Global mean SST has increased since the beginning of the 20th century by 0.88°C (very likely range: 0.68–1.01°C), and it is virtually certain that the global ocean has warmed since at least 1971 (WGI AR6 Section 9.2; Fox-Kemper et al., 2021). A key characteristic of ocean temperature change relevant for ecosystems is climate velocity, a measure of the speed and direction at which isotherms move under climate change (Burrows et al., 2011), which gives the rate at which species must migrate to maintain constant climate conditions. It has been shown to be a useful and simple predictor of species distribution shifts in marine ecosystems (Chen et al., 2011; Pinsky et al., 2013; Lenoir et al., 2020). Median climate velocity in the surface ocean has been 21.7 km per decade since 1960, with higher values in the Arctic/sub-Arctic and within 15° of the Equator (Figure 3.3; Burrows et al., 2011). While climate velocity has been slower in the mesopelagic layer (200–1000 m) than in the epipelagic layer (0–200 m) over the past 50 years, it has been shown to be faster in the bathypelagic (1000–4000 m) and abyssopelagic (>4000 m) layers (Figure 3.4; Brito-Morales et al., 2020), suggesting that deep-ocean species could be as exposed to effects of warming as species in the surface ocean (Brito-Morales et al., 2020).

Marine heatwaves (MHWs) are periods of extreme seawater temperature relative to the long-term mean seasonal cycle, that persist for days to months, and that may carry severe consequences for marine ecosystems and their services (WGI AR6 Box 9.2; Hobday et al., 2016a; Smale et al., 2019; Fox-Kemper et al., 2021). MHWs became more frequent over the 20th century (high confidence) and into the beginning of the 21st century, approximately doubling in frequency (high confidence) and becoming more intense and longer since the 1980s (medium confidence) (WGI AR6 Box 9.2; Fox-Kemper et al., 2021). These trends in MHWs are explained by an increase in ocean mean temperatures (Oliver et al., 2018), and human influence has very likely contributed to 84–90% of them since at least 2006 (WGI AR6 Box 9.2; Fox-Kemper et al., 2021). The probability of occurrence (as well as duration and intensity) of the largest and most impactful MHWs that have occurred in the past 30 years has increased more than 20-fold due to anthropogenic climate change (Laufkötter et al., 2020).

Ocean warming will continue over the 21st century (virtually certain), with the rate of global ocean warming starting to be scenario-dependent from about the mid-21st century (medium confidence). At the ocean surface, it is virtually certain that SST will continue to increase throughout the 21st century, with increasing hazards to many marine ecosystems (WGI AR6 Box 9.2; Fox-Kemper et al., 2021). The future global mean SST increase projected by CMIP6 models for the period 1995–2014 to 2081–2100 is 0.86°C (very likely range: 0.43–1.47°C) under SSP1-2.6, 1.51°C (1.02–2.19°C) under SSP2-4.5, 2.19°C (1.56–3.30°C) under SSP3-7.0 and 2.89°C (2.01–4.07°C) under SSP5-8.5 (WGI AR6 Section 9.2.1; Fox-Kemper et al., 2021). Stronger surface warming occurs in parts of the tropics, in the North Pacific, and in the Arctic Ocean, where SST increases by >4°C in 2080–2099 under SSP5-8.5 (Kwiatkowski et al., 2020). The CMIP6 climate models also project ocean warming at the seafloor, with the magnitude of projected changes being less than that of surface waters but having larger uncertainties (Kwiatkowski et al., 2020). The projected end-of-the-century warming in CMIP6 as reported here is greater than assessed with Coupled Model Intercomparison Project 5 (CMIP5) models in AR5 and in SROCC for similar radiative forcing scenarios (Figure 3.5; Kwiatkowski et al., 2020), because of greater climate sensitivity in the CMIP6 model ensemble than in CMIP5 (WGI AR6 Chapter 4; Forster et al., 2020; Lee et al., 2021).

Marine heatwaves will continue to increase in frequency, with a likely global increase of 2–9 times in 2081–2100 compared with 1995–2014 under SSP1-2.6, and 3–15 times under SSP5-8.5, with the largest increases in tropical and Arctic oceans (WGI AR6 Box 9.2; Frölicher et al., 2018; Fox-Kemper et al., 2021).

Global mean sea level (GMSL) (Cross-Chapter Box SLR in Chapter 3) has risen by about 0.20 m since 1901 and continues to accelerate (WGI AR6 Section 2.3.3.3; Church and White, 2011; Jevrejeva et al., 2014; Hay et al., 2015; Kopp et al., 2016; Dangendorf et al., 2017; WCRP Global Sea Level Budget Group, 2018; Kemp et al., 2018; Ablain et al., 2019; Gulev et al., 2021).

Most coastal ecosystems (mangroves, seagrasses, salt marshes, shallow coral reefs, rocky shores and sandy beaches) are affected by changes in relative sea level (RSL, the change in the mean sea level relative to the land; Section 3.4.2). Regional rates of RSL rise differ from the global mean due to a range of factors, including local subsidence driven by anthropogenic activities such as groundwater and hydrocarbon extraction (WGI AR6 Box 9.1; Fox-Kemper et al., 2021). In many deltaic regions, anthropogenic subsidence is currently the dominant driver of RSL rise (WGI AR6 Section 9.6.3.2; Tessler et al., 2018; Fox-Kemper et al., 2021). RSL rise is driving a global increase in the frequency of extreme sea levels (high confidence) (WGI AR6 Section 9.6.4.1; Fox-Kemper et al., 2021).

GMSL rise through the middle of the 21st century exhibits limited dependence on emissions scenario; between 1995–2014 and 2050, GMSL is likely to rise by 0.15–0.23 m under SSP1-1.9 and 0.20–0.30 m under SSP5-8.5 (WGI AR6 Section 9.6.3; Fox-Kemper et al., 2021). Beyond 2050, GMSL and RSL projections are increasingly sensitive to the differences among emission scenarios. Considering only processes in which there is at least medium confidence (e.g., thermal expansion, land-water storage, land-ice surface mass balance and some ice-sheet dynamic processes), GMSL between 1995–2014 and 2100 is likely to rise by 0.28–0.55 m under SSP1-1.9, 0.33–0.61 m under SSP1-2.6, 0.44–0.76 m under SSP2-4.5, 0.55–0.90 m under SSP3-7.0 and 0.63–1.02 m under SSP5-8.5 (Figure 3.5). Under high-emission scenarios, ice-sheet processes in which there is low confidence and deep uncertainty might contribute more than one additional metre to GMSL rise by 2100 (WGI AR6 Chapter 9; Fox-Kemper et al., 2021).

Rising mean RSL will continue to drive an increase in the frequency of extreme sea levels (high confidence). The expected frequency of the current 1-in-100-year extreme sea level is projected to increase by a median of 20–30 times across tide-gauge sites by 2050, regardless of emission scenario (medium confidence). In addition, extreme-sea-level frequency may be affected by changes in tropical cyclone climatology (low confidence), wave climatology (low confidence) and tides (high confidence) associated with climate change and sea level change (WGI AR6 Section 9.6.4.2; Fox-Kemper et al., 2021).

Ocean circulation and its variations are key to the evolution of the physical, chemical and biological properties of the ocean. Vertical mixing and upwelling are critical factors affecting the supply of nutrients to the sunlit ocean and hence the magnitude of primary productivity. Ocean currents not only transport heat, salt, carbon and nutrients, but they also control the dispersion of many organisms and the connectivity between distant populations.

Ocean stratification is an important factor controlling biogeochemical cycles and affecting marine ecosystems. WGI AR6 Section 9.2.1.3 (Fox-Kemper et al., 2021) assessed that it is virtually certain that stratification in the upper 200 m of the ocean has been increasing since 1970. Recent evidence has strengthened estimates of the rate of change (Yamaguchi and Suga, 2019; Li et al., 2020a; Sallée et al., 2021), with an estimated increase of 1.0 ± 0.3% (very likely range) per decade over the period 1970–2018 (high confidence) (WGI AR6 Section 9.2.1.3; Fox-Kemper et al., 2021), higher than assessed in SROCC. It is very likely that stratification in the upper few hundred metres of the ocean will increase substantially in the 21st century in all ocean basins, driven by intensified surface warming and near-surface freshening at high latitudes (WGI AR6 Section 9.2.1.3; Capotondi et al., 2012; Fu et al., 2016; Bindoff et al., 2019a; Kwiatkowski et al., 2020; Fox-Kemper et al., 2021).

Contrasting changes among the major eastern boundary coastal upwelling systems (EBUS) were identified in AR5 (Hoegh-Guldberg et al., 2014). While SROCC assessed with high confidence that three (Benguela, Peru-Humboldt, California) out of the four major EBUS have experienced upwelling-favourable wind intensification in the past 60 years (Sydeman et al., 2014; Bindoff et al., 2019a), WGI AR6 revisited this assessment based on evidence showing low agreement between studies that have investigated trends over past decades (Varela et al., 2015). WGI AR6 assessed that only the California Current system has undergone large-scale upwelling-favourable wind intensification since the 1980s (medium confidence) (WGI AR6 Section 9.2.1.5; García-Reyes and Largier, 2010; Seo et al., 2012; Fox-Kemper et al., 2021).

While no consistent pattern of contemporary changes in upwelling-favourable winds emerges from observation-based studies, numerical and theoretical work projects that summertime winds near poleward boundaries of upwelling zones will intensify, while winds near equatorward boundaries will weaken (high confidence) (WGI AR6 Section 9.2.3.5; García-Reyes et al., 2015; Rykaczewski et al., 2015; Wang et al., 2015; Aguirre et al., 2019; Fox-Kemper et al., 2021). Nevertheless, projected future annual cumulative upwelling wind changes at most locations and seasons remain within ±10–20% of present-day values (medium confidence) (WGI AR6 Section 9.2.3.5; Fox-Kemper et al., 2021).

Continuous observation of the Atlantic meridional overturning circulation (AMOC) has improved the understanding of its variability (Frajka-Williams et al., 2019), but there is low confidence in the quantification of AMOC changes in the 20th century because of low agreement in quantitative reconstructed and simulated trends (WGI AR6 Sections 2.3.3, 9.2.3.1; Fox-Kemper et al., 2021; Gulev et al., 2021). Direct observational records since the mid-2000s remain too short to determine the relative contributions of internal variability, natural forcing and anthropogenic forcing to AMOC change (high confidence) (WGI AR6 Sections 2.3.3, 9.2.3.1; Fox-Kemper et al., 2021; Gulev et al., 2021). Over the 21st century, AMOC will very likely decline for all SSP scenarios but will not involve an abrupt collapse before 2100 (WGI AR6 Sections 4.3.2, 9.2.3.1; Fox-Kemper et al., 2021; Lee et al., 2021).

Sea ice is a key driver of polar marine life, hosting unique ecosystems and affecting diverse marine organisms and food webs through its impact on light penetration and supplies of nutrients and organic matter (Arrigo, 2014). Since the late 1970s, Arctic sea ice area has decreased for all months, with an estimated decrease of 2 million km 2 (or 25%) for summer sea ice (averaged for August, September and October) in 2010–2019 as compared with 1979–1988 (WGI AR6 Section 9.3.1.1; Fox-Kemper et al., 2021). For Antarctic sea ice there is no significant global trend in satellite-observed sea ice area from 1979 to 2020 in either winter or summer, due to regionally opposing trends and large internal variability (WGI AR6 Section 9.3.2.1; Maksym, 2019; Fox-Kemper et al., 2021).

CMIP6 simulations project that the Arctic Ocean will likely become practically sea ice free (area below 1 million km 2) for the first time before 2050 and in the seasonal sea ice minimum in each of the four emission scenarios SSP1-1.9, SSP1-2.6, SSP2-4.5 and SSP5-8.5 (Figure 3.7; WGI AR6 Section 9.3.2.2; Notze and SIMIP Community, 2020; Fox-Kemper et al., 2021). Antarctic sea ice area is also projected to decrease during the 21st century, but due to mismatches between model simulations and observations, combined with a lack of understanding of reasons for substantial inter-model spread, there is low confidence in model projections of future Antarctic sea ice changes, particularly at the regional level (WGI AR6 Section 9.3.2.2; Roach et al., 2020; Fox-Kemper et al., 2021).

