Chapter 5) and protection from hazards ( [[#Thom--2016|Thom and Seidl, 2016]] ). In addition, deforestation in tropical and temperate forests can increase local temperatures by 0.3°C–2°C ( [[#Hesslerová--2018|Hesslerová et al., 2018]] ; [[#Lejeune--2018|Lejeune et al., 2018]] ; [[#Zeppetello--2020|Zeppetello et al., 2020]] ) and this effect can extend up to 50 km ( [[#Cohn--2019|Cohn et al., 2019]] ).

In Amazon rainforests, the relatively lower buffering capacity for plant moisture during drought increases the risk of tree mortality and, combined with increased heat from climate change and fire from deforestation, the possibility of a tipping point of extensive forest dieback and a biome shift to grassland ( [[#Oyama--2003|Oyama and Nobre, 2003]] ; [[#Sampaio--2007|Sampaio et al., 2007]] ; [[#Lenton--2008|Lenton et al., 2008]] ; [[#Nepstad--2008|Nepstad et al., 2008]] ; [[#Malhi--2009|Malhi et al., 2009]] ; [[#Salazar--2010|Salazar and Nobre, 2010]] ; [[#Settele--2014|Settele et al., 2014]] ; [[#Lyra--2016|Lyra et al., 2016]] ; [[#Zemp--2017b|Zemp et al., 2017b]] ; [[#Brando--2020|Brando et al., 2020]] ). This could occur at a 4°C–5°C temperature increase above that of the pre-industrial period ( [[#Salazar--2010|Salazar and Nobre, 2010]] ). Under RCP8.5, half the Amazon tropical evergreen forest could turn into grassland through drought-induced tree mortality and wildfire, but lower emissions (RCP4.5) could limit this loss to ~5% ( [[#Lyra--2016|Lyra et al., 2016]] ). The decline in precipitation due to reduced evapotranspiration inputs after forest loss could cause additional Amazon forest loss of one-quarter to one-third ( [[#Zemp--2017a|Zemp et al., 2017a]] ). Similarly, in Guinean tropical deciduous forest in Africa, climate change under RCP8.5 could increase mortality 700% by 2100 or 400% under lower emissions (RCP4.5; ( [[#Claeys--2019|Claeys et al., 2019]] ). These projections indicate risks of climate change-induced tree mortality reducing tropical forest areas in Africa and South America by up to half under a 4°C increase above the pre-industrial period, but a lower projection of a 2°C increase could limit the projected increases in tree mortality ( ''robust evidence'' , ''high agreement'' ).

Temperate and boreal forests possess greater diversity of physiological traits related to plant hydraulics, so they are more buffered against drought than tropical forests ( [[#Anderegg--2018|Anderegg et al., 2018]] ). Nevertheless, in temperate forests, drought-induced tree mortality under RCP8.5 could cause the loss of half the Northern Hemisphere conifer forest area by 2100 ( [[#McDowell--2016|McDowell et al., 2016]] ). In the western USA, under RCP8.5, one-tenth of forest area is highly vulnerable to drought-induced mortality by 2050 ( [[#Buotte--2019|Buotte et al., 2019]] ). In California, increased evapotranspiration in Sierra Nevada conifer forests increases the potential fraction of the area at risk of tree mortality by 15–20% per degree Celsius ( [[#Goulden--2019|Goulden and Bales, 2019]] ). In Alaska, fire-induced tree mortality from climate change under RCP8.5 could reduce the extent of spruce forest ( ''Picea'' sp.) by 8–44% by 2100 ( [[#Pastick--2017|Pastick et al., 2017]] ). Under RCP8.5, tree mortality from drought, wildfire and bark beetles could reduce the timber productivity of boreal forests in Canada by 2100 below the current levels ( [[#Boucher--2018|Boucher et al., 2018]] ; [[#Chaste--2019|Chaste et al., 2019]] ; [[#Brecka--2020|Brecka et al., 2020]] ). In Tasmania, projected increases in wildfire ( [[#Fox-Hughes--2014|Fox-Hughes et al., 2014]] ) increase the risk of mortality of mesic vegetation ( [[#Harris--2018b|Harris et al., 2018b]] ) and threaten the disappearance of the long-lived endemic pencil pine ( ''Athrotaxis cupressoides'' ) ( [[#Holz--2015|Holz et al., 2015]] ; [[#Worth--2016|Worth et al., 2016]] ) and temperate montane rainforest ( [[#Mariani--2019|Mariani et al., 2019]] ). These projections indicate risks of climate change-induced tree mortality reducing some temperate forest areas by half under emissions scenarios of 2.5°C–4°C above the pre-industrial period ( ''medium evidence'' , ''high agreement'' ).

<div id="2.5.3.4" class="h3-container"></div>

<span id="risk-to-terrestrial-ecosystem-carbon-stocks"></span>
==== 2.5.3.4 Risk to Terrestrial-Ecosystem Carbon Stocks ====

<div id="h3-46-siblings" class="h3-siblings"></div>

Globally, increasing atmospheric CO 2 enhances the terrestrial sink but temperature increases constrain it, reflecting the biological process understanding highlighted in previous IPCC reports ( ''high confidence'' ). Analyses of atmospheric inversion model output and spatial climate data indicate a sensitivity of net ecosystem productivity to CO 2 fertilisation of 3.1 ± 0.1 Gt to 8.1 ± 0.3 Gt per 100 ppm CO 2 (~1°C increase) and a sensitivity to temperature of -0.5 ± 0.2 Gt to -1.1 ± 0.1 Gt per degree Celsius ( [[#Fernandez-Martinez--2019|Fernandez-Martinez et al., 2019]] ). The future of the global land carbon sink ( [[#2.4.4.4|Section 2.4.4.4]] ) nevertheless remains highly uncertain because (i) of regionally complex interactions of climate change and changes in atmospheric CO 2 with vegetation, soil and aquatic processes, (ii) episodic events such as heat waves or droughts (and related impacts through mortality, wildfire or insects, pests and diseases) ( [[#2.5.3.2|Section 2.5.3.2]] , 2.5.3.3) are so far only incompletely captured in carbon cycle models, (iii) the legacy effects from historic LUC and environmental changes are incompletely captured but likely to decline in future and (iv) lateral carbon transport processes such as the export of inland waters and erosion are incompletely understood and modelled ( [[#Pugh--2019a|Pugh et al., 2019a]] ; [[#Friedlingstein--2020|Friedlingstein et al., 2020]] ; [[#Krause--2020|Krause et al., 2020]] ; [[#Canadell--2021|Canadell et al., 2021]] ).

Enhanced carbon losses from terrestrial systems further limit the available carbon budget for global warming staying below 1.5°C ( [[#Rogelj--2018|Rogelj et al., 2018]] ). Analyses of satellite remote sensing and ground-based observations have indicated that, between 1982 and 2015, the CO 2 fertilisation effect has already declined, implying a negative climate system feedback ( [[#Wang--2020c|Wang et al., 2020c]] ). Peatlands, permafrost regions and tropical ecosystems are particularly vulnerable due to their large carbon stocks, in combination with over-proportional warming, increases in heat waves and droughts and/or a complex interplay of climate change and increasing atmospheric CO 2 (Sections 2.5.2.8, 2.5.2.9, 2.5.3.2).

Model projections suggest a reduction of permafrost extent and potentially large carbon losses for all warming scenarios ( [[#Canadell--2021|Canadell et al., 2021]] ). Already a mean temperature increase of 2°C could reduce the total permafrost area extent by about 5–20% by 2100 ( [[#Comyn-Platt--2018|Comyn-Platt et al., 2018]] ; [[#Yokohata--2020|Yokohata et al., 2020]] ). Associated CO 2 losses in the order of 15 Gt up to nearly 70 Gt by 2100 have been projected across a number of modelling studies ( [[#Schneider%20von%20Deimling--2015|Schneider von Deimling et al., 2015]] ; [[#Comyn-Platt--2018|Comyn-Platt et al., 2018]] ; [[#Yokohata--2020|Yokohata et al., 2020]] ). Limiting the global temperature increase to 1.5°C versus 2°C could reduce projected permafrost CO 2 losses by 2100 by 24.2 Gt (median, calculated for a 3-m depth) ( [[#Comyn-Platt--2018|Comyn-Platt et al., 2018]] ). Losses are possibly underestimated in the studies that consider only the upper permafrost layers. Likewise, the actual committed carbon loss may well be larger (e.g., eventually a loss of approx. 40% of today’s permafrost area extent if climate is stabilised at 2°C above pre-industrial levels) due to the long time scale of warming in deep permafrost layers ( [[#Chadburn--2017|Chadburn et al., 2017]] ). It is not known at which level of global warming an abrupt permafrost collapse (estimated to enhance CO 2 emissions by 40% in 2300 in a high-emissions scenario) compared to gradual thaw ( [[#Turetsky--2020|Turetsky et al., 2020]] ) would have to be considered an important additional risk. Large uncertainties arise also from interactions with changes in surface hydrology and/or northward migrating woody vegetation as climate warms, which could dampen or even reverse projected net carbon losses in some regions ( [[#McGuire--2018a|McGuire et al., 2018a]] ; [[#Mekonnen--2018|Mekonnen et al., 2018]] ; [[#Pugh--2018|Pugh et al., 2018]] ). Overall, there is ''low confidence'' on how carbon–permafrost interactions will affect future carbon cycle and climate, although net carbon losses and thus positive (amplifying) feedbacks are ''likely'' (Sections 2.5.2.10, 2.5.3.5) ( [[#Shukla--2019|Shukla et al., 2019]] ). See also WGI AR6 ( [[#Canadell--2021|Canadell et al., 2021]] ) for a discussion on impacts of higher-emission and warming scenarios.