The ocean’s uptake of anthropogenic carbon affects its chemistry in a process referred to as ocean acidification, which increases the concentrations of aqueous CO2, bicarbonate and hydrogen ions, and decreases pH, carbonate ion concentrations and calcium carbonate mineral saturation states (Doney et al., 2009). Ocean acidification affects a variety of biological processes with, for example, lower calcium carbonate saturation states reducing net calcification rates for some shell-forming organisms and higher CO2 concentrations increasing photosynthesis for some phytoplankton and macroalgal species (Section 3.3.2).

Direct measurements of ocean acidity from ocean time series, as well as pH changes determined from other shipboard studies, show consistent decreases in ocean surface pH over the past few decades (virtually certain) (WGI AR6 Section 5.3.2.2; Takahashi et al., 2014; Bindoff et al., 2019a; Sutton et al., 2019; Canadell et al., 2021).

Since the 1980s, surface ocean pH has declined by a very likely rate of 0.016–0.020 per decade in the subtropics and 0.002–0.026 per decade in the subpolar and polar zones (WGI AR6 Section 5.3.2.2; Canadell et al., 2021). Typically, the pH of global surface waters has decreased from 8.2 to 8.1 since the pre-industrial era (1750 CE), a trend attributable to rising atmospheric CO2 (virtually certain) (Orr et al., 2005; Jiang et al., 2019).

Ocean acidification is also developing in the ocean interior (very high confidence) due to the transport of anthropogenic CO2 to depth by ocean currents and mixing (WGI AR6 Section 5.3.3.1; Canadell et al., 2021). There, it leads to the shoaling of saturation horizons of aragonite and calcite (high confidence) (WGI AR6 Section 5.3.3.1; Canadell et al., 2021), below which dissolution of these calcium carbonate minerals is thermodynamically favoured. The calcite or aragonite saturation horizons have migrated upwards in the North Pacific (1–2 m yr –1 over 1991–2006) (Feely et al., 2012) and in the Irminger Sea (10–15 m yr –1 for the aragonite saturation horizon over 1991–2016) (Perez et al., 2018). In some locations of the western Atlantic Ocean, calcite saturation depth has risen by ~300 m since the pre-industrial era due to increasing concentrations of deep-ocean dissolved inorganic carbon (Sulpis et al., 2018). In the Arctic, where some coastal surface waters are already undersaturated with respect to aragonite due to the degradation of terrestrial organic matter (Mathis et al., 2015; Semiletov et al., 2016), the deep aragonite saturation horizon shoaled on average 270 ± 60 m during 1765–2005 (Terhaar et al., 2020).

Detection and attribution of ocean acidification in coastal environments are more difficult than in the open ocean due to larger spatio-temporal variability of carbonate chemistry (Duarte et al., 2013; Laruelle et al., 2017; Torres et al., 2021) and to the influence of other natural acidification drivers such as freshwater and high-nutrient riverine inputs (Cai et al., 2011; Laurent et al., 2017; Fennel et al., 2019; Cai et al., 2020) or anthropogenic acidification drivers (Section 3.1) like atmospherically deposited nitrogen and sulphur (Doney et al., 2007; Hagens et al., 2014). Since AR5, the observing network in coastal oceans has expanded substantially, improving understanding of both the drivers and amplitude of observed variability (Sutton et al., 2016). Recent studies indicate that two more decades of observations may be required before anthropogenic ocean acidification emerges over natural variability in some coastal sites and regions (WGI AR6 Section 5.3.5.2; Sutton et al., 2019; Turk et al., 2019; Canadell et al., 2021).

Mean open-ocean surface pH is projected to decline by 0.08 ± 0.003 (very likely range), 0.17 ± 0.003, 0.27 ± 0.005 and 0.37 ± 0.007 pH units in 2081–2100 relative to 1995–2014, for SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, respectively (Figure 3.5; WGI AR6 Section 4.3.2; Kwiatkowski et al., 2020; Lee et al., 2021). Projected changes in surface pH are relatively uniform in contrast with those of other surface-ocean variables, but they are largest in the Arctic Ocean (Figure 3.6; WGI AR6 Section 5.3.4.1; Canadell et al., 2021). Similar declines in the concentration of carbonate ions are projected by Earth system models (ESMs; Bopp et al., 2013; Gattuso et al., 2015; Kwiatkowski et al., 2020). The North Pacific, the Southern Ocean and Arctic Ocean regions will become undersaturated for calcium carbonate minerals first (Orr et al., 2005; Pörtner et al., 2014). Concurrent impacts on the seasonal amplitude of carbonate chemistry variables are anticipated (i.e., increased amplitude for pCO2 and hydrogen ions, decreased amplitude for carbonate ions; McNeil and Sasse, 2016; Kwiatkowski and Orr, 2018; Kwiatkowski et al., 2020).

Future declines in subsurface pH (Figure 3.6) will be modulated by changes in ocean overturning and water-mass subduction (Resplandy et al., 2013), and in organic matter remineralisation (Chen et al., 2017). In particular, decreases in pH will be less consistent at the seafloor than at the surface and will be linked to the transport of surface anomalies to depth. For example, >20% of the North Atlantic seafloor deeper than 500 m, including canyons and seamounts designated as marine protected areas (MPAs), will experience pH reductions >0.2 by 2100 under RCP8.5 (Gehlen et al., 2014). Changes in pH in the abyssal ocean (>3000 m deep) are greatest in the Atlantic and Arctic Oceans, with lesser impacts in the Southern and Pacific Oceans by 2100, mainly due to ventilation time scales (Sweetman et al., 2017).

Ocean deoxygenation, the loss of oxygen in the ocean, results from ocean warming, through a reduction in oxygen saturation, increased oxygen consumption, increased ocean stratification and ventilation changes (Keeling et al., 2010; IPCC, 2019a). In recent decades, anthropogenic inputs of nutrients and organic matter (Section 3.1) have increased the extent, duration and intensity of coastal hypoxia events worldwide (Diaz and Rosenberg, 2008; Rabalais et al., 2010; Breitburg et al., 2018), while pollution-induced atmospheric deposition of soluble iron over the ocean has accelerated open-ocean deoxygenation (Ito et al., 2016). Deoxygenation and acidification often coincide because biological consumption of oxygen produces CO2. Deoxygenation can have a range of detrimental effects on marine organisms and reduce the extent of marine habitats (Sections 3.3.2, 3.4.3.1; Vaquer-Sunyer and Duarte, 2008; Chu and Tunnicliffe, 2015).

Changes in ocean oxygen concentrations have been analysed from compilations of in situ data dating back to the 1960s (Helm et al., 2011; Ito et al., 2017; Schmidtko et al., 2017). SROCC concluded that a loss of oxygen had occurred in the upper 1000 m of the ocean (medium confidence), with a global mean decrease of 0.5–3.3% (very likely range) over 1970–2010 (Bindoff et al., 2019a). Based on new regional assessments (Queste et al., 2018; Bronselaer et al., 2020; Cummins and Ross, 2020; Stramma et al., 2020), WGI AR6 assesses that ocean deoxygenation has occurred in most regions of the open ocean since the mid-20th century (high confidence), but it is modified by climate variability on interannual and inter-decadal time scales (medium confidence) (WGI AR6 Sections 2.3.3.6, 5.3.3.2; Canadell et al., 2021; Gulev et al., 2021). New findings since SROCC also confirm that the volume of oxygen minimum zones (OMZs) are expanding at many locations (high confidence) (WGI AR6 Section 5.3.3.2; Canadell et al., 2021).

The most recent estimates of future oxygen loss in the subsurface ocean (100–600 m), using CMIP6 models, amount to −4.1 ± 4.2 (very likely range), −6.6 ± 5.7, −10.1 ± 6.7 and −11.2 ± 7.7% in 2081–2100 relative to 1995–2014 for SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, respectively (Figure 3.5; Kwiatkowski et al., 2020). Based on these CMIP6 projections, WGI AR6 concludes that the oxygen content of the subsurface ocean is projected to decline to historically unprecedented conditions over the 21st century (medium confidence) (WGI AR6 Section 5.3.3.2; Canadell et al., 2021). These declines are greater (by 31–72%) than simulated by the CMIP5 models in their Representative Concentration Pathway (RCP) analogues, a likely consequence of enhanced surface warming and stratification in CMIP6 models (Figure 3.5; Kwiatkowski et al., 2020). At the regional scale and for subsurface waters, projected changes are not spatially uniform, and there is lower agreement among models than they show for the global mean trend (Bopp et al., 2013; Kwiatkowski et al., 2020). In particular, large uncertainties remain for these future projections of ocean deoxygenation in the subsurface tropical oceans, where the major OMZs are located (Cabré et al., 2015; Bopp et al., 2017).

The availability of nutrients in the surface ocean often limits primary productivity, with implications for marine food webs and the biological carbon pump. Nitrogen availability tends to limit phytoplankton productivity throughout most of the low-latitude ocean, whereas dissolved iron availability limits productivity in high-nutrient, low-chlorophyll regions, such as in the main upwelling region of the Southern Ocean and the Eastern Equatorial Pacific (high confidence) (Moore et al., 2013; IPCC, 2019b). Phosphorus, silicon, other micronutrients such as zinc, and vitamins can also co-limit marine phytoplankton productivity in some ocean regions (Moore et al., 2013). Whereas some studies have shown coupling between climate variability and nutrient trends in specific regions, such as in the North Atlantic (Hátún et al., 2016), North Pacific (Di Lorenzo et al., 2009; Yasunaka et al., 2014) and tropical (Stramma and Schmidtko, 2021) Oceans, very few studies have been able to detect long-term changes in ocean nutrient concentrations (but see Yasunaka et al., 2016).

Future changes in nutrient concentrations have been estimated using ESMs, with future increases in stratification generally leading to decreased nutrient levels in surface waters (IPCC, 2019b). CMIP6 models project a decline in the nitrate concentration of the upper 100 m in 2080–2099 relative to 1995–2014 of −0.46 ± 0.45 (very likely range), −0.60 ± 0.58, −0.80 ± 0.77 and −1.00 ± 0.78 mmol m –3 under SSP1-2.6, SSP2-4.5 and SSP5-8.5, respectively (Figure 3.5; Kwiatkowski et al., 2020). These declines in nitrate concentration are greater than simulated by the CMIP5 models in their RCP analogues, a likely consequence of enhanced surface warming and stratification in CMIP6 models (Figure 3.5; Kwiatkowski et al., 2020). It is concluded that the surface ocean will encounter reduced nitrate concentrations in the 21st century (medium confidence).

Earth system models project distinct regional evolutions of the different CIDs over the 21st century (very high confidence) (Figures 3.5, 3.6, 3.7; Kwiatkowski et al., 2020). Tropical and subtropical oceans are characterised by projected warming and acidification, accompanied by declining nitrate concentrations in equatorial upwelling regions. The North Atlantic is characterised by a high exposure to acidification and declining nitrate concentrations. The North Pacific is characterised by high sensitivity to compound changes, with high rates of warming, acidification, deoxygenation and nutrient depletion. In contrast, the development of compound hazards is limited in the Southern Ocean, where rates of warming and nutrient depletion are lower. The Arctic Ocean is characterised by the highest rates of acidification and warming, strong nutrient depletion, and it will likely become practically sea ice free in the September mean for the first time before the year 2050 in all SSP scenarios (high confidence) (Figures 3.5, 3.6, 3.7; Sections 3.2.2, 3.2.3).

In general, the projected changes in climate-induced drivers are less in absolute terms in the deep-sea (mesopelagic and bathypelagic domains and deep-sea habitats) than in the surface ocean and in shallow-water habitats (e.g., kelp ecosystems, warm-water corals) (very high confidence) (Figures 3.6, 3.7; Mora et al., 2013; Sweetman et al., 2017). The mesopelagic domain will be nevertheless exposed to high rates of deoxygenation (Figure 3.6) and high climate velocities (Figure 3.4; Section 3.2.2.1), as well as impacted by the shoaling of aragonite or calcite saturation horizon (Section 3.2.3.2). Significant differences in projected trends between the SSPs show that mitigation strategies will limit exposure of deep-sea ecosystems to potential warming, acidification and deoxygenation during the 21st century (very high confidence) (Figure 3.6; Kwiatkowski et al., 2020).

Anthropogenic changes in climate-induced drivers assessed here exhibit vastly distinct times of emergence, which is the time scale over which an anthropogenic signal related to climate change is statistically detected to emerge from the background noise of natural climate for a specific region (Christensen et al., 2007; Hawkins and Sutton, 2012). SROCC concluded that for ocean properties, the time of emergence ranges from under a decade (e.g., surface ocean pH) to over a century (e.g., net primary production; see Section 3.4.3.3.4 for time of emergence of biological properties; Bindoff et al., 2019a).

The literature assessed in SROCC mainly focused on surface ocean properties and gradual mean changes. Since then, the time of emergence has also been investigated for subsurface properties, ocean extreme events and particularly vulnerable regions, such as the Arctic Ocean (Hameau et al., 2019; Oliver et al., 2019; Burger et al., 2020; Landrum and Holland, 2020; Schlunegger et al., 2020), but subsequent assessments are low confidence due to limited evidence. Below the surface, changes in temperature typically emerge from internal variability prior to changes in oxygen; however, in about a third of the global thermocline, deoxygenation emerges prior to warming (Hameau et al., 2019). Permanent MHW states, defined as when SST exceeds the MHW threshold continuously over a full calendar year, will emerge during the 21st century in many parts of the surface ocean (Oliver et al., 2019). Ocean acidification extremes have already emerged from background natural internal variability during the 20th century in most of the surface ocean (Burger et al., 2020). In the Arctic, anthropogenic sea ice changes have already emerged from the background internal variability, and anthropogenic alteration of air temperatures will emerge in the early- to mid-21st century (Landrum and Holland, 2020).

Paleoclimatology observations are useful to assess multiple hazards of environmental change while excluding direct anthropogenic impacts (Section 3.4.3.3). Ancient intervals of rapid climate warming that occurred between 300 and 50 million years ago (Ma) were triggered by the release of greenhouse gases (high confidence). The sources of greenhouse gases varied but include volcanic degassing from continental flood basalts and methane hydrates stored in marine sediments and soils (Foster et al., 2018). Six extreme ancient hyperthermal events are known from the last 300 Ma, when tropical SSTs reached 1.5°C–10°C warmer than pre-industrial conditions, and with substantial impacts on ancient life (Cross-Chapter Box PALEO in Chapter 1). Warming and deoxygenation in the oceans were closely associated in hyperthermal events (high confidence), with anoxia reaching the photic zone and abyssal depths (Kaiho et al., 2014; Müller et al., 2017; Penn et al., 2018; Weissert, 2019), whereas ocean acidification has not been demonstrated consistently (medium confidence) (Hönisch et al., 2012; Penman et al., 2014; Clarkson et al., 2015; Harper et al., 2020a; Jurikova et al., 2020; Müller et al., 2020).

Greenhouse gases also contributed substantially to shaping the longer-term climate trends over the past 50 million years, although changes in continental configuration and ocean circulation as well as planetary orbital cycles were equally important (WGI AR6 Cross-Chapter Box 2.1 in Chapter 2; Westerhold et al., 2020; Gulev et al., 2021). There is little evidence for ocean acidification in the past 2.6 Ma (low confidence) (Hönisch et al., 2012), but ocean ventilation was highly sensitive to even modest warming such as observed in the past 10,000 years (medium confidence) (Jaccard and Galbraith, 2012; Lembke-Jene et al., 2018).

Warming and rising CO2 concentrations enhance growth and/or photosynthetic rates in many species of cyanobacteria, picoeukaryotes, coccolithophores, dinoflagellates and diatoms (high confidence) (Fu et al., 2007; Sett et al., 2014; Hoppe et al., 2018a; Wolf et al., 2018; Brandenburg et al., 2019), and the optimum pCO2 for growth and/or primary production shifts upward under warming (medium confidence) (Sett et al., 2014; Hoppe et al., 2018a). Warming and ocean acidification appear to jointly favour the proliferation and toxicity of harmful algal bloom (HAB) species (limited evidence, high agreement ) (Section 3.5.5.3; Bindoff et al., 2019a; Brandenburg et al., 2019; Griffith et al., 2019a; Wells et al., 2020), but a 2021 analysis found no uniform global trend in HABs or their distribution over 1985–2018 once field data were adjusted for regional variations in monitoring effort (Hallegraeff et al., 2021). The predominantly detrimental impacts of ocean acidification on coccolithophores can partly be offset by warming (Seifert et al., 2020) but also be exacerbated, depending on the magnitudes of drivers (D’Amario et al., 2020). For non-calcifying macroalgae, responses are highly species specific and often indicate synergistic interactions between warming and acidification (Kram et al., 2016; Falkenberg et al., 2018). Ocean acidification poses a large risk for coralline algae that is further amplified by warming (medium confidence) (Section 3.4.2.2; Cornwall et al., 2019). However, temperatures up to 5°C above ambient do not decrease calcification (Cornwall et al., 2019), and there is limited evidence that some species have the physiological capacity to resist acidification via pH upregulation at the calcification site (Cornwall et al., 2017a). For seagrass, warming beyond a species’ thermal tolerance will limit growth and impact germination, but ocean acidification appears to increase thermal tolerance of some eelgrass species by increasing the photosynthesis-to-respiration ratio (medium confidence) (Egea et al., 2018; Scalpone et al., 2020; Zimmerman, 2021).

Thermal sensitivity of pelagic primary producers changes with nutrient supply (high confidence) (Thomas et al., 2017; Marañón et al., 2018; Fernández et al., 2020). Phosphorus limitation lowers the temperature optimum for growth of phytoplankton, making these organisms more prone to heat stress (Thomas et al., 2017; Bestion et al., 2018). This trend may hold for open-ocean phytoplankton, which are often iron-limited (medium confidence) (Boyd, 2019). Such temperature-nutrient interactions might be especially relevant during summer MHWs (Section 3.2.2.1; Cross-Chapter Box EXTREMES in Chapter 2; IPCC, 2018; Holbrook et al., 2019; DeCarlo et al., 2020; Hayashida et al., 2020), when primary producers are often nutrient-limited and near their thermal limits. Increasingly frequent and intense MHWs along with projected decreases in nutrient availability (Section 3.2.3.3) may push some primary producers beyond tolerance thresholds. Temperature–nutrient interactions can also alter the photosynthesis-to-respiration ratio in phytoplankton (Marañón et al., 2018). Overall, rising metabolic rates due to warming will be restricted to primary producers in high-nutrient regions (medium confidence) (Thomas et al., 2017; Marañón et al., 2018). For zooxanthellae-containing corals, nutrient supply from upwelling or from runoff can increase coral susceptibility to bleaching during warm-season MHWs (DeCarlo et al., 2020; Wooldridge, 2020).

The effects of ocean acidification on growth, metabolic rates or elemental composition of primary producers changes with nutrient availability and light conditions (high confidence) (Gao et al., 2019; Seifert et al., 2020). While interactions with nutrients are often additive in phytoplankton, diatoms revealed predominantly synergistic interactions (Seifert et al., 2020). Growth or photosynthesis of some diatom and HAB species, for instance, are stimulated by ocean acidification only if nutrients are replete (Hoppe et al., 2013; Boyd et al., 2015b; Eberlein et al., 2016; Griffith et al., 2019a). Interactions with light are more complex because relative effects of ocean acidification are larger under limiting irradiances, while saturating light levels decrease beneficial or detrimental effects on these processes (Kranz et al., 2010; Garcia et al., 2011; Rokitta and Rost, 2012; Heiden et al., 2016). For the coccolithophore Emiliania huxleyi, for example, the impacts of ocean acidification are less detrimental under high light availability, which could partly explain why this species is moving poleward (Winter et al., 2014; Kondrik et al., 2017; Neukermans et al., 2018), although acidification is more pronounced in polar waters (Section 3.2.3.1; Cross-Chapter Paper 6). Under excess light, however, the detrimental impacts of ocean acidification are amplified for many species (high confidence) (Gao et al., 2012; Li and Campbell, 2013; Zhang et al., 2015; Kottmeier et al., 2016; Gafar et al., 2019). Lowered photo-physiological capacity to cope with high-light stress and avoid photodamage (Gao et al., 2012; Li and Campbell, 2013; Hoppe et al., 2015; Kvernvik et al., 2020) is also consistent with observations that dynamic light regimes can become more stressful under ocean acidification (Jin et al., 2013; Hoppe et al., 2015). Given the expected mixed-layer shallowing in some regions (Section 3.2.2.3), the exposure to overall higher mean irradiances could shift the effects of acidification from beneficial to detrimental for some primary producers, depending on species and organismal traits (medium confidence) (Gao et al., 2019; Seifert et al., 2020).

Studies investigating two drivers provide most of the information on the wide range of interactive effects of drivers on phytoplankton (Gao et al., 2019; Seifert et al., 2020), although climate change alters several oceanic drivers concurrently (Section 3.2). The few experimental studies that have addressed three or more drivers (Xu et al., 2014; Boyd et al., 2015b; Brennan and Collins, 2015; Brennan et al., 2017; Hoppe et al., 2018b; Moreno-Marín et al., 2018) indicate that one or two drivers generally dominate the cumulative outcome, with others playing a subordinate role (medium confidence). In these studies, temperature had a disproportionately large influence, while other drivers differed in importance, depending on the type of primary producer, ecosystem characteristics and selected driver values.

When changing CO2 concentrations affect marine ectotherms, they typically combine additively or synergistically with warming (medium confidence) (e.g., Lefevre, 2016; Reddin et al., 2020; Sampaio et al., 2021), and their cumulative effects can lead to detrimental, neutral or beneficial effects (high confidence) (Figure 3.9a; Bennett et al., 2017; Büscher et al., 2017; Dahlke et al., 2017; Foo and Byrne, 2017; Johnson et al., 2017b; Cominassi et al., 2019). Higher ocean CO2 influences the thermal tolerance of species adapted to extreme but stable habitats in tropical and polar regions, more than that of thermally tolerant generalists (high confidence) (Byrne et al., 2013; Schiffer et al., 2014; Flynn et al., 2015; Kunz et al., 2016; Pörtner et al., 2017; Kunz et al., 2018; Bindoff et al., 2019a; but see Ern et al., 2017), especially in early life stages (Dahlke et al., 2020a). In thermal generalists from temperate and subtropical species, warming and ocean acidification generally have detrimental effects on growth and survival (e.g., Gao et al., 2020), but warming can also alleviate the detrimental effects of ocean acidification by increasing metabolic rate and/or growth (Garzke et al., 2020), provided that other conditions (e.g., thermal niche, food availability) are beneficial. For example, larval growth and survival of Australasian snapper (Pagrus auratus) appear to benefit from combined acidification and warming (but see Watson et al., 2018; McMahon et al., 2020), introducing major uncertainties to population modelling (Section 3.3.4; Parsons et al., 2020).

As with ocean acidification, reduced oxygen availability further alters the influence of warming on metabolic rates (high confidence). Acidification and hypoxia can contribute to a decrease or shift in thermal tolerance, while the magnitude of this effect depends on the duration of exposure (Tripp-Valdez et al., 2017; Cattano et al., 2018; Calderón-Liévanos et al., 2019; Schwieterman et al., 2019). Warming and hypoxia are mostly positively correlated and tolerances to both phenomena are often linked after long-term acclimation (e.g., Bouyoucos et al., 2020). Acute short-term heat shocks can impair hypoxia tolerance, for instance, in intertidal fish (McArley et al., 2020). This is relevant for shallow waters, specifically for MHWs (Section 3.2.2.1; Hobday et al., 2016a; IPCC, 2018; Collins et al., 2019a). Ocean acidification can increase hypoxia tolerance in some cases, possibly by downregulating activity (Faleiro et al., 2015) and/or changing blood oxygenation (Montgomery et al., 2019). Other studies, however, reported additive negative effects of acidification and warming on hypoxia tolerance (Schwieterman et al., 2019; Götze et al., 2020), in line with the oxygen- and capacity-limited thermal tolerance (OCLTT) hypothesis presented in AR5 (Pörtner et al., 2014): Warming causes increased metabolic rates and oxygen demand in ectotherms, which at some point exceed supply capacities (which also depend on environmental oxygen availability) and reduce aerobic scope. In consequence, expansion of OMZs and other regions where warming, hypoxia and acidification combine will further reduce habitat for many fish and invertebrates (high confidence) (Sections 3.4.3.2, 3.4.3.3).