Peatland carbon is estimated as about 550–1000 Gt in northern latitudes (many of these peatlands would be found in permafrost regions) ( [[#Turetsky--2015|Turetsky et al., 2015]] ; [[#Nichols--2019|Nichols and Peteet, 2019]] ) and &gt;100 Gt in tropical regions ( [[#Turetsky--2015|Turetsky et al., 2015]] ; [[#Dargie--2017|Dargie et al., 2017]] ). For both northern mid- and high-latitude and tropical peatlands, a shift from contemporary CO 2 sinks to sources were simulated in high-warming scenarios ( [[#Wang--2018a|Wang et al., 2018a]] ; [[#Qiu--2020|Qiu et al., 2020]] ). Due to the lack of large-scale modelling studies, there is ''low confidence'' for climate change impacts on peat carbon uptake and emissions. The largest risk to tropical peatlands is expected to arise from drainage and conversion to forestry or agriculture, which would outpace the impacts of climate change ( [[#Page--2016|Page and Baird, 2016]] ; [[#Leifeld--2019|Leifeld et al., 2019]] ; [[#Cooper--2020|Cooper et al., 2020]] ). The magnitude of possible carbon losses is uncertain, however, and depends strongly on socioeconomic scenarios (Sections 2.4.3.8, 2.4.4.2; 2.4.4.4.2, 2.5.2.8).

For tropical and subtropical regions, the interplay of atmospheric CO 2 with precipitation and temperature becomes of particular importance for future carbon uptake, since in warm and dry environments, elevated CO 2 fosters plants with C3 photosynthesis and enhances their water-use efficiency relative to C4 species ( [[#Moncrieff--2014a|Moncrieff et al., 2014a]] ; [[#Midgley--2015|Midgley and Bond, 2015]] ; [[#Knorr--2016a|Knorr et al., 2016a]] ). As a consequence, enhanced woody cover is expected to occur in the future, especially in mesic savannas, while in xeric savannas an increase in woody cover would occur in regions with enhanced precipitation ( [[#Criado--2020|Criado et al., 2020]] ). Even though semiarid regions have dominated the global trend in land CO 2 uptake in recent decades ( [[#Ahlström--2015|Ahlström et al., 2015]] ), so far, most studies that investigated future climate change impacts on savanna ecosystems have concentrated on changes in the extent of land area affected (2.5.2.5) rather than on carbon cycling, with ''medium confidence'' for increasing woody cover:grass ratios ( [[#Moncrieff--2014a|Moncrieff et al., 2014a]] ; [[#Midgley--2015|Midgley and Bond, 2015]] ; [[#Moncrieff--2016|Moncrieff et al., 2016]] ; [[#Criado--2020|Criado et al., 2020]] ). Increases in woody vegetation in what is now grass-dominated would possibly come with a carbon benefit, for instance, it was found that a broad range of future climate and CO 2 changes would enhance vegetation carbon storage on Australian savannas ( [[#Scheiter--2015|Scheiter et al., 2015]] ). Results from a number of field experiments indicate, however, that impacts on total ecosystem carbon storage may be smaller due to a loss in below-ground carbon ( [[#Coetsee--2013|Coetsee et al., 2013]] ; [[#Wigley--2020|Wigley et al., 2020]] ). [[#Nunez--2021|Nunez et al. (2021)]] critique existing incentives to promote the invasion of non-native trees into treeless areas as a means of carbon sequestration, raising doubts about the effects on fire, albedo, biodiversity and water yield (see Box 2.2).

Substantial climate change-driven impacts on tropical tree cover and vegetation type are projected in all studies, irrespective of whether or not the degree amounts to a forest “dieback” (Sections 2.4.3.6, 2.4.4.3, 2.5.2.6, 2.5.3.3) ( [[#Davies-Barnard--2015|Davies-Barnard et al., 2015]] ; [[#Wu--2016a|Wu et al., 2016a]] ; [[#Zemp--2017a|Zemp et al., 2017a]] ; [[#Canadell--2021|Canadell et al., 2021]] ) . Accordingly, models also suggest a continuation of tropical forests acting as carbon sinks ( [[#Huntingford--2013|Huntingford et al., 2013]] ; [[#Mercado--2018|Mercado et al., 2018]] ). A recent study combining field plot data with statistical models ( [[#Hubau--2020|Hubau et al., 2020]] ) indicates that, in the Amazonian and possibly also in the African forest, the carbon sink in above-ground biomass already declined in the three decades up to 2015. This trend is distinct in the Amazon whereas data from Africa suggests a possible decline after 2010. The authors estimate the vegetation carbon sink in 2030–2040 to decline to zero±0.205 PgC yr -1 in the Amazon and to 0.26±0.215 PgC yr -1 in Africa (a loss of 14% compared to the present). Their results suggest that, over time, CO 2 fertilisation is outweighed by the impacts of higher temperatures and drought that enhance tree mortality and diminish growth. The degree of thermal resilience of tropical forests is still uncertain, however ( [[#Sullivan--2020|Sullivan et al., 2020]] ).

The lack of simulation studies that seek to quantify all important interacting factors (CO 2 , drought and fire) for future carbon cycling in savannas and tropical forests and the apparent disagreement between trends projected in models compared to data-driven estimates result in ''low confidence'' regarding the direction or magnitude of carbon flux and pool-size changes. Similar to tropical peatlands, given projected human population growth and socioeconomic changes, the continued conversion of forests and savannas into agricultural or pasture systems ''very likely'' poses a significant risk of rapid carbon loss which will amplify the climate change-induced risks substantially ( ''high confidence'' ) (2.5.2.10, 2.5.3.5) ( [[#Aragao--2014|Aragao et al., 2014]] ; [[#Searchinger--2015|Searchinger et al., 2015]] ; [[#Aleman--2016|Aleman et al., 2016]] ; [[#Nobre--2016|Nobre et al., 2016]] ).

The impacts of climate-induced altered animal composition and trophic cascades on land-ecosystem carbon cycling globally are as yet unquantified ( [[#Schmitz--2018|Schmitz et al., 2018]] ), even though climate change is expected to lead to shifts in consumer–resource interactions that also contribute to losses of top predators or top herbivores (Sections 2.4.2.2, 2.5.1.3, 2.5.4; ( [[#Lurgi--2012|Lurgi et al., 2012]] ; [[#Damien--2019|Damien and Tougeron, 2019]] ). Cascading trophic effects triggered by top predators or the largest herbivores propagate through food webs and reverberate through to the functioning of whole ecosystems, notably altering productivity, carbon and nutrient turnover and net carbon storage ( ''medium confidence'' ) ( [[#Wilmers--2016|Wilmers and Schmitz, 2016]] ; [[#Sobral--2017|Sobral et al., 2017]] ; [[#Stoner--2018|Stoner et al., 2018]] ). Across different field experiments, the ecosystem consequences of the presence or absence of herbivores and carnivores have been found to be quantitatively as large as the effects of other environmental change drivers such as warming, enhanced CO 2 , fire and variable nitrogen deposition ( ''medium confidence'' ) ( [[#Hooper--2012|Hooper et al., 2012]] ; [[#Smith--2015|Smith et al., 2015]] ). Some local and regional modelling experiments have begun to explore animal impacts on vegetation dynamics and carbon and nutrient cycling ( [[#Pachzelt--2015|Pachzelt et al., 2015]] ; [[#Dangal--2017|Dangal et al., 2017]] ; [[#Berzaghi--2019|Berzaghi et al., 2019]] ). Turnover rate is the chief factor that determines future land-ecosystem carbon dynamics and hence carbon–climate feedbacks ( [[#Friend--2014|Friend et al., 2014]] ). To improve projections, it is imperative to better quantify the broader role of carnivores, grazers and browsers and the way these interact in global studies of how ecosystems respond to climate change.

<div id="2.5.3.5" class="h3-container"></div>

<span id="feedbacks-between-ecosystems-and-climate"></span>
==== 2.5.3.5 Feedbacks between Ecosystems and Climate ====

<div id="h3-47-siblings" class="h3-siblings"></div>

The possibility of feedbacks and interactions between climate drivers and biological systems or ecological processes was identified as a significant emerging issue in AR5, and has since also been highlighted in the SRCCL and the SR1.5. It is virtually certain that land cover changes affect regional and global climate through changes to albedo, evapotranspiration and roughness ( ''very high confidence'' ) ( [[#Perugini--2017|Perugini et al., 2017]] ). There is growing evidence that biosphere-related climate processes are being affected by climate change in combination with disturbance and LULCC ( ''high confidence'' ) ( [[#Jia--2019|Jia et al., 2019]] ). It is virtually certain that land surface change caused by disturbances such as forest fires, hurricanes, phenological changes, insect outbreaks and deforestation affect carbon, water and energy exchanges, thereby influencing weather and climate ( ''very high confidence'' ) (Table 2.4; Figure 2.10) ( [[#Bright--2013|Bright et al., 2013]] ; [[#Brovkin--2013|Brovkin et al., 2013]] ; [[#Naudts--2016|Naudts et al., 2016]] ; [[#Prăvălie--2018|Prăvălie, 2018]] ).