Food availability modulates, and may be more influential than, other driver responses by affecting the energetic and nutritional status of animals (Cole et al., 2016; Stiasny et al., 2019; Cominassi et al., 2020). Laboratory studies conducted under an excess of food risk underestimating the ecological effects of climate-induced drivers, because increased feeding rates may help mitigate adverse effects (Nowicki et al., 2012; Towle et al., 2015; Cominassi et al., 2020). Lowered food availability from reduced open-ocean primary production (Sections 3.2.3.3, 3.4.4.2.1) will act as an additional driver, amplifying the detrimental effects of other drivers. However, warming and higher CO2 availability may increase primary productivity in some coastal areas (Section 3.4.4.1), ameliorating the adverse direct effects on animals (e.g., Sswat et al., 2018). Due to the few studies addressing food availability under multiple-driver scenarios (Thomsen et al., 2013; Pistevos et al., 2015; Towle et al., 2015; Ramajo et al., 2016; Brown et al., 2018a; Cominassi et al., 2020), there is medium confidence in its modulating effect on climate-induced driver responses.

Animal behaviour can be affected by ocean acidification, warming and hypoxia. While warming and hypoxia mostly induce avoidance behaviour, potentially leading to migration and habitat compression (Section 3.4; McCormick and Levin, 2017; Limburg et al., 2020), the effects of acidification appear more complex. Some studies reported that acidification dominates behavioural effects (Schmidt et al., 2017), although outcomes vary with experimental design and duration of exposure (low confidence, low agreement ) (Maximino and de Brito, 2010; Munday et al., 2016; Laubenstein et al., 2018; Munday et al., 2019; Sundin et al., 2019; Clark et al., 2020; Munday et al., 2020; Williamson et al., 2021). Behaviour represents an integrated phenomenon that can be influenced both directly and indirectly by multiple drivers. For instance, increased pCO2 can directly act on neuronal signalling pathways (e.g., Gamma-aminobutyric acid hypothesis; Nilsson et al., 2012; Thomas et al., 2020) and influence learning (Chivers et al., 2014), vision (Chung et al., 2014), and choice and escape behaviour (Watson et al., 2014; Wang et al., 2017b). There is further evidence that observed alterations in fish olfactory behaviour under ocean acidification may result from physiological and molecular changes of the olfactory epithelium, influencing olfactory receptors (Roggatz et al., 2016; Porteus et al., 2018; Velez et al., 2019; Mazurais et al., 2020). Temperature mainly drives metabolic processes and thus energetic requirements, which can indirectly influence behaviour, including increased risk-taking during feeding (Marangon et al., 2020). Ocean warming also accelerates the biochemical reactions and metabolic processes that are primarily influenced by acidification. It is therefore difficult to generalise to what extent co-occurring ocean warming ameliorates or exacerbates effects of acidification on behaviour (Laubenstein et al., 2019); outcomes depend upon species and life stage (Faleiro et al., 2015; Chan et al., 2016; Tills et al., 2016; Wang et al., 2018b; Jarrold et al., 2020), interactions between species (e.g., Paula et al., 2019) along with confounding factors including food availability and salinity (medium confidence) (Ferrari et al., 2015; Pistevos et al., 2015; Pimentel et al., 2016; Pistevos et al., 2017; Horwitz et al., 2020).

While hypoxia can dominate multiple-driver responses locally (Sampaio et al., 2021), warming is the fundamental physiological driver for most marine ectotherms, globally, as it directly affects their entire biochemistry and energy metabolism. Other influential drivers include ocean acidification, salinity (high confidence) (Lefevre, 2016; Whiteley et al., 2018; Reddin et al., 2020) or food availability/quality (medium confidence) (Nagelkerken and Munday, 2016; Gao et al., 2020). Fluctuating and decreasing salinity may aggravate the detrimental effects of warming and elevated CO2, because dilution with freshwater lowers acid–base buffering capacity, resulting in lower pH and calcium carbonate saturation state (Dickinson et al., 2012; Shrivastava et al., 2019; Melzner et al., 2020).

Warm-water coral reef ecosystems house one-quarter of the marine biodiversity and provide services in the form of food, income and shoreline protection to coastal communities around the world. These ecosystems are threatened by climate-induced and non-climate drivers, especially ocean warming, MHWs, ocean acidification, SLR, tropical cyclones, fisheries/overharvesting, land-based pollution, disease spread and destructive shoreline practices (Hoegh-Guldberg et al., 2018a; Bindoff et al., 2019a; Hughes et al., 2020). Warm-water coral reefs face near-term threats to their survival (Table 3.3), but research on observed and projected impacts is very advanced.

Table 3.3 | Summary of previous IPCC assessments of coral reefs

Global analyses published since AR5 show that mass coral bleaching events and disease outbreaks have increased due to more frequent and severe heat stress associated with ocean warming (very high confidence, virtually certain) (Donner et al., 2017; Hughes et al., 2018a; DeCarlo et al., 2019; Sully et al., 2019; Tracy et al., 2019). The mass coral bleaching, which occurred continuously across different parts of the tropics from 2014 to 2016, is considered the longest and most severe global coral bleaching event on record (Section 10.4.3; see Box 15.2; Eakin et al., 2019). The Great Barrier Reef underwent mass bleaching three times between 2016 and 2020 (see Box 11.2; Pratchett et al., 2021), validating past model projections that some warm-water coral reefs would encounter bleaching-level heat stress multiple times per decade by the 2020s (Hoegh-Guldberg, 1999; Donner, 2009).

Heat stress and mass bleaching events caused decreases in live coral cover (virtually certain) (Graham et al., 2014; Hughes et al., 2018b), loss of sensitive species (extremely likely ) (Donner and Carilli, 2019; Lange and Perry, 2019; Toth et al., 2019; Courtney et al., 2020), vulnerability to disease (extremely likely ) (van Woesik and Randall, 2017; Hadaidi et al., 2018; Brodnicke et al., 2019; Howells et al., 2020) and declines in coral recruitment in the tropics (medium confidence) (Hughes et al., 2019; Price et al., 2019). Recent observations also suggest that excess nutrients can increase the susceptibility of corals to heat stress (DeCarlo et al., 2020). Changes in coral community structure due to bleaching have caused declines in reef carbonate production (high confidence) (Perry and Morgan, 2017; Lange and Perry, 2019; Perry and Alvarez-Filip, 2019; Courtney et al., 2020; van Woesik and Cacciapaglia, 2021) and in reef structural complexity (high confidence, very likely ) (Couch et al., 2017; Leggat et al., 2019; Magel et al., 2019), which increases water depth, reduces wave attenuation and increases coastal flood risk (Yates et al., 2017; Beck et al., 2018). Corals may also lose reproductive synchrony through climate change (Shlesinger and Loya, 2019), adding to their vulnerability. Bleaching and other drivers promote phase shifts to ecosystems dominated by macroalgae or other stress-tolerant species (very high confidence) (Graham et al., 2015; Stuart-Smith et al., 2018), leading to changes in reef-fish species assemblages (high confidence) (Richardson et al., 2018; Robinson et al., 2019a; Stuart-Smith et al., 2021).

Ocean acidification and associated declines in aragonite saturation state (Ωaragonite) decrease rates of calcification by corals and other calcifying reef organisms (very high confidence), reduce coral settlement (medium confidence) and increase bioerosion and dissolution of reef substrates (high confidence) (Hoegh-Guldberg et al., 2018a; Bindoff et al., 2019a; Kline et al., 2019; Pitts et al., 2020). Warming can exacerbate the coral response to ocean acidification (Kornder et al., 2018) and accelerate the decrease in coral skeletal density (Guo et al., 2020). In addition, reefs with lower coral cover and a higher proportion of slow-growing species, because of bleaching, are more sensitive to acidification (net dissolution occurs Ωaragonite= 2.3 for 100% coral cover, and Ωaragonite>3.5 for 30% coral cover; Kline et al., 2019). However, experimental evidence suggests that coral responses to ocean acidification are species specific (medium confidence) (Fabricius et al., 2011; DeCarlo et al., 2018; Comeau et al., 2019). Evidence from experiments suggests that crustose coralline algae, which contribute to reef structure and integrity and may be resistant to warming at the RCP8.5 level by 2100 (Cornwall et al., 2019), are also sensitive to declines in Ωaragonite (high confidence) (Section 3.4.2.3; Fabricius et al., 2015; Smith et al., 2020). The integrated effect of acidification, bleaching, storms and other non-climate drivers on corals, coralline algae and other calcifiers can further compromise reef integrity and ecosystem services (Rivest et al., 2017; Cornwall et al., 2018; Perry and Alvarez-Filip, 2019).

Since SROCC, there have been advances in experimental, field and modelling research on the projected response of coral cover and reef growth to bleaching and ocean acidification (Cziesielski et al., 2019; Morikawa and Palumbi, 2019; Cornwall et al., 2021; Klein et al., 2021; Logan et al., 2021; McManus et al., 2021), and on the effect of possible human interventions like assisted evolution on coral resilience (Section 3.6.3.2.2; Condie et al., 2021; Hafezi et al., 2021; Kleypas et al., 2021). New model projections incorporating physiological acclimation, larval dispersal and evolutionary processes find limited ability to adapt this century at rates of warming at or exceeding that in RCP4.5 (high confidence, very likely ) (Bay et al., 2017; Kubicek et al., 2019; Matz et al., 2020; McManus et al., 2020; Logan et al., 2021; McManus et al., 2021). For example, a global analysis (Logan et al., 2021) finds that increased thermal tolerance via evolution or switching to more stress-tolerant algal symbionts enable most (73–81%) coral to survive through 2100 under RCP2.6, but coral-dominated communities with a historical mix of coral taxa still disappear (0–8% coral survival) under RCP6.0 in simulations with adaptive mechanisms (Figure 3.13). Due to the impacts of warming, and to a lesser extent ocean acidification, global reef carbonate production is estimated to decline 71% by 2050 in SSP1-2.6, and the rate of SLR is estimated to exceed that of reef growth for 97% of reefs assessed, without adaptation by corals and their symbionts (WGI AR6 Table 9.9; Cornwall et al., 2021; Fox-Kemper et al., 2021). The increased water depth due to coral loss and reef erosion, as well as reduced structural complexity, will limit wave attenuation and exacerbate the risk of flooding from SLR on reef-fringed shorelines and reef islands (Yates et al., 2017; Beck et al., 2018; Harris et al., 2018). Local coral reef fish species richness is projected to decline due to the impacts of warming on coral cover and diversity (high confidence), with declines up to 40% by 2060 in SSP5-8.5 (Strona et al., 2021).

These observed and projected impacts are supported by geological and paleo-ecological evidence showing a decline in coral reef extent and species richness under previous episodes of climate change and ocean acidification (Kiessling and Simpson, 2011; Pandolfi et al., 2011; Kiessling et al., 2012; Pandolfi and Kiessling, 2014; Kiessling and Kocsis, 2015). Major reef crises in the past 300 million years were governed by hyperthermal events (medium confidence) (Section 3.2.4.4; Cross-Chapter Box PALEO in Chapter 1) longer in time scale than anthropogenic climate change, during which net coral reef accretion was more strongly affected than biodiversity (medium confidence).

In response to the global-scale decline in coral reefs and high future risk, recent literature focuses on finding thermal refuges and identifying uniquely resilient species, populations or reefs for targeted restoration and management (Hoegh-Guldberg et al., 2018b). Reefs exposed to internal waves (Storlazzi et al., 2020), turbidity (Sully and van Woesik, 2020) or warm-season cloudiness (Gonzalez-Espinosa and Donner, 2021) are expected to be less sensitive to thermal stress. Mesophotic reefs (30–150 m) have also been proposed as thermal refugia (Bongaerts et al., 2010), although evidence from recent bleaching events, subsurface temperature records and species overlap is mixed (Frade et al., 2018; Rocha et al., 2018b; Eakin et al., 2019; Venegas et al., 2019; Wyatt et al., 2020). A study of 2584 reef sites across the Indian and Pacific oceans estimated that 17% had sufficient cover of framework-building corals to warrant protection, 54% required recovery efforts and 28% were on a path to net erosion (Darling et al., 2019). There is medium evidence for greater bleaching resistance among reefs subject to temperature variability or frequent heat stress (Barkley et al., 2018; Gintert et al., 2018; Hughes et al., 2018a; Morikawa and Palumbi, 2019), but with trade-offs in terms of diversity and structural complexity (Donner and Carilli, 2019; Magel et al., 2019). There is limited agreement about the persistence of thermal tolerance in response to severe heat stress (Le Nohaïc et al., 2017; DeCarlo et al., 2019; Fordyce et al., 2019; Leggat et al., 2019; Schoepf et al., 2020). Recovery and restoration efforts that target heat-resistant coral populations and culture heat-tolerant algal symbionts have the greatest potential of effectiveness under future warming (high confidence) (see Box 5.5 in SROCC Chapter 5; Bay et al., 2017; Darling and Côté, 2018; Baums et al., 2019; Bindoff et al., 2019a; Howells et al., 2021); however, there is low confidence that enhanced thermal tolerance can be sustained over time (Section 3.6.3.3.2; Buerger et al., 2020). The effectiveness of active restoration and other specific interventions (e.g., reef shading) are further assessed in Section 3.6.3.3.2.