<div id="_idContainer050" class="Figure"></div>

[[File:d856dff730369fa63df2c84157b62ea0 IPCC_AR6_WGII_Figure_2_010.png]]

'''Figure 2.10 | Terrestrial ecosystem feedbacks, which affect the Earth’s climate system dynamics.''' Perturbations and implications for climate system dynamics (warming/cooling) are shown for the three global forest biomes (adapted from Figure 5 in ( [[#Prăvălie--2018|Prăvălie, 2018]] ). The strength of the mechanism is estimated in general terms, based on the magnitude of carbon storage and evaporative cooling processes that characterise each forest biome ( [[#Bonan--2008|Bonan, 2008]] ). Carbon storage includes forest biomass, without accounting for carbon dynamics in soil, peat and underlying permafrost deposits. Implications of bio-geochemical shifts were only estimated in relation to the intensification of the carbon cycle and increase in biomass at high latitudes, assuming nitrogen availability for the stoichiometric demands of forest vegetation.

Feedbacks can be positive or negative (i.e., amplify or dampen the original forcing), vary spatially and seasonally, and act over large geographic areas and long time periods (more than decades), making them difficult to observe and quantify directly ( [[#Schimel--2015|Schimel et al., 2015]] ; [[#Canadell--2021|Canadell et al., 2021]] ). Due to the positive impacts of CO 2 on vegetation growth and ecosystem carbon storage ( ''high confidenc'' e) (Sections 2.4.4.4, 2.5.5.4) ( [[#Canadell--2021|Canadell et al., 2021]] ), the associated climate feedback is negative (i.e., increased removal of atmospheric CO 2 and dampened warming, compared to an absence of the feedback). By contrast, projected global losses of carbon in warmer climates ( [[#Canadell--2021|Canadell et al., 2021]] ) imply a positive climate feedback. WGI ( [[#Canadell--2021|Canadell et al., 2021]] ) assesses an overall increase in land carbon uptake through the 21st century. However, the overall strength of the carbon cycle–climate feedback remains very uncertain. One of the underlying reasons may be complex interactions with ecosystem water balance and nitrogen and phosphorous availability, which are poorly constrained by observational evidence and incompletely captured in ESMs ( [[#2.5.2.10|Section 2.5.2.10]] ) ( [[#Huntzinger--2017|Huntzinger et al., 2017]] ; [[#Canadell--2021|Canadell et al., 2021]] ).

Land ecosystems contribute substantially to global emissions of nitrous oxide and methane . As with CO 2 , these emissions respond both directly and indirectly to atmospheric CO 2 concentration and climate change, and this gives rise to potential additional bio-geochemical feedbacks in the climate system. A large part of these emissions stem from land and water management, such as fertilizer application, rice production, aquaculture or animal husbandry ( [[#Jia--2019|Jia et al., 2019]] ). However, nearly 60% of total nitrous oxide emissions (in 2007–2016) has been estimated to stem from natural ecosystems, especially in the Tropics ( [[#Tian--2019|Tian et al., 2019]] ; [[#Canadell--2021|Canadell et al., 2021]] ), while freshwater wetlands and peatlands are estimated to contribute between 83% (top-down estimates) and 40% (bottom-up estimates) of total natural CH 4 (and 31 and 20% of total methane emissions, respectively) for the period 2008–2017 ( [[#Canadell--2021|Canadell et al., 2021]] ). Median CH 4 emissions from northern-latitude wetlands in 2100 were estimated to be 12.1 and 13.5 PgC in emission scenarios leading to 1.5°C and 2°C warming, respectively ( [[#Comyn-Platt--2018|Comyn-Platt et al., 2018]] ). Likewise, global warming has been attributed to soil N 2 O emission increases since the pre-industrial period of 0.8 (0.3–1.3) TgN yr -1 ( [[#Tian--2020|Tian et al., 2020]] ). Overall, climate feedbacks from future altered land ecosystem emissions of CH 4 or N 2 O are uncertain, but are expected to be small ( [[#Canadell--2021|Canadell et al., 2021]] ).

Changes in regional biodiversity are integral parts of ecosystem–climate feedback loops, including and beyond carbon cycle processes (Figure 2.10; Table 2.4). For instance, the impacts of climate-induced altered animal composition and trophic cascades on ecosystem carbon turnover (see Sections 2.4.4.4, 2.5.3.4) could be a substantive contribution to carbon–climate feedbacks ( ''low confidence'' ). Additional surface–atmosphere feedbacks that arise from changes in vegetation cover and subsequently altered albedo, evapotranspiration or roughness (often summarised as biophysical feedbacks) can be regionally relevant and could amplify or dampen vegetation cover changes ( [[#Jia--2019|Jia et al., 2019]] ).

Climate-induced shifts towards forests in what is currently tundra would be expected to reduce regional albedo especially in spring, but also during parts of winter when trees are snow-free (whereas tundra vegetation would be covered in snow), which amplifies warming regionally ( ''high confidence'' ) ( [[#Perugini--2017|Perugini et al., 2017]] ; [[#Jia--2019|Jia et al., 2019]] ). Trees would also enhance momentum absorption compared to low tundra vegetation, thus impacting surface–atmosphere mixing of latent and sensible heat fluxes ( [[#Jia--2019|Jia et al., 2019]] ). Boreal forests insulate and stabilize permafrost and reduce fluctuations of ground temperature: the amplitude of variation of ground surface temperatures was 28°C at a forested site, compared to 60°C in nearby grassland ( [[#2.5.2.7|Section 2.5.2.7]] ) ( [[#Bonan--1989|Bonan, 1989]] ; [[#Stuenzi--2021a|Stuenzi et al., 2021a]] ; [[#Stuenzi--2021b|Stuenzi et al., 2021b]] ). Likewise, a shift in moist tropical forests towards vegetation with drought-tolerant traits could possibly reduce evapotranspiration, increase albedo, alter heat transfer at the surface and lead to a negative feedback to precipitation ( [[#2.5.2.6|Section 2.5.2.6]] ) ( [[#Jia--2019|Jia et al., 2019]] ). In savannas, restoration of woody vegetation has been shown to enhance cloud formation and precipitation in response to enhanced transpiration and turbulent mixing, leading to a positive feedback on woody cover ( [[#Syktus--2016|Syktus and McAlpine, 2016]] ). While this has not yet been systematically explored, similar feedbacks might also emerge from a CO 2 -induced woody cover increase in savannas ( ''low confidence'' ) ( [[#2.5.2.5|Section 2.5.2.5]] ).

Since biophysical feedbacks can contribute to both surface temperature warming or cooling, analyses so far suggest that, on a global scale, the net impact on climate change is small ( [[#Perugini--2017|Perugini et al., 2017]] ; [[#Jia--2019|Jia et al., 2019]] ), unless these feedbacks also accelerate vegetation mortality and lead to substantive carbon losses ( [[#Zemp--2017a|Zemp et al., 2017a]] ; [[#Lemordant--2019|Lemordant and Gentine, 2019]] ). More than one-third of the Earth’s land surface has at least 50% of its evapotranspiration regulated by vegetation, and in some regions between 40 and &gt;80% of the land’s evaporated water is returned to land as precipitation. Locally, both directly human-mediated and climate change-mediated changes in vegetation cover can therefore notably affect annual average freshwater availability to human societies, especially if negative feedbacks amplify the reduction of vegetation cover, evapotranspiration and precipitation ( ''medium confidence'' ) ( [[#Keys--2016|Keys et al., 2016]] ; [[#Keys--2018|Keys and Wang-Erlandsson, 2018]] ).

Since AR5, freshwater ecosystems (lakes, reservoirs, rivers and ponds) have been increasingly recognised as important sources of GHG emissions (CO 2 , CH 4 and N 2 O) into the atmosphere. Key mechanisms which contribute to rising GHG emissions from freshwater ecosystems are the temperature imbalance between photosynthesis and respiration (respiration increases more than photosynthesis with rising temperature), CO 2 and CH 4 emissions from exposed sediments during droughts, increased transport of matter from land to water, changes in water retention time in rivers and lakes and the effects of temperature on lake stratification and anoxia that favour CH 4 emissions.

DelSontro et al. (2018) assembled the largest global data set to date on emission rates from lakes of CO 2 , CH 4 and N 2 O and found that they co-vary with lake size and trophic state. They estimated that moderate global increases in eutrophication of lakes could translate to 5–40% increases in the GHG effect in the atmosphere. Moreover, they estimated that GHG emissions from lakes and impoundments in past decades accounted for 1.25–2.30 PgCO 2 yr -1 ( [[#DelSontro--2018|DelSontro et al., 2018]] ), thus around 20% of global CO 2 emissions from the burning of fossil fuels (9.4 PgCO 2 yr -1 ) ( [[#Friedlingstein--2020|Friedlingstein et al., 2020]] ).

Global warming will strongly enhance freshwater CH 4 emissions through a disproportionate increase in ebullition (gas flux) by 6–20% per 1°C increase in water temperature ( [[#Aben--2017|Aben et al., 2017]] ). It can be expected that ongoing eutrophication enhanced by climate change-related increases in the release of sediment nutrients and the loading of organic carbon and nutrients from catchments will enhance CH 4 ebullition on a global scale ( [[#Aben--2017|Aben et al., 2017]] ; [[#DelSontro--2018|DelSontro et al., 2018]] ; [[#Bartosiewicz--2019|Bartosiewicz et al., 2019]] ; [[#Beaulieu--2019|Beaulieu et al., 2019]] ; [[#Sanches--2019|Sanches et al., 2019]] ). The strongest increase in ebullition is expected in shallow waters where sediment temperatures are strongly related to atmospheric temperature ( [[#Aben--2017|Aben et al., 2017]] ). Given that small ponds and shallow lakes are the most abundant freshwater ecosystems globally, these may become hot spots of CH 4 ebullition in the future ( [[#Aben--2017|Aben et al., 2017]] ). On average, CH 4 , CO 2 and N 2 O account for 75, 23 and 2% of the total CO 2 -equivalent emissions, respectively, in lakes ( [[#DelSontro--2018|DelSontro et al., 2018]] ).