In summary, additional evidence since SROCC and SR15 (Table 3.3) finds that living coral and reef growth are declining due to warming and MHWs (very high confidence). Coral reefs are under threat of transitioning to net erosion with >1.5°C of global warming (high confidence), with impacts expected to occur fastest in the Atlantic Ocean. The effectiveness of conservation efforts to sustain living coral area, coral diversity and reef growth is limited for the majority of the world’s reefs with >1.5°C of global warming (high confidence) (Section 3.6.3.3.2; Hoegh-Guldberg et al., 2018b; Bruno et al., 2019; Darling et al., 2019).

Rocky shore ecosystems refer to a range of temperate intertidal and shallow coastal ecosystems that are dominated by different foundational organisms, including mussels, oysters, fleshy macroalgae, hard and soft corals, coralline algae, bryozoans and sponges, which create habitat for species-rich assemblages of invertebrates, fish, marine mammals and other organisms. Rocky shores provide services including wave attenuation, habitat provision and food resources, and these support commercial, recreational and Indigenous fisheries and shellfish aquaculture.

Table 3.4 | Summary of previous IPCC assessments of rocky shores

Observations since AR5 and SROCC (Table 3.4) find increased impacts of ocean warming on rocky shores. This includes extirpation of species at the warm edge of their ranges (Yeruham et al., 2015; Martínez et al., 2018), extension of poleward range boundaries (Sanford et al., 2019), mortality from climate extremes (Seuront et al., 2019), reduction in survival at shallower depths (Sorte et al., 2019; Wallingford and Sorte, 2019) and reorganisation of communities (Wilson et al., 2019; Mulders and Wernberg, 2020; Albano et al., 2021). Data collected after MHWs find ecological phase shifts (moderate evidence, high agreement ) (e.g., California; Rogers-Bennett and Catton, 2019; McPherson et al., 2021) and homogenisation of communities (limited evidence) (e.g., Alaska; Weitzman et al., 2021). For example, the collapse of sea star populations in the Northeast Pacific due to a MHW-related disease outbreak (Hewson et al., 2014; Menge et al., 2016; Miner et al., 2018; Schiebelhut et al., 2018), including 80–100% loss of the common predatory sunflower star, Pycnopodia helianthoides (very high confidence) (Harvell et al., 2019), triggered shifts from kelp- to urchin-dominated ecosystems (Schultz et al., 2016; Gravem and Morgan, 2017; McPherson et al., 2021).

Multiple lines of evidence find that foundational calcifying organisms such as mussels are at high risk of decline due to both the individual and synergistic effects of warming, acidification and hypoxia (high confidence) (Sunday et al., 2016; Sorte et al., 2017; Sorte et al., 2019; Newcomb et al., 2020). Warmer temperatures reduce mussel and barnacle recruitment (e.g., northwest Atlantic; Petraitis and Dudgeon, 2020) and the upper vertical limit of mussels (e.g., northeast Pacific, Harley, 2011; and southwest Pacific, Sorte et al., 2019). Experiments show that ocean acidification negatively impacts mussel physiology (very high confidence), with evidence of reduced growth (Gazeau et al., 2010), attachment (Newcomb et al., 2020), biomineralisation (Fitzer et al., 2014) and shell thickness (Pfister et al., 2016; McCoy et al., 2018). Net calcification and abundance of mussels and other foundational species, including oysters, are expected to decline due to ocean acidification (very high confidence) (Kwiatkowski et al., 2016; Sunday et al., 2016; McCoy et al., 2018; Meng et al., 2018), causing the reorganisation of communities (high confidence) (Kroeker et al., 2013b; Linares et al., 2015; Brown et al., 2016; Sunday et al., 2016; Agostini et al., 2018; Teixidó et al., 2018). Experiments indicate that acidification can interact with warming and hypoxia to increase the detrimental effects on mussels (Gu et al., 2019; Newcomb et al., 2020). In regions where food is readily available to mussels, detrimental effects of ocean acidification may be dampened (Kroeker et al., 2016); however, recent findings are inconclusive (Brown et al., 2018a).

Coralline algae, foundational taxa that create habitat for sea urchins and abalone, form rhodolith beds in temperate to Arctic habitats and bind together substrates, are expected to be highly susceptible to ocean acidification because they precipitate soluble magnesium calcite (Kuffner et al., 2008; Williams et al., 2021). Damage from acidification varies among species and regions, and can be due to direct physiological stress (Marchini et al., 2019) or interactions with non-calcifying competitors such as fleshy macroalgae (Smith et al., 2020). Experiments indicate that warming reduces calcification by coralline algae (high confidence) (Cornwall et al., 2019) and exacerbates the effect of acidification (Kim et al., 2020; Williams et al., 2021).

In contrast to warm-water coral reefs, there are no regional or global numerical models of rocky shore ecosystem response to projected climate change and acidification. Experiments suggest that existing genetic variation could be sufficient for some mussels (Bitter et al., 2019) and coralline algae (Cornwall et al., 2020) to adapt over generations to ocean acidification. Populations exposed to variable environments often have a greater capacity for phenotypic plasticity and resilience to environmental change [e.g., urchins (Gaitan-Espitia et al., 2017b) and coralline algae (Section 3.3.2; Rivest et al., 2017; Cornwall et al., 2018)]. Although parental conditioning within and across generations is an acclimatisation mechanism to global change, there is limited evidence from experimental studies that this is applicable for marine invertebrates on rocky shores (Byrne et al., 2020).

This assessment concludes that MHWs attributable to climate change (Section 3.2.2.1) can cause fatal disease outbreaks or mass mortality among some key foundational species (high confidence) and contribute to ecological phase shifts (medium confidence). The upper vertical limits of some species will also be constrained by climate change (high confidence). Experimental evidence since previous assessments further indicates that acidification decreases abundance and richness of calcifying species (high confidence), although there is limited evidence for acclimation in some species. Synergistic effects of warming and acidification will promote shifts towards macroalgal dominance in some ecosystems (medium confidence) and lead to reorganisation of communities (medium confidence).

Kelp are temperate, habitat-forming marine macroalgae or seaweeds, mostly of the order Laminariales, which extend across one-quarter of the world’s coastlines (Assis et al., 2020; Jayathilake and Costello, 2020). The perennial species form dense underwater forest canopies and three-dimensional habitat that provides refuge for fish, crustaceans, invertebrates and marine mammals (Filbee-Dexter et al., 2016; Wernberg et al., 2019). Kelp ecosystems support fisheries, aquaculture, fertiliser and food provision, including for local and Indigenous Peoples, along with regulating services in the form of wave attenuation and habitat provision. Kelp aquaculture can also buffer against local acidification (Xiao et al., 2021) and contribute to carbon storage (Froehlich et al., 2019).

Recent research (Straub et al., 2019; Butler et al., 2020; Filbee-Dexter et al., 2020b; Tait et al., 2021) supports the findings of previous assessments (Table 3.5) that kelp and other seaweeds in most regions are undergoing mass mortalities from high temperature extremes and range shifts from warming (very high confidence). Kelp are highly sensitive to the direct effect of high temperature on survival (Nepper-Davidsen et al., 2019) and indirect impact of temperature on herbivorous species (Ling, 2008; Vergés et al., 2016), upwelling and nutrient availability (Carr and Reed, 2015; Schiel and Foster, 2015). Synergies between warming, storms, pollution and intensified herbivory (due to removal or loss of predators including sea stars and otters that constrain herbivory by fish and urchin populations) can also cause physiological stress and physical damage in kelp, reducing productivity and reproduction (Rogers-Bennett and Catton, 2019; Beas-Luna et al., 2020; McPherson et al., 2021).

Table 3.5 | Summary of previous IPCC assessments of kelp ecosystems

Trends in kelp abundance since 1950 are uneven globally (Krumhansl et al., 2016; Wernberg et al., 2019), with population declines (e.g., giant kelp Macrocystis pyrifera in Tasmania, Butler et al., 2020; and sugar kelp Saccharina latissima in the North Atlantic, Filbee-Dexter et al., 2020b) more common than increases or no change (e.g., giant kelp Macrocystis pyrifera in southern Chile; Friedlander et al., 2020). Warming is driving range contraction and extirpation at the warm edge of species’ ranges and expansions at the cold range edge (very high confidence) (Smale, 2019; Filbee-Dexter et al., 2020b). Local declines in populations of kelp and other canopy-forming seaweeds driven by MHWs and other stressors have caused irreversible shifts to turf- or urchin-dominated ecosystems, with lower productivity and biodiversity (high confidence) (Filbee-Dexter and Scheibling, 2014; Filbee-Dexter and Wernberg, 2018; Rogers-Bennett and Catton, 2019; Beas-Luna et al., 2020; Stuart-Smith et al., 2021), ecosystems dominated by warm-affinity seaweeds or coral (high confidence) (Vergés et al., 2019), and loss of genetic diversity (Coleman et al., 2020a; Gurgel et al., 2020).

Species distribution models of kelp project range shifts and local extirpations with increasing levels of warming (Japan: Takao et al., 2015, Sudo et al., 2020; Australia: see Table 11.6, and Assis et al., 2018, Martínez et al., 2018, Castro et al., 2020; Europe: de la Hoz et al., 2019; North America: Wilson et al., 2019; South America: see Figure 12.3). There is high agreement on the direction but not the magnitude of change (Martínez et al., 2018; Castro et al., 2020), but effects of MHWs are not simulated. Where the length of higher-latitude coastlines is limited, range contractions are projected to occur, even with 2°C of global warming (i.e., SSP1-2.6) due to loss of habitat at the warm edge of species’ ranges (Martínez et al., 2018). Poleward expansion of warm-affinity herbivores, including urchins, could further reduce warm-edge kelp populations (Castro et al., 2020; Mulders and Wernberg, 2020). Evidence from natural temperate CO2 seeps suggests that ocean acidification at levels above those in RCP4.5 in 2100 could offset the increase in urchin abundance (Coni et al., 2021). Genetic analyses suggest that kelp populations at the midpoint of species’ ranges will have lower tolerance of warming than that implied by species distribution models, without assisted gene flow from warm-edge populations (King et al., 2019; Wood et al., 2021).

While reducing non-climate drivers can help prevent kelp loss from warming and MHWs, there is limited potential for restoration of kelp ecosystems after transition to urchin-dominant ecosystems (high confidence). Current restoration efforts are generally small scale (<0.1 km 2) and less advanced than those in ecosystems like coral reefs (Coleman et al., 2020b; Eger et al., 2020; Layton et al., 2020). Although abundance of herbivores limits kelp populations, there is limited evidence that restoring predators of herbivores by creating marine reserves, or directly removing grazing species, will increase kelp forest resilience to warming and extremes (Vergés et al., 2019; Wernberg et al., 2019). Active reseeding of wild kelp populations through transplantation and propagation of warm-tolerant genotypes (Coleman et al., 2020b; Alsuwaiyan et al., 2021) can overcome low dispersal rates of many kelp species and facilitate effective restoration (medium confidence) (Morris et al., 2020c).

Building on the conclusions of SROCC, this assessment finds that kelp ecosystems are expected to decline and undergo changes in community structure in the future due to warming and increasing frequency and intensity of MHWs (high confidence). Risk of loss of kelp ecosystems and shifts to turf- or urchin-dominated ecosystems are highest at the warm edge of species’ ranges (high confidence) and risks increase under RCP6.0 and RCP8.5 by the end of the century (high confidence).

Estuaries, deltas and lagoons encounter environmental gradients over small spatial scales, generating diverse habitats that support myriad ecosystem services, including food provision, regulation of erosion, nutrient recycling, carbon sequestration, recreation and tourism, and cultural significance (D’Alelio et al., 2021; Keyes et al., 2021). Although these coastal ecosystems have historically been sensitive to erosion-accretion cycles driven by sea level, drought and storms (high confidence) (Peteet et al., 2018; Wang et al., 2018c; Jones et al., 2019b; Urrego et al., 2019; Hapsari et al., 2020; Zhao et al., 2020b), they were impacted for much of the 20th century primarily by non-climate drivers (very high confidence) (Brown et al., 2018b; Ducrotoy et al., 2019; Elliott et al., 2019; He and Silliman, 2019; Andersen et al., 2020; Newton et al., 2020; Stein et al., 2020). Nevertheless, the influence of climate-induced drivers has become more apparent over recent decades (medium confidence) (Table 3.6).

Table 3.6 | Summary of previous IPCC assessments of estuaries, deltas and coastal lagoons

Estuarine biota are sensitive to warming (high confidence), with recent responses including changes in abundance of some fish stocks (Erickson et al., 2021; Woodland et al., 2021), poleward shifts in distributions of fish species, communities and associated biogeographic transition zones (Table 12.3; Franco et al., 2020; Troast et al., 2020), recruits of warm-affinity species persisting into winter (Kimball et al., 2020) and changes in seasonal timing of peaks in species abundance (Kimball et al., 2020). MHWs can be more severe in estuaries than in adjacent coastal seas (Lonhart et al., 2019), causing conspicuous impacts (very high confidence), including mass mortality of intertidal vegetation (Section 3.4.2.5), range shifts in algae and animals (Lonhart et al., 2019) and reduced spawning success among invertebrates (Shanks et al., 2020).

Relative SLR extends the upstream limit of saline waters (high confidence) (Harvey et al., 2020; Jiang et al., 2020) and alters tidal ranges (high confidence) (Idier et al., 2019; Talke et al., 2020). Elevated water levels also alter submergence patterns for intertidal habitat (high confidence) (Andres et al., 2019), moving high-water levels inland (high confidence) (Peteet et al., 2018; Appeaning Addo et al., 2020; Liu et al., 2020e) and increasing the salinity of coastal water tables and soils (high confidence) (Eswar et al., 2021). These processes favour inland and/or upstream migration of intertidal habitat, where it is unconstrained by infrastructure, topography or other environmental features (high confidence) (Kirwan and Gedan, 2019; Parker and Boyer, 2019; Langston et al., 2020; Magolan and Halls, 2020; Saintilan et al., 2020). The spread of ‘ghost forests’ along the North American east coast (Kirwan and Gedan, 2019) and elsewhere (Grieger et al., 2020) illustrates this phenomenon. Along estuarine shorelines, changing submergence patterns and upstream penetration of saline waters interact synergistically to stress intertidal plants, changing species composition and reducing above-ground biomass, in some cases favouring invasive species (Xue et al., 2018; Buffington et al., 2020; Gallego-Tévar et al., 2020). Overall, changing salinity and submergence patterns decrease the ability of shoreline vegetation to trap sediment (Xue et al., 2018), reducing accretion rates and increasing the vulnerability of estuarine shorelines to submergence by SLR and erosion by wave action (medium confidence) (Zhu et al., 2020b).

Drought and freshwater abstraction can reduce freshwater inflows to estuaries and lagoons, increasing salinity, reducing water quality (Brooker and Scharler, 2020) and depleting resident macrophyte communities (Scanes et al., 2020b). Changes in freshwater input and SLR, combined with land-use change, can alter inputs of land-based sediments, causing expansion (Suyadi et al., 2019; Magolan and Halls, 2020) or contraction (Andres et al., 2019; Appeaning Addo et al., 2020; Li et al., 2020b) of intertidal habitats. The same phenomena alter salinity gradients, which are the primary drivers of estuarine species distributions (high confidence) (Douglass et al., 2020; Lauchlan and Nagelkerken, 2020). Extreme reduction of freshwater input can extend residence time of estuarine water, leading to persistent HABs (Lehman et al., 2020) and converting estuaries to lagoons if the mouth clogs with sediment (Thom et al., 2020).

Acidification of estuarine water is a growing hazard (medium confidence) (Doney et al., 2020; Scanes et al., 2020a; Cai et al., 2021), and resident organisms display sensitivity to altered pH in laboratory settings (medium confidence) (Young et al., 2019a; Morrell and Gobler, 2020; Pardo and Costa, 2021). However, attribution of the biological effects of acidification is difficult because many biogeochemical processes affect estuarine carbon chemistry (Sections 3.2.3.1, 3.3.2). Warming can exacerbate the impacts of both acidification and hypoxia on estuarine organisms (Baumann and Smith, 2018; Collins et al., 2019b; Ni et al., 2020). These effects are further complicated by eutrophication, with high nitrogen loads associated with lower pH (Rheuban et al., 2019). Warming (including MHWs) and eutrophication interact to decrease estuarine oxygen content and pH, increasing the vulnerability of animals to MHWs (Brauko et al., 2020) and exacerbating the incidence and impact of dead zones (medium confidence) (Altieri and Gedan, 2015). The impacts of storms on estuaries are variable and are described in SM3.3.1.

All these impacts are projected to escalate under future climate change, but their magnitude depends on the amount of warming, the socioeconomic development pathway and implementation of adaptation strategies (medium confidence). Modelling studies (Lopes et al., 2019; Rodrigues et al., 2019; White et al., 2019; Zhang and Li, 2019; Hong et al., 2020; Krvavica and Ružić, 2020; Liu et al., 2020e; Shalby et al., 2020) suggest that responses of estuaries to SLR will be complex and context dependent (Khojasteh et al., 2021), but project that salinity, tidal range, storm-surge amplitude, depth and stratification will increase with SLR (medium confidence), and that marine-dominated waters will penetrate farther upstream (high confidence). Without careful management of freshwater inputs, sediment augmentation and/or the restoration of shorelines to more natural states, transformation and loss of intertidal areas and wetland vegetation will increase with SLR (high confidence) (Doughty et al., 2019; Leuven et al., 2019; Yu et al., 2019; Raw et al., 2020; Shih, 2020; Stein et al., 2020), with small, shallow microtidal estuaries being more vulnerable to impacts than deeper estuaries with well-developed sediments (medium confidence) (Leuven et al., 2019; Williamson and Guinder, 2021). Warming and MHWs will enhance stratification and deoxygenation in shallow lagoons (medium confidence) (Derolez et al., 2020) and will continue to drive range shifts among estuarine biota (medium confidence) (Veldkornet and Rajkaran, 2019; Zhang et al., 2020c), resulting in extirpations where thermal habitat is lost and potentially generating new habitat for warm-affinity species (limited evidence, medium agreement ) (Veldkornet and Rajkaran, 2019).

Mangroves, salt marshes and seagrass beds (wetland ecosystems) are considered ‘blue carbon’ ecosystems due to their capacity to accumulate and store organic-carbon rich sediments (see Box 3.4; Macreadie et al., 2019; Rogers et al., 2019) and provide an extensive range of other ecosystem services (see Box 3.4). Because these ecosystems are often found within estuaries and along sheltered coastlines, they share vulnerabilities, climate-induced drivers (Table 3.7) and non-climate drivers with estuaries and coastal lagoons (Section 3.4.2.4).

Table 3.7 | Summary of previous IPCC assessments of mangroves, salt marshes and seagrass beds

Since AR5 and SROCC, syntheses have emphasised that the vulnerability of rooted wetland ecosystems to climate-induced drivers is exacerbated by non-climate drivers (high confidence) (Elliott et al., 2019; Ostrowski et al., 2021; Williamson and Guinder, 2021) and climate variability (high confidence) (Day and Rybczyk, 2019; Kendrick et al., 2019; Shields et al., 2019). Global rates of mangrove loss have been extensive but are slowing (high confidence) at least partially due to management interventions (Friess et al., 2020b; Goldberg et al., 2020). From 2000 to 2010 mangrove loss averaged 0.16% yr –1, globally, but with greatest loss in Southeast Asia (high confidence) (Hamilton and Casey, 2016; Friess et al., 2019; Goldberg et al., 2020) and ubiquitous fragmentation leaving few mangroves intact (Bryan-Brown et al., 2020). Salt-marsh ecosystems have also suffered extensive losses (up to 60% in places since the 1980s), especially in developed and rapidly developing countries (medium confidence) (Table 12.3; Gu et al., 2018; Stein et al., 2020). Similarly, 29% of seagrass meadows were lost from 1879–to 2006 due primarily to coastal development and degradation of water quality, with climate-change impacts escalating since 1990 (medium confidence) (Waycott et al., 2009; Sousa et al., 2019; Derolez et al., 2020; Green et al., 2021a). Local examples of habitat stability or growth (e.g., de los Santos et al., 2019; Laengner et al., 2019; Sousa et al., 2019; Suyadi et al., 2019; Derolez et al., 2020; Goldberg et al., 2020; McKenzie and Yoshida, 2020) indicate some resilience to climate change in the absence of non-climate drivers (high confidence). Nevertheless, previous declines have left wetland ecosystems more vulnerable to impacts from climate-induced drivers and non-climate drivers (high confidence) (Friess et al., 2019; Williamson and Guinder, 2021).

Warming and MHWs have affected the range, species composition and survival of some wetland ecosystems. Warming is allowing some, but not all (Rogers and Krauss, 2018; Saintilan et al., 2018), mangrove species to expand their ranges poleward (high confidence) (Friess et al., 2019; Whitt et al., 2020). This expansion can affect species interactions (Guo et al., 2017; Friess et al., 2019), and enhance sediment accretion and carbon storage rates in some instances (medium confidence) (Guo et al., 2017; Kelleway et al., 2017; Chen et al., 2018b; Coldren et al., 2019; Raw et al., 2019). Drought, low sea levels and MHWs can cause significant die-offs among mangroves (medium confidence) (Lovelock et al., 2017b; Duke et al., 2021). Seagrasses are similarly vulnerable to warming (high confidence) (Repolho et al., 2017; Duarte et al., 2018; Jayathilake and Costello, 2018; Savva et al., 2018), which has been attributed as one cause of observed changes in distribution and community structure (medium confidence) (Hyndes et al., 2016; Nowicki et al., 2017). MHWs, together with storm-driven turbidity and structural damage, can cause seagrass die-offs (high confidence) (Arias-Ortiz et al., 2018 ; Kendrick et al., 2019; Smale et al., 2019; Strydom et al., 2020), shifts to small, fast-growing species (high confidence) (Kendrick et al., 2019; Shields et al., 2019; Strydom et al., 2020) and ecosystem collapse (Serrano et al., 2021).

The sensitivity of salt marshes and mangroves to RSLR depends on whether they accrete inorganic sediment and/or organic material at rates equivalent to rising water levels (very high confidence) (Peteet et al., 2018; FitzGerald and Hughes, 2019; Friess et al., 2019; Gonneea et al., 2019; Leo et al., 2019; Marx et al., 2020; Saintilan et al., 2020). Otherwise, wetland ecosystems must migrate either inland or upstream, or face gradual submergence in deeper, increasingly saline water (very high confidence) (Section 3.4.2.4; Andres et al., 2019; Jones et al., 2019b; Cohen et al., 2020; Mafi-Gholami et al., 2020; Magolan and Halls, 2020; Sklar et al., 2021). Ability to migrate depends on local topography, the positioning of anthropogenic infrastructure and structures placed to defend such infrastructure (Schuerch et al., 2018; Fagherazzi et al., 2020; Cahoon et al., 2021). Submergence drives changes in community structure (high confidence) (Jones et al., 2019b; Yu et al., 2019; Douglass et al., 2020; Langston et al., 2020) and functioning (high confidence) (Charles et al., 2019; Buffington et al., 2020; Stein et al., 2020), and will eventually lead to extirpation of the most sensitive vegetation (medium confidence) (Schepers et al., 2017; Scalpone et al., 2020) and associated animals (low confidence) (Rosencranz et al., 2018). The impacts of storms on wetlands are variable and described in SM3.3.1.