Furthermore, the exposure of lake and river sediments during droughts activates the decomposition of buried organic carbon. In dry river beds, mineralisation of buried organic matter is likely to increase with climate change as anoxic sediments are oxygenated downwards during drying, along with pulses of microbial activity following re-wetting of desiccated sediment. Conservative estimates indicate that adding emissions from exposed sediments of dry inland waters across diverse ecosystem types and climate zones to current global estimates of CO 2 emissions could result in a 6% (~0.12 PgC yr −1 ) increase of total inland water CO 2 emission rates covering streams and rivers (334 mmol m -2 day -1 ), lakes and reservoirs (320 mmol m -2 day -1 ) and small ponds (148 mmol m -2 day -1 ) ( [[#Marcé--2019|Marcé et al., 2019]] ; [[#Keller--2020|Keller et al., 2020]] ).

Overall, uncertainty as to the quantity of carbon fluxes within freshwater ecosystems and between terrestrial and freshwater systems, and subsequent emissions to the atmosphere remains very ''high'' ( [[#Raymond--2013|Raymond et al., 2013]] ; [[#Catalán--2016|Catalán et al., 2016]] ; [[#Stanley--2016|Stanley et al., 2016]] ; [[#Evans--2017|Evans et al., 2017]] ; [[#Drake--2018|Drake et al., 2018]] ; [[#Seekell--2018|Seekell et al., 2018]] ; [[#Sanches--2019|Sanches et al., 2019]] ; [[#Bodmer--2020|Bodmer et al., 2020]] ; [[#Keller--2020|Keller et al., 2020]] ; [[#Canadell--2021|Canadell et al., 2021]] ) (see Table SM2.1.). Projections of carbon fluxes are, for example, challenged by the complex interaction between rising water temperatures, loss of ice, changes in hydrology, ecosystem productivity, increased extreme events and variation in terrestrial-matter transport. While we are still short of empirical data, particularly in the Tropics ( [[#DelSontro--2018|DelSontro et al., 2018]] ), improvements in sensor technology ( [[#Eugster--2011|Eugster et al., 2011]] ; [[#Gonzalez-Valencia--2014|Gonzalez-Valencia et al., 2014]] ; [[#Maeck--2014|Maeck et al., 2014]] ; [[#Delwiche--2015|Delwiche et al., 2015]] ) and the use of statistically robust survey designs ( [[#Beaulieu--2016|Beaulieu et al., 2016]] ; [[#Wik--2016|Wik et al., 2016]] ) have improved the accuracy of measurements of GHG emissions in freshwater ecosystems. Global networks such as the Global Lakes Ecological Observatory Network (GLEON) increasingly allow a global view of carbon fluxes, thereby improving estimates of the contribution of freshwater ecosystems to global GHG emissions to the atmosphere.

In summary for freshwater systems, Drake et al. (2018) aggregated contemporary estimates of CO 2 and CH 4 emissions from freshwater ecosystems with global estimates made by [[#Raymond--2013|Raymond et al. (2013)]] , and arrived at an estimate of 3.9 PgC yr -1 . Rivers and streams accounted for 85% and lakes and reservoirs for 15% of the emissions ( [[#Raymond--2013|Raymond et al., 2013]] ). This trend will continue under scenarios of nutrient loading to inland waters over the next century where increased CH 4 emission of inland water has an atmospheric impact of 1.7–2.6 PgC/CO 2 -eq y −1 , which is equivalent to 18–33% of annual CO 2 emissions from burning fossil fuels ( ''medium evidence'' , ''medium agreement'' ) ( [[#Beaulieu--2019|Beaulieu et al., 2019]] ). For comparison, annual uptake of CO 2 in land ecosystems is estimated as 3.4 (± 0.9) PgC yr −1 ( [[#Friedlingstein--2020|Friedlingstein et al., 2020]] ). The freshwater numbers combine CO 2 and CH 4 and are thus not directly comparable. However, they are indicative of the importance of better accounting for freshwater systems in global carbon budgets.

'''Table 2.4 |''' Terrestrial and freshwater ecosystem feedbacks which affect the Earth’s climate system dynamics, according to ( [[#Prăvălie--2018|Prăvălie, 2018]] ).

{| class="wikitable"
|-
! '''Perturbation'''

! '''Implications for warming/feedback mechanism'''

'''The Earth’s climate system dynamics'''

|-
| ''Phenological changes''

(sections 2.4.2.4, 2.4.2.5)

| Increased primary productivity and plant growth with CO 2 fertilisation ( [[#Mao--2016|Mao et al., 2016]] ; [[#Wang--2018a|Wang et al., 2018a]] ); increasing growing season length ( [[#Peñuelas--2009|Peñuelas et al., 2009]] ; [[#Barichivich--2013|Barichivich et al., 2013]] ); reduced diurnal temperature range through evapotranspiration (mid latitudes) and albedo (high latitudes) caused by vegetation greening ( [[#Jeong--2011|Jeong et al., 2011]] ); increased CO 2 storage in biomass (cooling) ( [[#Keenan--2014|Keenan et al., 2014]] ); reduced albedo in snow-covered regions as canopies become taller and darker (warming); increased evapotranspiration, a key component of the global water cycle and energy balance which influences global rainfall, temperature and atmospheric motion ( [[#Zeng--2017|Zeng et al., 2017]] )

|-
| ''Insect outbreaks''

(sections 2.4.4.2

| Reduced carbon uptake and storage (warming); increased surface albedo (cooling) ( [[#Landry--2016|Landry et al., 2016]] ); increased CO 2 emissions (warming); decreased LAI and gross primary productivity ( [[#Ghimire--2015|Ghimire et al., 2015]] ), leading to reduced evapotranspiration and increased land surface temperature ( [[#Bright--2013|Bright et al., 2013]] )

|-
| ''Range shifts''

(sections 2.4.2.1, 2.4.2.2, 2.4.2.3, 2.4.2.5, 2.4.3)

| Reduced albedo in snow-covered regions as trees expand polewards (warming) ( [[#Chae--2015|Chae et al., 2015]] ); enhanced permafrost thawing; expansion of insect outbreak range, increasing forest impact ( [[#Pureswaran--2018|Pureswaran et al., 2018]] ); biome-dependent changes in albedo and evapotranspiration regimes ( [[#Naudts--2016|Naudts et al., 2016]] ); reduction in snow and ice albedo in freshwater due to loss of ice (warming) ( [[#Lang--2018|Lang et al., 2018]] )

|-
| ''Die-off and large-scale mortality events''

(sections 2.4.2.2, 2.4.4.3)

| Decreased GPP; decline in carbon storage (warming); increased CO 2 emissions; increased solar radiation, reduced soil moisture and higher surface runoff; albedo effects ( [[#Lewis--2011|Lewis et al., 2011]] ; [[#Prăvălie--2018|Prăvălie, 2018]] )

|-
| ''Deforestation''

(sections 2.4.3.6, 2.4.3.7)

| Reduced carbon storage (warming) ( [[#Pugh--2019a|Pugh et al., 2019a]] ); increase in (regional) surface air temperature due to reduced evaporation (less cooling); increased albedo in high-latitude systems (regional radiative cooling) ( [[#Loranty--2014|Loranty et al., 2014]] ); increased air temperature and diurnal temperature variation ( [[#Alkama--2016|Alkama and Cescatti, 2016]] ), locally and globally ( [[#Winckler--2019|Winckler et al., 2019]] ); reduced precipitation ( [[#Perugini--2017|Perugini et al., 2017]] ); decreased biogenic volatile organic compounds (BVOC) and aerosol emissions (warming through direct and indirect aerosol effects; cooling associated with reduction in atmospheric methane ( [[#Jia--2019|Jia et al., 2019]] )

|-
| ''Forest degradation''

(sections 2.4.3.6, 2.4.3.7)

| Reduced carbon storage (warming) ( [[#de%20Paula--2015|de Paula et al., 2015]] ; [[#Bustamante--2016|Bustamante et al., 2016]] ; [[#de%20Andrade--2017|de Andrade et al., 2017]] ; [[#Mitchard--2018|Mitchard, 2018]] )

|-
| ''Fragmentation''

| Carbon losses because biomass is less developed at forest edges ( [[#Pütz--2014|Pütz et al., 2014]] ; [[#Chaplin-Kramer--2015|Chaplin-Kramer et al., 2015]] ; [[#Haddad--2015|Haddad et al., 2015]] )

|-
| ''Air pollution''

| Decreased plant productivity, transpiration and carbon sequestration in forests with lower biomass due to ozone toxicity ( [[#Sitch--2007|Sitch et al., 2007]] ; [[#Ainsworth--2012|Ainsworth et al., 2012]] ); increased (regional) productivity due to increase in diffuse solar radiation caused by terrestrial aerosols ( [[#Xie--2021|Xie et al., 2021]] )

|-
| ''Declining populations of megafauna''

| Changes to physical and chemical properties of organic matter, soils and sediments influence carbon uptake and storage ( [[#Schmitz--2018|Schmitz et al., 2018]] ); increased or decreased carbon storage biomass and carbon storage, with differences across biomes determined by floristic structure and animal size ( [[#Bello--2015|Bello et al., 2015]] ; [[#Osuri--2016|Osuri et al., 2016]] ; [[#Peres--2016|Peres et al., 2016]] ; [[#He--2017|He et al., 2017]] ; [[#Berzaghi--2018|Berzaghi et al., 2018]] ; [[#Schmitz--2018|Schmitz et al., 2018]] ; [[#He--2019|He et al., 2019]] )

|-
| ''Fire''

(sections 2.4.4.2, 2.5.3.2)

| Increased carbon and aerosol emissions ( [[#van%20der%20Werf--2017|van der Werf et al., 2017]] ); surface warming ( [[#Liu--2019b|Liu et al., 2019b]] ); albedo effect dependent on ecosystem and species-level traits ( [[#Rogers--2015|Rogers et al., 2015]] ; [[#Chen--2018a|Chen et al., 2018a]] ) (initial albedo decreases post-fire; increased albedo where snow exposure is increased by canopy removal and species composition changes during recovery); black carbon deposition on snow and sea ice (short-term) ( [[#Randerson--2006|Randerson et al., 2006]] ); indirect increases in carbon emissions due to soil erosion ( [[#Caon--2014|Caon et al., 2014]] )

|-
| ''Changes in forest composition'' (sections 2.4.3.6, 2.4.3.7, 2.5.2.6, 2.5.2.7)

| Reduced carbon storage due to the decline of biomass (warming) ( [[#McIntyre--2015|McIntyre et al., 2015]] )

|-
| ''Woody encroachment in non-forested ecosystems''

(sections 2.4.3.3, 2.4.3.4, 2.4.3.5, 2.5.2.3, 2.5.2.4, 2.5.2.5, Box 2.1)

| Reduced production, increased water use, reduced albedo and altered land–atmosphere feedbacks; increased carbon storage in woody savannas ( [[#Zhou--2017|Zhou et al., 2017]] ; [[#Mureva--2018|Mureva et al., 2018]] ); uncertain feedbacks to the carbon cycle (some suggest an increase, others a decrease)

|-
| ''NPP shifts''

(section 2.4.4.5)

| Reduced albedo following high-latitude expansion of trees caused by photosynthetic enhancement of growth (cooling); increased photosynthesis and net ecosystem production (NEP) ( [[#Fernandez-Martinez--2019|Fernandez-Martinez et al., 2019]] ); increased NPP in nutrient-limited ecosystems due to increased nitrogen deposition from agriculture and combustion ( [[#Du--2018|Du and de Vries, 2018]] ; [[#Schulte-Uebbing--2018|Schulte-Uebbing and de Vries, 2018]] ); nutrient-limited lakes are likely to become less productive, while nutrient-rich lakes are likely to become more productive due to warming-induced prolongation of stable stratification ( [[#Adrian--2016|Adrian et al., 2016]] ; [[#Kraemer--2017|Kraemer et al., 2017]] )

|-
| ''Bio-geochemical shifts''

| Decline in carbon storage due to nitrogen limitation in nutrient-limited systems (warming) ( [[#Reich--2014|Reich et al., 2014]] ; [[#Wieder--2015|Wieder et al., 2015]] ); increased carbon storage on land ( [[#Peñuelas--2013|Peñuelas et al., 2013]] ) and in lakes ( [[#Heathcote--2015|Heathcote et al., 2015]] ; [[#Mendonça--2017|Mendonça et al., 2017]] ); increase in CO 2 and CH 4 emissions from freshwater ecosystems due to increased eutrophication ( [[#DelSontro--2018|DelSontro et al., 2018]] ), the imbalance between losses and gains of CO 2 by photosynthesis and respiration (the metabolic theory of ecology), enhanced emissions from exposed river and lake sediments during droughts and re-wetting ( [[#Marcé--2019|Marcé et al., 2019]] ; [[#Keller--2020|Keller et al., 2020]] ), enhanced CH 4 ebullition of seasonally hypoxic lakes ( [[#Aben--2017|Aben et al., 2017]] ; [[#DelSontro--2018|DelSontro et al., 2018]] ; [[#Bartosiewicz--2019|Bartosiewicz et al., 2019]] ; [[#Beaulieu--2019|Beaulieu et al., 2019]] ; [[#Sanches--2019|Sanches et al., 2019]] ) and increased transfer of organic carbon from land to water (particularly in permafrost areas) ( [[#Wauthy--2018|Wauthy et al., 2018]] )

|}

<div id="2.5.3.6 " class="h3-container"></div>

<span id="risks-to-freshwater-ecosystem-services-drinking-water-fisheries-and-hydropower"></span>
==== 2.5.3.6 Risks to Freshwater Ecosystem Services: Drinking Water, Fisheries and Hydropower ====

<div id="h3-48-siblings" class="h3-siblings"></div>

AR5 named water supply and biodiversity as freshwater ecosystem services vulnerable to climate change. We discuss the risks to these and to additional services identified by model projections based both on climate-change scenarios ( [[#Schröter--2005|Schröter et al., 2005]] ; [[#Boithias--2014|Boithias et al., 2014]] ; [[#Huang--2019|Huang et al., 2019]] ; [[#Jorda-Capdevila--2019|Jorda-Capdevila et al., 2019]] ) and on the Common International Classification of Ecosystem Services ( ''high confidence'' ) (CICES, 2018). The effects of floods, droughts, permafrost and glacier-melting on global changes in water quality, particularly with respect to contamination with pollutants, are described in [[IPCC:Wg2:Chapter:Chapter-4#4.2.6|Section 4.2.6]] .

<div id="2.5.3.6.1" class="h4-container"></div>

<span id="risks-to-the-quantity-and-quality-of-drinking-water"></span>
===== 2.5.3.6.1 Risks to the quantity and quality of drinking water =====

<div id="h4-39-siblings" class="h4-siblings"></div>

Forests and other vegetated ecosystems assist the production of drinkable water by facilitating the infiltration of rainfall and snowfall into the ground, where water either moves through the saturated soil zone to supply streams and other surface waters or infiltrates further to recharge groundwater aquifers ( [[#Ellison--2012|Ellison et al., 2012]] ; [[#Bonnesoeur--2019|Bonnesoeur et al., 2019]] ). Globally, 4 billion people depend on forested watersheds for drinking water ( [[#Mekonnen--2016|Mekonnen and Hoekstra, 2016]] ). [[IPCC:Wg2:Chapter:Chapter-4|Chapter 4]] assesses the physical science of water supply, including precipitation, runoff and hydrology as well as the social aspects of human water use. This section assesses the ecological aspects of risks to freshwater supplies for people.

Diminished vegetation cover following wildfires ( [[#2.5.3.2|Section 2.5.3.2]] ) and tree mortality ( [[#2.5.3.3|Section 2.5.3.3]] ) can reduce long-term water infiltration, increase soil erosion and flash floods and release sediment that degrades drinking water quality. Widlfires increase impacts of extreme precipitation events due to climate change, which contribute to increased surface runoff and hence increased risks of land erosion, landslides and flooding ( [[#Ebel--2012|Ebel et al., 2012]] ; [[#Robinne--2020|Robinne et al., 2020]] ). Under current conditions, nearly half the global land area is at a moderate-to-high risk of water scarcity due to wildfires ( [[#Robinne--2018|Robinne et al., 2018]] ; [[#Robinne--2020|Robinne et al., 2020]] ). From 1984 to 2014, wildfires in the western USA affected 6–11% of stream and river length ( [[#Ball--2021|Ball et al., 2021]] ). Under a high-emissions scenario of a 3.5°C temperature increase, post-fire erosion across the western USA could double sedimentation and degrade drinking water quality in one-third of watersheds by 2050 ( [[#Sankey--2017|Sankey et al., 2017]] ). In Brazil, post-fire vegetation loss tends to increase runoff, reduce infiltration and reduce groundwater recharge and flow of springs ( [[#Rodrigues--2019|Rodrigues et al., 2019]] ). Runoff from wildfires can contain DOC precursors for the formation of carcinogenic trihalomethanes during chlorination of water for drinking ( [[#Uzun--2020|Uzun et al., 2020]] ) as well as chromium, mercury, selenium and other toxic trace metals ( [[#Burton--2016|Burton et al., 2016]] ; [[#Burton--2019|Burton et al., 2019]] ).

Net effects of deforestation and afforestation on runoff and water supply depend on local factors, leading to conflicting evidence of effects of land cover change ( [[#Ellison--2012|Ellison et al., 2012]] ; [[#Chen--2021b|Chen et al., 2021b]] ), but combinations of climate change and deforestation are projected to reduce water flows ( [[#Olivares--2019|Olivares et al., 2019]] ). In southern Thailand, the combination of the conversion of forest to rubber plantations and a one-third increase in rainfall could increase erosion and sediment load by 15% ( [[#Trisurat--2016|Trisurat et al., 2016]] ). In the watershed that supplies São Paulo, Brazil, afforestation could increase water quantity and quality ( [[#Ferreira--2019|Ferreira et al., 2019]] ). In most regions with dry or Mediterranean subtropical climates, projected climate change can reduce surface water and groundwater resources ( [[#Doell--2015|Doell et al., 2015]] ). In northeast Spain, reduced precipitation and vegetation cover under the high-emissions scenario of a 3.5°C temperature increase could reduce drinking water supplies by half by 2100 ( [[#Bangash--2013|Bangash et al., 2013]] ).

Changes in algal biomass development and the spread of cyanobacteria blooms, related to global warming, resemble those triggered by eutrophication with the well-known negative effects on the services lakes provide, particularly for drinking water provision and recreation ( ''robust evidence'' , ''high agreement'' , ''high confidence'' ) ( [[#Carvalho--2013|Carvalho et al., 2013]] ; [[#Adrian--2016|Adrian et al., 2016]] ; [[#Gozlan--2019|Gozlan et al., 2019]] ).