Table 3.8 | Estimates of vulnerability of coastal wetlands to sea level rise (SLR) on the basis of sediment cores

Notes:

(a) Estimate digitised from published figure

(b) Published figure digitised and remodelled as binomial generalised linear model (number drowned as compared with not drowned)

As noted in SROCC, given the diversity of coastal wetlands as well as the dependence of their future vulnerability to climate change on adaptation pathways (Krauss, 2021; Rogers, 2021), projections of future impacts based on shoreline elevation estimated from satellite data and CMIP5 projections (Spencer et al., 2016; Schuerch et al., 2018) vary greatly. Although all approaches have individual strengths and weaknesses (Törnqvist et al., 2021), paleorecords provide some clarity because they yield estimates of wetland responses to changes in climate in the absence of other anthropogenic drivers and are therefore inherently conservative. On the basis of paleorecords (Table 3.8), we assess that mangroves and salt marshes are likely at high risk from future SLR, even under SSP1-1.9, with impacts manifesting in the mid-term (medium confidence). Under SSP5-8.5, wetlands are very likely at high risk from SLR, with larger impacts manifesting before 2040 (medium confidence). By 2100, these ecosystems are at high risk of impacts under all scenarios except SSP1-1.9 (high confidence), with impacts most severe along coastlines with gently sloping shorelines, limited sediment inputs, small tidal ranges and limited space for inland migration (very high confidence) (Cross-Chapter Box SLR in Chapter 3; Schuerch et al., 2018; FitzGerald and Hughes, 2019; Leo et al., 2019; Schuerch et al., 2019; Raw et al., 2020; Saintilan et al., 2020).

For seagrasses, recent projections for climate-change impacts vary by species and region. Warming is projected to increase the habitat available to Zostera marina on the east coast of the USA by 2100 but contract its southern range edge by 150–650 km under RCP2.6 and RCP8.5, respectively (Wilson and Lotze, 2019 ). Other species, such as Posidonia oceanica in the Mediterranean, might lose as much as 75% of their habitat by 2050 under RCP8.5 and become functionally extinct (low confidence) by 2100 (Chefaoui et al., 2018). Observed impacts of MHWs (Kendrick et al., 2019; Strydom et al., 2020; Serrano et al., 2021) indicate that increasing intensity and frequency of MHWs (Section 3.2.2.1) will have escalating impacts on seagrass ecosystems (high confidence). Habitat suitability can also be reduced by moderate RSLR, due to its impact on light attenuation (medium confidence) (Aoki et al., 2020; Ondiviela et al., 2020; Scalpone et al., 2020).

Overall, warming will drive range shifts in wetland species (medium to high confidence), but SLR poses the greatest risk for mangroves and salt marshes, with significant losses projected under all future scenarios by mid-century (medium confidence) and substantially greater losses by 2100 under all scenarios except SSP1-1.9 (high confidence). MHWs pose the greatest risk to seagrasses (high confidence). In all cases, losses will be greatest where accommodation space is constrained or where other non-climate drivers exacerbate risk from climate-induced drivers (very high confidence).

Table 3.9 | Summary of previous IPCC assessments of sandy beaches.

Sandy beaches comprise unvegetated, fine- to medium-grained sediments in the intertidal zones that line roughly one-third of the length of the world’s ice-free coastlines (Luijendijk et al., 2018). The amenity value of beaches as cultural, recreational and residential destinations has driven extensive urbanisation of beach-associated coastlines (Todd et al., 2019). Beaches also provide habitat for many resident species, nesting habitat for marine vertebrates, filtration of coastal waters and protection of the coastline from erosion (McLachlan and Defeo, 2018). These soft-sediment coastal ecosystems are particularly vulnerable to habitat loss caused by erosion, especially where landward transgression is inhibited by infrastructure (Table 3.9).

Since SROCC, observed trends in coastal erosion continue to be obscured by beach nourishment that replaces eroded sediment or by coastal protection of areas at risk of erosion (Section 3.6.3.1.1; Cross-Chapter Box SLR in Chapter 3). Nevertheless, RSLR, increases in wave energy and/or changes in wave direction, disruptions to sediment supplies (including sand mining) and other anthropogenic modifications of the coast have driven localised beach erosion (very high confidence) at rates up to 0.5–3 m yr –1 (Vitousek et al., 2017a; Vitousek et al., 2017b; Cambers and Wynne, 2019; Enríquez-de-Salamanca, 2020; Sharples et al., 2020). Corresponding analyses of coarse-scale (30-m resolution) global data estimate that 15% of tidal flats (including beaches) have been lost since 1984 (medium confidence) (Mentaschi et al., 2018; Murray et al., 2019) but with a corresponding number of the world’s beaches accreting (28%) as eroding (24%) (medium confidence) (Luijendijk et al., 2018).

Progress is being made towards models that can project beach erosion under future scenarios despite inherent uncertainties and the presence of multiple confounding drivers in the coastal zone (Vitousek et al., 2017b; Le Cozannet et al., 2019; Cooper et al., 2020a; Vousdoukas et al., 2020b; Vousdoukas et al., 2020a). In the interim, models with varying levels of complexity estimate local loss of beach area to SLR by 2100 under RCP8.5-like scenarios, assuming minimal human intervention, ranging 30–70% (low confidence) (Vitousek et al., 2017b; Mori et al., 2018; Ritphring et al., 2018; Hallin et al., 2019; Kasmi et al., 2020). Within regions, projected impacts scale negatively with beach width and positively with the magnitude of projected SLR. None of these local studies, however, considered high-energy storm events, which are known to also impact sandy coasts (high confidence) (e.g., Burvingt et al., 2018; Garrote et al., 2018; Duvat et al., 2019; Sharples et al., 2020), and model structure often had more influence on projected shoreline responses than did physical drivers (Le Cozannet et al., 2019). Nevertheless, the most-advanced available models, which incorporate multiple coastal processes, including SLR, project that without anthropogenic barriers to erosion, 13.6–15.2% and 35.7–49.5% of the world’s beaches likely risk undergoing at least 100 m of shoreline retreat (relative to 2010) by 2050 and 2100, respectively (low confidence) (Vousdoukas et al., 2020b). Aggregating these trends regionally suggests that relative rates of shoreline change under RCP4.5 and RCP8.5 diverge strongly after mid-century, with trends towards erosion escalating under RCP8.5 by 2100 (medium confidence) (Figure 3.14; Vousdoukas et al., 2020b). This trend supports the WGI AR6 assessment that projected SLR will contribute to erosion of sandy beaches, especially under high-emissions futures (high confidence) (WGI AR6 Technical Summary; Arias et al., 2021).

For beach fauna, emerging evidence links range shifts, increasing representation by warm-affinity species and mass mortalities to ocean warming (limited evidence, high agreement ) (McLachlan and Defeo, 2018; Martin et al., 2019). But even amongst the best-studied taxa, such as turtles, vulnerability to warming (high confidence) and SLR (medium confidence) anticipated on the basis of theory (Poloczanska et al., 2009; Saba et al., 2012; Pike, 2013; Laloë et al., 2017; Tilley et al., 2019) yields only a few detected impacts in the field associated mainly with feminisation (female-skewed sex ratios driven by warmer nest temperatures) (Jensen et al., 2018; Colman et al., 2019; Tilley et al., 2019), phenology (Monsinjon et al., 2019), reproductive success (Bladow and Milton, 2019) and inter-nesting period (Valverde-Cantillo et al., 2019). Moreover, although established vulnerabilities imply high projected future risk for turtles (high confidence) (e.g., Almpanidou et al., 2019; Monsinjon et al., 2019; Patrício et al., 2019; Varela et al., 2019; Santidrián Tomillo et al., 2020), many populations remain resilient to change (Fuentes et al., 2019; Valverde-Cantillo et al., 2019; Laloë et al., 2020; Lamont et al., 2020), perhaps because variation in sand temperatures at nesting depth among beaches very likely exceeds the magnitude of warming anticipated by 2100, even under RCP8.5 (medium confidence) (Bentley et al., 2020a). As expected for a taxon with a long evolutionary history, turtles display natural adaptation, not only by virtue of broad geographic distributions that include natural climate-change refugia (Boissin et al., 2019; Jensen et al., 2019), but also because some initial responses to warming might counteract anticipated impacts. For example, although feminisation poses a significant long-term risk to turtle populations (high confidence), it might contribute to population growth in the near to mid-term (medium confidence) (Patrício et al., 2019). Resilience to climate change might be further enhanced by range extensions, alterations in nesting phenology and fine-scale nest-site selection (medium confidence) (Abella Perez et al., 2016; Santos et al., 2017; Almpanidou et al., 2018; Rivas et al., 2019; Laloë et al., 2020).

New literature since SROCC on climate impacts and risks has been scarce for most beach taxa besides turtles. (The impacts of storms on beach fauna are variable and are described in SM3.3.1.) Nevertheless, theoretical sensitivity to warming (Section 3.3.2), together with the projected loss of habitat under future climate scenarios, suggest substantial impacts for populations and communities of beach fauna into the future (high confidence). These impacts will be exacerbated by coastal squeeze along urbanised coastlines (high confidence), albeit with magnitudes that cannot yet be accurately projected (McLachlan and Defeo, 2018; Le Cozannet et al., 2019; Leo et al., 2019).

This section assesses impacts on five SES, or seas larger than 200,000 km 2 with single entrances <120 km wide, including the Persian Gulf, the Red Sea, the Black Sea, the Baltic Sea and the Mediterranean Sea. These SES are largely landlocked and are thus heavily influenced by surrounding landscapes, local and global climate-induced drivers, as well as non-climate drivers (Section 3.1), making them highly vulnerable to cumulative threats. Key climate-induced drivers in SES are warming, increasing frequency and duration of MHWs, acidification and the increasing in size and number of OMZs (Figure 3.12; Hoegh-Guldberg et al., 2014). In AR5, SES were recognised as regionally significant for fisheries and tourism but highly exposed to both local and global stressors, offering limited options for organisms to migrate in response to climate change (Table 3.10).

Table 3.10 | Summary of past IPCC assessments of semi-enclosed seas (SES)

Since AR5, there is evidence for increasing frequency and duration of MHWs, extreme-weather events and a diversity of threats across depth strata causing mass-mortality events, local extirpations and coral reef decline (high confidence) (Section 3.4.2.1; SM3.3.2; Buchanan et al., 2016a; Shlesinger et al., 2018; Wabnitz et al., 2018b; Garrabou et al., 2019). In most SES, non-climate drivers, including pollution, habitat destruction and especially overfishing, are decreasing the local adaptive capacity of organisms and the ability of ecosystems to cope with climate-change impacts (high confidence) (Cramer et al., 2018; Hidalgo et al., 2018; Ben-Hasan and Christensen, 2019). The SLR is accelerating faster than expected (high confidence) (Kulp and Strauss, 2019), posing a key risk to SES’ coastal ecosystems and the services they provide in urban areas, including drinking water provision, housing and recreational activities, among others (Hérivaux et al., 2018; Reimann et al., 2018).

The size and number of OMZs are increasing worldwide and in most SES (high confidence) (Global Ocean Oxygen Network, 2018), with growing impacts on fish species diversity and ecosystem functioning. In the Persian Gulf and Red Sea, increasing nutrient loads associated with coastal activities and warming has increased the size of OMZs (high confidence) (Al-Said et al., 2018; Lachkar et al., 2019). OMZs represent an even greater problem in the Black and Baltic seas, with broad implications for ecosystem function and services (Levin et al., 2009), especially where actions to reduce nutrient loading from land have been unable to reduce the OMZ coverage (high confidence) (Carstensen et al., 2014; Miladinova et al., 2017; Global Ocean Oxygen Network, 2018). In the Baltic Sea, OMZs are affecting the extent of suitable spawning areas of cod, Gadus morhua (high confidence) (Hinrichsen et al., 2016), while in the Black Sea, the combined effect of OMZs and warming is influencing the distribution and physiology of fish species, and their migration and schooling behaviour in their overwintering grounds (medium confidence) (Güraslan et al., 2017). Cascading effects on food webs have been reported in the Baltic, where detrimental effects of changing oxygen levels on zooplankton production, pelagic and piscivorous fish are influencing seasonal succession and species composition of phytoplankton (high confidence) (Viitasalo et al., 2015).