Based on a 10% increase in precipitation, ( [[#de%20Wit--2016|de Wit et al., 2016]] ) estimated an increased mobilisation of organic carbon from soils to freshwaters of at least 30%, demonstrating the importance of climate wetting for the carbon cycle. Browning negatively affects the taste of drinking water and this may be difficult to address ( [[#Kothawala--2015|Kothawala et al., 2015]] ; [[#Kritzberg--2020|Kritzberg et al., 2020]] ). It also often reduces attractiveness for recreational purposes, especially swimming ( [[#Arthington--2003|Arthington and Hadwen, 2003]] ; [[#Keeler--2015|Keeler et al., 2015]] ). Based on a worst-case climate scenario until 2030, ( [[#Weyhenmeyer--2016|Weyhenmeyer et al., 2016]] ) projected an increase in the browning of lakes and rivers in boreal Sweden by a factor of 1.3. The chemical character of DOM, as modified by climate change ( [[#Kellerman--2014|Kellerman et al., 2014]] ), determines its amenability to removal by water treatment ( [[#Ritson--2014|Ritson et al., 2014]] ). Therefore, in order to provide safe and acceptable drinking water, more advanced, more expensive and more energy/resource-intensive technical solutions may be required ( [[#Matilainen--2010|Matilainen et al., 2010]] ).

In summary, climate change increases risks to the integrity of watersheds and the provision of safe, acceptable freshwater to people ( ''medium evidence'' , ''medium agreement'' ).

<div id="2.5.3.6.2" class="h4-container"></div>

<span id="risks-to-freshwater-fisheries-and-biodiversity"></span>
===== 2.5.3.6.2 Risks to freshwater fisheries and biodiversity =====

<div id="h4-40-siblings" class="h4-siblings"></div>

Climate change will increase water temperatures and decrease dissolved oxygen levels ( [[#2.3.1|Section 2.3.1]] ), impacting freshwater fisheries which form an important ecosystem service ( [[#Vári--2022|Vári et al., 2022]] ). People living in the vicinity of cold lakes will be affected by projected losses of ice. In a worst-case scenario (an air temperatures increase of 8°C), 230,400 lakes and 656 million people in 50 countries will be impacted ( [[#Reid--2019|Reid et al., 2019]] ; [[#Sharma--2019|Sharma et al., 2019]] ). Winter ice-fishing ( [[#Orru--2014|Orru et al., 2014]] ), transportation via ice roads ( [[#Prowse--2011|Prowse et al., 2011]] ) and cultural activities ( [[#Magnuson--2014|Magnuson and Lathrop, 2014]] ) are ecosystem services at stake from the ongoing loss of lake ice.

Eutrophication of central European lakes has wiped out a significant proportion of the endemic fish fauna ( [[#Vonlanthen--2012|Vonlanthen et al., 2012]] ), so climate-induced further eutrophication is expected to represent an additional threat to fish fauna and commercial fisheries ( [[#Ficke--2007|Ficke et al., 2007]] ). Given that the ecological consequences of lake warming may be especially strong in the Tropics ( [[#2.3.1|Section 2.3.1.1]] ), ecosystem services may be most affected there. Tropical lakes support important fisheries ( [[#Lynch--2016a|Lynch et al., 2016a]] ; [[#McIntyre--2016|McIntyre et al., 2016]] ) that provide a critical source of nutrition to adjacent human populations. These lakes are especially prone to the loss of deep-water oxygen due to warming, with adverse consequences for the productivity of fisheries and for biodiversity ( ''medium evidence'' , ''medium agreement'' ) ( [[#Lewis%20Jr--2000|Lewis Jr, 2000]] ; [[#Van%20Bocxlaer--2012|Van Bocxlaer et al., 2012]] ).

Tropical lakes tend to be hotspots of freshwater biodiversity ( [[#Vadeboncoeur--2011|Vadeboncoeur et al., 2011]] ; [[#Brawand--2014|Brawand et al., 2014]] ; [[#Sterner--2020|Sterner et al., 2020]] ); ancient tropical lakes such as Malawi, Tanganyika, Victoria, Titicaca, Towuti and Matano hold thousands of animal species found nowhere else ( [[#Vadeboncoeur--2011|Vadeboncoeur et al., 2011]] ). While biodiversity and several ecosystem services can be considered synergistic (food webs, tourism and of aesthetic and spiritual value) ( [[#Langhans--2019|Langhans et al., 2019]] ), others can be considered antagonistic in case of a strong ecosystem service demand (such as water abstraction, water use and food security in terms of overexploitation). Here, the balance between biodiversity and ecosystem services is key ( [[#Langhans--2019|Langhans et al., 2019]] ), where biodiversity can be integrated into water policy by means of integrated water resource management (IWRM) towards NbS ( [[#Ligtvoet--2017|Ligtvoet et al., 2017]] )

<div id="2.5.3.6.3" class="h4-container"></div>

<span id="risks-to-hydropower-and-erosion-control"></span>
===== 2.5.3.6.3 Risks to hydropower and erosion control =====

<div id="h4-41-siblings" class="h4-siblings"></div>

River banks, riparian vegetation and macrophyte beds play important roles in erosion control through reducing current velocities, increasing sedimentation and reducing turbidity ( [[#Madsen--2001|Madsen et al., 2001]] ). Rates of flow in rivers affect inland navigation ( [[#Vári--2022|Vári et al., 2022]] ). Changing seasonality in snow-dominated basins is expected to enhance hydropower production in winter but decrease it during summer ( [[#Doell--2015|Doell et al., 2015]] ). Glacier melt changes hydrological regimes, sediment transport and bio-geochemical and contaminant fluxes from rivers to oceans, profoundly influencing ecosystem services that glacier-fed rivers provide, particularly the provision of water for agriculture, hydropower and consumption ( [[#Milner--2017|Milner et al., 2017]] ). Loss of glacial mass and snowpack has already impacted flow rates, quantities and seasonality (Chapter 4, in this report) ( [[#Hock--2019|Hock et al., 2019]] ). Meltwater yields from glacier ice are likely to increase in many regions during the next decades but decrease thereafter, as glaciers become smaller and smaller and finally disappear ( [[#Hock--2019|Hock et al., 2019]] ).

<div id="2.5.4" class="h2-container"></div>

<span id="key-risks-to-terrestrial-and-freshwater-ecosystems-from-climate-change"></span>
=== 2.5.4 Key Risks to Terrestrial and Freshwater Ecosystems from Climate Change ===

<div id="h2-15-siblings" class="h2-siblings"></div>

Among numerous risks to terrestrial and freshwater ecosystems from climate change, this chapter identified five phenomena as the most fundamental risks of climate change to ecosystem integrity and the ecosystem services that support human well-being that are also quantified sufficiently to estimate risk thresholds with at least ''medium'' confidence : Biodiversity loss (global losses of species from ecosystems), ecosystem structure change, increased tree mortality, increased wildfire, and ecosystem carbon losses and (Table 2.5, Table SM2.5; Figure 2.11). These key risks form part of the overall assessment of key risks in Chapter 16. The AR5 chapter on terrestrial ecosystems ( [[#Settele--2014|Settele et al., 2014]] ) had also identified three of these key risks—species extinctions, tree mortality and ecosystem carbon losses—and a fourth—invasion by non-native species. This chapter assesses, in multiple sections, the impacts of climate change on invasive species with respect to different processes or systems (e.g., in [[#2.4.2.3.3|Section 2.4.2.3.3]] ), and includes this aspect here in a new broader key risk of ecosystem structure change. The AR5 included wildfire as a mechanism of the key risk of ecosystem carbon loss. Based on additional research and field experience with major wildfires since then, this chapter sets wildfire apart as a specific key risk to ecosystem integrity and human well-being. These different measures of risk are interconnected, but approach the assessment of the risks to terrestrial and freshwater ecosytems from different angles, using complementary metrics.

Species are the fundamental unit of ecosystems. As species become rare, their roles in the functioning of the ecosystem diminishes and disappears altogether if they become locally extinct ( ''high confidence'' ) ( [[#Isbell--2015|Isbell et al., 2015]] ; [[#Chen--2018b|Chen et al., 2018b]] ; [[#van%20der%20Plas--2019|van der Plas, 2019]] ; [[#Wang--2021b|Wang et al., 2021b]] ). Loss of species and functional groups reduces the ability of an ecosystem to provide services, and lowers its resilience to climate change ( ''high confidence'' ) ( [[#2.6.7|Section 2.6.7]] ) ( [[#Elmqvist--2003|Elmqvist et al., 2003]] ; [[#Cadotte--2011|Cadotte et al., 2011]] ; [[#Harrison--2014|Harrison et al., 2014]] ; [[#Carlucci--2020|Carlucci et al., 2020]] ). For example, among crop systems, a key factor to succesful pollination is the phylogenetic diversity of bee species available, more than total abundances ( [[#Drossart--2020|Drossart and Gérard, 2020]] ). Because many species have obligate interactions with, or are resources for, other species (e.g., predators and their prey, insects and their host plants, plants and their mycorrhizae symbionts), the loss of one species affects the risk to another species, and, ultimately, ecosystem functioning ( [[#Mahoney--2017|Mahoney and Bishop, 2017]] )

Global rates of species extinction are accelerating dramatically ( [[#Barnosky--2011|Barnosky et al., 2011]] ), with approximately 10% of species having been driven extinct by humans since the late Pleistocene, principally by overexploitation and habitat destruction, a rate estimated to be 1000 times higher than pre-Anthropocene (natural) background extinction rates ( [[#De%20Vos--2015|De Vos et al., 2015]] ). Therefore, this level—10%—of species becoming “endangered” (sensu IUCN),and therefore at ''high'' risk of extinction, due to the loss of suitable climate space (Figure 2.8b), is used here as a threshold, moving the risk to biodiversity from ''moderate'' to ''high'' , and twice that (20%) as the threshold from ''high'' to ''very high'' .

Key risks assessed here are interconnected. Extinction of species is an irreversible impact of climate change and has negative consequences on ecosystem integrity and functioning, and the risks increase steeply with even small rises in global temperature ( [[#2.5.1.3|Section 2.5.1.3]] , Figure 2.6, Figure 2.7, Figure 2.8). Continued climate change substantially increases the risk of carbon losses due to wildfires, tree mortality from drought and insect pest outbreaks, peatland drying, permafrost thaw and changes in the structure of ecosystems; these could exacerbate self-reinforcing feedbacks between emissions from high-carbon ecosystems and increasing global temperatures ( ''medium confidence'' ). Thawing of Arctic permafrost alone could release 11–200 GtC ( ''medium confidence'' ). Complex interactions of climate changes, LULCC, carbon dioxide fluxes and vegetation changes will regulate the future carbon balance of the biosphere, processes incompletely represented in ESMs. The exact timing and magnitude of climate–biosphere feedbacks and the potential tipping points of carbon loss are characterised by broad ranges of the estimates, but studies indicate that increased ecosystem carbon losses could cause extreme future temperature increases ( ''medium confidence'' ). (Sections 2.5.2.7, 2.5.2.8, 2.5.2.9, 2.5.3.2, 2.5.3.3, 2.5.3.4, 2.5.3.5, Figure 2.10, Figure 2.11, Table 2.4, Table 2.5, Table SM2.2, Table SM2.5)

'''Table 2.5 |''' Key risks to terrestrial and freshwater ecosystems from climate change. This IPCC chapter assesses these as the most fundamental risks of climate change to ecosystem integrity and the ecosystem services that support human well-being. Climate factors include the primary variables governing the risk. Non-climate factors include other phenomena that can dominate or contribute to the risk. Detection and attribution comprise cases of observed changes attributed predominantly, or in part, to climate change, with some cases being attributed to anthropogenic climate change (Sections 2.4.2, 2.4.3, 2.4.4, 2.4.5, Table 2.2, Table 2.3, Table SM2.1). Adaptation includes options to address the risk (Section 2.6). Risk transitions (defined in Figure 2.11) indicate an approximate GSAT increase, relative to the pre-industrial period (1850–1900), to move from one level of risk to the other as well as assessed confidence. Table SM2.5 provides details of the temperature levels for risk transitions. Both tables provides details for the key risk burning embers diagram (Figure 2.11).

{| class="wikitable"
|-
| colspan="5"| '''Global biodiversity loss:''' Increasing numbers of plant and animal species at ''high'' extinction risk (species becoming endangered with projected loss of &gt;50% of range). The transition from non-detectable risk to moderate risk was based on the observed documentation of hundreds of local population extinctions, major declines in many sub-species and two to 92 global species extinctions that are attributable to climate change (with ''medium confidence'' or higher). The transition from ''moderate risk'' to ''high risk'' of biodiversity loss is centred around 1.5°C, based on a few taxa that are known from their basic biology and habitat requirements to be at ''high'' risk of extinction (endangered) at 1.5°C, and on the increasing number of taxa that are projected to have a ''high'' extinction risk affecting &gt;10% of the species in that taxa (1000 times the natural background rates of extinction). The transition to ''very high risk'' of biodiversity loss comes from the increasing number of taxa projected to have &gt;20% of species at a ''high'' risk of extinction. In the worst-case scenario (10th percentile of the models), some of the taxa show &gt;50% of the species at a ''high'' risk of extinction. These assessments are also weighted by role the species in the taxa play in performing ecosystem services (both to the ecosystems and to humans, e.g., pollinators, detritivores). There is ''high confidence'' for the moderate risk threshold because it is based on observed trends attributed to climate change. There is ''medium confidence'' for future projections since, for the purpose of developing this burning embers diagram, these risk thresholds are based on one large study (covering &gt;119,000 species) for which there were multiple warming scenarios considered, and primarily on the loss of suitable climate. Based on Sections 2.4.2, 2.5.1, 2.6.1, 2.6.6, Table 2.3, Figure 2.6, Table SM2.1 and Table SM2.2.

|-
| '''Climate factors'''

| '''Non-climate factors'''

| '''Detection and attribution'''

| '''Adaptation'''

| '''Risk transitions'''

'''(''' '''''confidence''''' ''')'''

|-
| Shifts in geographic placements of climate space; loss of climate space globally; emergence of non-analogue climates, increases in extreme climate events

| LUC, habitat degradation (e.g. from pollution, fertilisation, and invasive species)

| Already observed: many cases of population extinctions; 2 to 92 cases of species extinctions (2.4.2.2, 2.4.2.7.1);

species have tracked their climate niches raising confidence in SDM projections (2.4.2.1, 2.4.2.3, 2.4.2.5)

| Habitat restoration, habitat creation, increased connectivity of habitats and protected areas, increase in protected areas, assisted colonisation

| 0.8°C undetectable risk to moderate risk ( ''high confidence'' )

1.58°C moderate risk to high risk ( ''medium confidence'' )

2.07°C high risk to very high risk ( ''medium confidence'' )

|-
| colspan="5"|
|-
| colspan="5"| '''Ecosystem structure change:''' increasing risk of large-scale changes in ecosystem structure. Ecosystem structural change with most information derived for tropical forests, boreal forests, savannas and tundra for both observations and future projections. The transition from ''non-detectable risk'' to ''moderate risk'' is based on detected changes attributable to climate change or to interactions between changing disturbance regime, climate and rising CO 2 . These changes have already been observed at 0.5°C above pre-industrial levels, with shifts initially detected in boreal forests, tundra and tropical grassy ecosystems. The transition from ''moderate risk'' to ''high risk'' is centred around 1.5°C, based on widespread global observations (at a current GSAT of 1.09°C above pre-industrial levels) that agree with projected future impacts with at least 10% area of key ecosystems being affected (Box 2.1). Overall, there is ''medium confidence'' in projections. This is based on existing observations and some projections that have a ''high confidence'' of risk for several ecosystems, but data and projections are not available for all biomes, thus lowering overall confidence to ''medium confidence'' . The transition from ''high risk'' to ''very high risk'' occurs when &gt;50% of multiple ecosystems are projected to experience shifts in structure. (Sections 2.4.2.3, 2.4.3, 2.4.5, 2.5.2, Box 2.1, Figure Box 2.1.1, Table Box 2.1.1, Table SM2.2, Table SM2.3, Table SM2.4, Table SM2.5)

|-
| '''Climate factors'''

| '''Non-climate factors'''

| '''Detection and attribution'''

| '''Adaptation'''

| '''Risk transitions'''

'''(''' '''''confidence''''' ''')'''

|-
| Increases in average and extreme temperatures, changes in precipitation volume and timing, increased atmospheric CO 2

| LUC, livestock grazing, deforestation, fire suppression, loss of native herbivores, food, fiber and wood production

| Individual species range shifts, biome shifts

| Conservation of potential refugia, habitat restoration, increasing connectivity of habitats and protected areas, increase in protected areas, changes in grazing and fire management

| 0.5°C undetectable rsik to moderate risk ( ''high confidence'' )

1.5°C moderate risk to high risk ( ''medium confidence'' )

2.5°C high risk to very high risk ( ''medium confidence'' )

|-
| colspan="5"|
|-
| colspan="5"| '''Tree mortality:''' tree mortality that exceeds natural levels degrades habitat for plant and animal species, increases carbon emissions and reduces water supplies for people. Anthropogenic climate change caused three cases of drought-induced tree mortality in the period 1945–2007 in western North America, the African Sahel and north Africa in temperate and tropical ecosystems. Increased pest infestations and wildfires due to climate change also caused much of the recent tree mortality in North America. These changes occurred at GMST increases of 0.3°C–0.9°C above those in the pre-industrial period. Models project increasingly extensive drought-induced tree mortality at continued temperature increases of 1°C–2°C. Models project risks of mortality of up to half the forest area in different biomes at temperature increases of 2.5°C–4.5°C. In Amazon rainforests, insufficient plant moisture reserves during drought increase the risk of tree mortality, and, combined with increased fire from climate change and deforestation, the risk of a tipping point of massive forest dieback and a biome shift to grassland. (Sections 2.4.4.3, 2.5.2.6, 2.5.3.3, 2.5.3.5)

|-
| '''Climate factors'''

| '''Non-climate factors'''

| '''Detection and attribution'''

| '''Adaptation'''

| '''Risk transitions'''

'''(''' '''''confidence''''' ''')'''

|-
| Increase in temperature, decrease in precipitation, increase in aridity, increase in the frequency and severity of drought

| Deforestation, LUC

| Tree mortality up to 20% in three regions in Africa and North America

| Reduce deforestation, reduce habitat fragmentation, encourage natural regeneration, restore fragmented habitats

| 0.6°C undetectable risk to moderate risk ( ''high confidence'' )

1.5°C moderate risk to high risk ( ''medium confidence'' )

3.5°C high risk to very high risk ( ''medium confidence'' )

|-
| colspan="5"|
|-
| colspan="5"| '''Wildfire:''' increasing risk of wildfire that exceeds natural levels, damaging ecosystems, increasing human diseases and deaths and increasing carbon emissions. Field evidence shows that anthropogenic climate change has increased the area burned by wildfire above natural levels across western North America in the period 1984–2017, increasing burned area up to 11 times in one extreme year and doubling burned area over natural levels in a 32-year period. Burned area has increased in the Amazon, the Arctic, Australia and parts of Africa and Asia, consistent with but not formally attributed to anthropogenic climate change. These changes have occurred at GMST increases of 0.6°C–0.9°C. Empirical and dynamic global vegetation models project increases in burned area and fire frequency above natural levels on all continents under continued climate change, the emergence of an anthropogenic signal from natural variation in fire weather for a third of the global area and increases of burned area in regions where fire was previously rare or absent, particularly the Arctic tundra and Amazon rainforest, at global temperature increases of 1.5°C–2.5°C. Models project up to a doubling of burned area globally and wildfire-induced conversion of up to half the area of the Amazon rainforest to grassland at temperature increases of 3°C–4.5°C. (Sections 2.4.4.2, 2.5.3.2)

|-
| '''Climate factors'''

| '''Non-climate factors'''

| '''Detection and attribution'''

| '''Adaptation'''

| '''Risk transitions'''

'''(''' '''''confidence''''' ''')'''

|-
| Increase in the magnitude and duration of high temperatures, decrease in precipitation, decrease in relative humidity

| Deforestation, agricultural burning, peatland burning

| Increased burned area in western North America above natural levels

| Reduce deforestation, reduce the use of fire in tropical forests, use prescribed burning and allow naturally ignited fires to burn in targeted areas to reduce fuel loads, encourage settlement in non-fire-prone areas

| 0.75°C undetectable risk to moderate risk ( ''high confidence'' )

2.0°C moderate risk to high risk ( ''medium confidence'' )

4.0°C high risk to very high risk ( ''medium confidence'' )

|-
| colspan="5"|
|-
| colspan="5"| '''Ecosystem carbon loss''' : increasing risk of ecosystem carbon losses that could substantially raise the atmospheric carbon dioxide level. Measurements have detected emissions of carbon from boreal, temperate and tropical ecosystems in places where increases in wildfire and tree mortality have been attributed to anthropogenic climate change, at GMST increases of 0.6°C–0.9°C above the pre-industrial period. Many factors govern the carbon balance of ecosystems, so changes have not been attributed to climate change. Tropical forests and Arctic permafrost contain the highest ecosystem stocks of above- and below-ground carbon, respectively. Due to deforestation and forest degradation, primary tropical forests currently emit more carbon to the atmosphere than they remove. Wildfires in the Arctic are contributing to permafrost thaw and soil carbon release. An emissions scenario of 2°C increase could thaw ~15% of permafrost area and emit 20–100 GtC by 2100. Under emissions scenarios of 4°C global temperature increase, models project possible tipping points of conversion of half the Amazon rainforest to grassland and thawing of Arctic permafrost that could release 11–200 GtC which could substantially exacerbate climate change (Sections 2.4.3, 2.4.4.3, 2.4.4.4, 2.5.2.7–10, 2.5.3.2–5, Figure 2.9, Figure 2.10, Figure 2.11, Table 2.4, Table 2.5, Table SM2.2, Table SM2.3, Table SM2.5).

|-
| '''Climate factors'''

| '''Non-climate factors'''

| '''Detection and attribution'''

| '''Adaptation'''

| '''Risk transitions'''

'''(''' '''''confidence''''' ''')'''

|-
| Increase in temperature, increase in aridity, increase in the frequency and severity of drought

| Deforestation, road and infrastructure expansion, agricultural expansion

| Losses of carbon detected in boreal, temperate and tropical ecosystems due to wildfire and tree mortality, not formally attributed to climate change

| Reduce deforestation, especially in tropical forests, reduce road and infrastructure expansion, especially in the Arctic, reduce the use of fire to clear agricultural land, increase protected areas

| 0.75°C undetectable risk to moderate risk ( ''medium confidence'' )

2°C: moderate risk to high risk ( ''medium confidence'' )

4°C high risk to very high risk ( ''low confidence'' )

|}

<div id="_idContainer053" class="Figure"></div>

[[File:9c3a25c8d5a4134bf5d1c11a873cfe74 IPCC_AR6_WGII_Figure_2_011.png]]

'''Figure 2.11 | Key risks to terrestrial and freshwater ecosystems from climate change.''' This IPCC chapter assesses these as fundamental risks of climate change to ecosystem integrity and the ecosystem services that support human well-being, based on observed impacts and future risks of: (far-left) “biodiversity loss” refers to losses of animal and plant species from different ecosystems globally, with resulting declines in ecosystem integrity, functioning and resilience ( [[#2.4.2.1|Section 2.4.2.1]] , 2.4.2.2, 2.5.1.3.3); (middle-left) “structure change” refers to major changes occurring in ecosystem structure (Sections 2.4.3, Box 2.1, 2.5.2, Figure 2.9, Figure Box 2.1.1, Table Box 2.1.1, Table SM2.5); (middle) “tree mortality” refers to tree mortality exceeding natural levels (2.4.4.3, 2.5.3.3); (middle- right) “wildfire increase” refers to wildfire exceeding natural levels ( [[#2.4.4.2|Section 2.4.4.2]] , 2.5.3.2); (far-right) “carbon loss” refers to ecosystem carbon losses that could occur abruptly and substantially raise atmospheric carbon dioxide (Sections 2.4.3.6–2.4.3.9, 2.4.4.4, 2.5.2.6–2.5.2.10, 2.5.3.4, 2.5.3.5). This burning embers diagram shows impacts and risks in relation to changes in GSAT, relative to the pre-industrial period (1850–1900). Risk levels reflect current levels of adaptation and do not include more interventions that could lower risk. The compound effects of climate change, combined with deforestation, agricultural expansion and urbanisation as well as air, water and soil pollution and other non-climate hazards could increase risks. Tables 2.5 and SM2.5 provide details of the key risks and temperature levels for the risk transitions.

<div id="FAQ 2.4" class="h2-container"></div>

<span id="faq-2.4-how-does-nature-benefit-human-health-and-well-being-and-how-does-climate-change-affect-this"></span>
=== FAQ 2.4 | How does nature benefit human health and well-being and how does climate change affect this? ===

<div id="h2-31-siblings" class="h2-siblings"></div>

''Human health and well-being are highly dependent on the ‘health’ of nature. Nature provides material and economic services that are essential for human health and productive livelihoods. Studies also show that being in ‘direct contact with natural environments’ has direct positive effects on well-being, health and socio-cognitive abilities. Therefore, the loss of species and biodiversity due to climate change will reduce natural spaces and, in turn, decrease human well-being and health worldwide.''

Human health and well-being are highly dependent on the ‘health’ of nature. Biodiversity—the variety of genes, species, communities and ecosystems—provides services that are essential for human health and productive livelihoods, such as breathable air, drinkable water, productive oceans and fertile soils for growing food and fuels. Natural ecosystems also help store carbon and regulate climate, floods, disease, pollution and water quality. The loss of species, leading to reduced biodiversity, has direct and measurable negative effects on all of these essential services, and therefore on humankind. A recent demonstration of this is the decline of pollinator species, with potential negative effects on crop pollination, a fundamental ecosystem function crucial for agriculture. The loss of wild relatives of the domesticated varieties that humans rely on for agriculture reduces the genetic variability that may be needed to support the adaptation of crops to future environmental and social challenges.

[[File:e83586be3c9477cb86557cee5b6be56c IPCC_AR6_WGII_Figure_2_FAQ_2.4.1.png]]

'''Figure FAQ2.4.1 |''' '''The positive relationship between human health and well-being and nature conservation.''' Nature provides essential services to humans including material and economic services (i.e., ecosystem services) as well as cultural, experiential and recreational services, which, in turn, enhance human psychological and physical health and well-being. People who are more connected to nature are not only happier and healthier but are also more likely to engage in pro-nature behaviours, making the enhancement of human–nature connectedness worldwide a valuable win–win solution for humans and nature to face environmental challenges.

The number of species that can be lost before negative impacts occur is not known and is likely to differ in different systems. However, in general, more diverse systems are more resilient to disturbances and able to recover from extreme events more quickly. Biodiversity loss means there are fewer connections within an ecosystem. A simpler food web with fewer interactions means less redundancy in the system, reducing the stability and ability of plants and animal communities to recover from disturbances and extreme weather events such as floods and drought.

In addition to ‘material’ and economic services such as eco-tourism, nature also provides cultural services such as recreation, spirituality and well-being. Specifically, being in ‘direct contact with natural environments’ (vs. an urban environment) has a high positive impact on human well-being (e.g., mood, happiness), psychological and physical health (energy, vitality, heart rate, depression) and socio-cognitive abilities (attention, memory, hyperactivity, altruism, cooperation). Therefore, the loss of species from climate change and urbanisation will reduce natural spaces, decrease biodiversity, and, in turn, decrease human well-being and health worldwide.

Finally, the extent to which humans consider themselves part of the natural world—known as human-nature connectedness—has been demonstrated to be closely associated with human health and well-being. Individuals who are more connected to nature are not only happier and healthier but also tend to engage more in pro-nature behaviours, making the enhancement of human–nature connectedness worldwide a valuable win–win solution for humans and nature to face environmental challenges.

<div id="2.6 " class="h1-container"></div>

<span id="climate-change-adaptation-for-terrestrial-and-freshwater-ecosystems"></span>