In the Mediterranean Sea (Cross-Chapter Paper 4), the increase in climate extremes and mass-mortality events reported in AR5 has continued (very high confidence) (Gómez-Gras et al., 2021). Extreme-weather events (including deep convection; González-Alemán et al., 2019) and MHWs have become more frequent (Darmaraki et al., 2019) and are associated with mass mortality of benthic sessile species across the basin (high confidence) (Garrabou et al., 2019; Gómez-Gras et al., 2021). Since AR5, in the Persian Gulf and Red Sea, extreme temperatures, together with disease and predation, have continued to cause bleaching-induced mortality of corals, along with declines in the average coral-colony size (high confidence) (Burt et al., 2019). Poleward migration and tropicalisation of species (Section 3.4.2.3) has also continued in the Mediterranean, and these phenomena have also become an issue in the Black Sea (high confidence) (Boltachev and Karpova, 2014; Hidalgo et al., 2018). Climate impacts on phytoplankton production and phenology show high spatial heterogeneity across the Mediterranean Sea (medium evidence) (Marbà et al., 2015b; Salgado-Hernanz et al., 2019), with consequent effects on the diversity and abundance of zooplankton and fish species (medium confidence) (Peristeraki et al., 2019). Changes in primary production and a decrease in river runoff have also altered the optimum habitats for small pelagic fish in the Mediterranean, from the local to the basin scale (Piroddi et al., 2017). Evidence of impacts from ocean acidification is increasing, with the rates of coral calcification showing major decline in the Red Sea (medium confidence) (Section 3.4.2.1; Steiner et al., 2018; Bindoff et al., 2019a). In the Mediterranean Sea, evidence of acidification events have been reported at the local scale (Hassoun et al., 2015), with impacts on bivalves and coralligenous species (medium confidence) (Lacoue-Labarthe et al., 2016).

Climate models project increasing frequency and intensity of MHWs (high confidence) (Section 3.2.2.1), which will exacerbate warming-driven impacts in the Red Sea and Persian Gulf regions, and erode the resilience of Red Sea coral reefs (high confidence) (Osman et al., 2018; Genevier et al., 2019; Kleinhaus et al., 2020). In the Persian Gulf region, extreme temperatures, >35°C (Pal and Eltahir, 2016), have been linked with high rates of extirpation and a decrease in fisheries catch potential (medium confidence) (Wabnitz et al., 2018b). In the Mediterranean Sea, east–west gradients in rates of warming are projected to trigger spatially different changes in primary production, which combined with the increasing arrival of non-indigenous species, may trigger biogeographic changes in fish diversity, increasing in the eastern and decreasing in the western Mediterranean (medium to high confidence) (Albouy et al., 2013; Macias et al., 2015). Projections also show greater impacts from SLR than originally expected in the Mediterranean and Baltic (e.g., Dieterich et al., 2019; Thiéblemont et al., 2019). In the Baltic Sea, under high nutrient load and warming climate scenarios, eutrophication is projected to increase in the future (2069–2098) compared with historical (1976–2005) periods. In contrast, under continued nutrient load reductions following present management regulations, environmental conditions and ecological state will continue to improve independently of the climate-warming scenarios (low to medium confidence) (Saraiva et al., 2019).

Shelf seas overlie the continental margin, often with maximum depths of <200 m, and represent 7% of the global ocean by area (Simpson and Sharples, 2012). These ecosystems are found offshore of every continent, generate 10–30% (Mackenzie et al., 2000; Andersson and Mackenzie, 2004) of global marine net primary production and play a key role in global biogeochemical cycling, including the export of land-borne carbon and nutrients (Johnson et al., 1999; Nishioka et al., 2011; Li et al., 2019) to the deep ocean and recycling of fixed nitrogen back to the atmosphere via denitrification (Devol, 2015). The shelf seas are home to several of the world’s major industrial capture fisheries, such as those of the mid-Atlantic Bight, Scotian Shelf, Eastern Bering Sea Shelf and North Brazil Shelf (Barange et al., 2018), and support other marine industries, including aquaculture, extractive industries (oil, gas and mining), shipping and renewable energy installations.

Table 3.11 | Summary of past IPCC assessments of shelf seas

Similar to other coastal ecosystems, evidence since SROCC (Table 3.11) suggests that shelf-sea ecosystems and the fisheries and aquaculture they support are sensitive to the interactive effects of climate-induced drivers, as well as non-climate drivers, including nutrient pollution, sedimentation, fishing pressure and resource extraction (Table 3.12; Figure 3.12). Changes in freshwater, nutrient and sediment inputs from rivers due to both climate-induced and non-climate drivers can influence productivity and nutrient limitation, ecosystem structure, carbon export and species diversity and abundance (Balch et al., 2012; Picado et al., 2014), and can result in reduced water clarity and light penetration (Dupont and Aksnes, 2013; McGovern et al., 2019). Seasonal bottom-water hypoxia occurs in some shelf seas (e.g., northern Gulf of Mexico, Bohai Sea, East China Sea) due to riverine inputs of freshwater and nutrients, promoting stratification, enhanced primary production and organic carbon export to bottom waters (high confidence) (Zhao et al., 2017; Wei et al., 2019; Del Giudice et al., 2020; Große et al., 2020; Jarvis et al., 2020; Rabalais and Baustian, 2020; Song et al., 2020a; Xiong et al., 2020; Zhang et al., 2020a).

Table 3.12 | Synthesis of interactive effects and their influence on shelf-sea ecosystems and the fisheries and aquaculture they support

Key risks to shelf seas include shifts or declines in marine micro- and macro-organism abundance and diversity driven by eutrophication, HABs and extreme events (storms and MHWs), and consequent effects on fisheries, resource extraction, transportation, tourism and marine renewable energy (Figure 3.12). The combined effects of deoxygenation and warming can affect the metabolism, growth, feeding behaviour and mobility of fish species (Section 3.3.3). The increasing availability of observations mean that ecosystem changes in shelf seas can be increasingly attributed to climate change (high confidence) (Liang et al., 2018; Maharaj et al., 2018; Ma et al., 2019; Meyer and Kröncke, 2019; Bargahi et al., 2020; Bedford et al., 2020; Friedland et al., 2020b; Mérillet et al., 2020). Eutrophication and seasonal bottom-water hypoxia in some shelf seas have been linked to warming (high confidence) (Wei et al., 2019; Del Giudice et al., 2020) and increased riverine nutrient loading (high confidence) (Wei et al., 2019; Del Giudice et al., 2020). Since SROCC, some severe HABs have been attributed to extreme events, such as MHWs (Section 14.4.2; Roberts et al., 2019; Trainer et al., 2019); however, a recent worldwide assessment of HABs attributed the increase in observed HABs to intensified monitoring associated with increased aquaculture production (high confidence) (Hallegraeff et al., 2021).

Since SROCC, changes in the community structure and diversity of plankton, macrofauna and infauna have been detected in some shelf seas, although attribution has been regionally specific (e.g . , bottom-water warming or hypoxia) (Meyer and Kröncke, 2019; Rabalais and Baustian, 2020). Detection of the picoplankton Synechococcus spp. in the North Sea is potentially linked to a summer decrease in copepod stocks and declining food-web efficiency (low confidence) (Schmidt et al., 2020). The seasonally distinct phytoplankton assemblages in the North Sea have begun to appear concurrently and homogenise (Nohe et al., 2020). Changes in abundance, species composition and size of zooplankton have been detected in some shelf seas (Yellow Sea, North Sea, Celtic Sea and Tasman Sea), including a decline in stocks of larger copepods, increased abundances of gelatinous and meroplankton, and a shift to smaller species due to warming, increased river discharge, circulation change and/or extreme events (high confidence) (Wang et al., 2018a; Bedford et al., 2020; Evans et al., 2020; Shi et al., 2020; Edwards et al., 2021).

Ocean warming has shifted distributions of fish (Free et al., 2019; Franco et al., 2020; Pinsky et al., 2020b; Fredston et al., 2021) and marine mammal species (Salvadeo et al., 2010; García-Aguilar et al., 2018; Davis et al., 2020) poleward (high confidence) or deeper (low to medium confidence) (Section 3.4.3.1; Nye et al., 2009; Pinsky et al., 2013; Pinsky et al., 2020b). Warming has also tropicalised the pelagic and demersal fish assemblages of mid- and high-latitude shelves (high confidence) (Montero-Serra et al., 2015; Liang et al., 2018; Maharaj et al., 2018; Ma et al., 2019; Friedland et al., 2020a; Kakehi et al., 2021; Punzón et al., 2021). Fisheries catch composition in many shelf-sea ecosystems has become increasingly dominated by warm-water species since the 1970s (high confidence) (Cheung et al., 2013; Leitão et al., 2018; Maharaj et al., 2018; McLean et al., 2019). Warming has taxonomically diversified fish communities along a latitudinal gradient in the North Sea but has homogenised functional diversity (McLean et al., 2019). However, in some regions, changing predator or prey distributions, temperature-dependent hypoxia, population changes, evolutionary adaptation and other biotic or abiotic processes, including species’ exploitation, confound responses to climate-induced drivers, which must therefore be interpreted with caution (Frank et al., 2018). For example, although, most species’ range edges are tracking temperature change on the northeast shelf of the USA (medium confidence) (Fredston-Hermann et al., 2020; Fredston et al., 2021), range edges of others are not.

A wide range of responses by fish and invertebrate populations to warming have been observed. The majority of responses have been detrimental, with the direction and magnitude of the response depending on ecoregion, taxonomy, life history and exploitation history (Free et al., 2019; Yati et al., 2020). For example, fisheries productivity has strongly decreased in the North Sea (Free et al., 2019), and fisheries yields have also decreased in the Celtic Sea, attributed primarily to warming and secondarily to long-term exploitation (Hernvann and Gascuel, 2020; Mérillet et al., 2020). Conversely, fish species diversity and overall productivity have increased in the Gulf of Maine, even with warming (Le Bris et al., 2018; Friedland et al., 2020a; Friedland et al., 2020b). Fisheries yields have decreased in the Yellow Sea, East China Sea and South China Sea partially due to overexploitation (Ma et al., 2019; Wang et al., 2019c), with warming exerting more influence on the yield of cold-water species than on temperate- and warm-water groups (Ma et al., 2019). The combined effects of exploitation and multi-decadal climate fluctuations make it difficult to assess global climate-change impacts on fisheries yields (Chapter 5; Ma et al., 2019; Bentley et al., 2020b; Johnson et al., 2020).

Since AR5, increasing spatio-temporal extent of hypoxia has been projected due to enhanced benthic respiration and reduced oxygen solubility from warming (Del Giudice et al., 2020). Similar to the open ocean, large shifts in the phenology of phytoplankton blooms have been projected for shelf seas throughout subpolar and polar waters (medium confidence) (Henson et al., 2018a; Asch et al., 2019). Zooplankton, which are important prey for many fish species and sea birds, are expected to decrease in abundance on the northeast shelf of the USA (Grieve et al., 2017); however, responses vary by shelf ecosystem (Chust et al., 2014b). Trends towards tropicalisation will continue in the future (high confidence) (Cheung et al., 2015; Stortini et al., 2015; Allyn et al., 2020; Maltby et al., 2020; Costa et al., 2021), but uncertainty of future projections of fisheries production increases substantially beyond 2040 (Maltby et al., 2020). Nevertheless, shelf-sea fisheries at lower latitudes are most vulnerable to climate change (Monnereau et al., 2017). Under future climate change marked by more frequent and intense extreme events and the influences of multiple drivers, more flexible and adaptive management approaches could reduce climate impacts on species while also supporting industry adaptation (high confidence) (Section 3.6.3.1.2; Shackell et al., 2014; Stortini et al., 2015; Hare et al., 2016; Stortini et al., 2017; Greenan et al., 2019; Ocaña et al., 2019; Maltby et al., 2020).

Eastern boundary upwelling systems (EBUS) comprise four important social–ecological systems in the Pacific (California and Peru-Humboldt) and Atlantic (Canary and Benguela) ocean basins. Each is characterised by high primary production, sustained by wind-driven upwelling that draws cold, nutrient-rich, generally low-pH and low-oxygen water to the surface (Bindoff et al., 2019a). Despite their small relative size, the primary productivity in EBUS supports a vast biomass of marine consumers, including some of the world’s most productive fisheries (Pauly and Zeller, 2016), along with many species of conservation significance (Bakun et al., 2015).

Although upwelling is important in many other oceanic regions, we focus here on the most documented examples provided by the EBUS. Yet even here, observed changes in upwelling, temperature, acidification and loss of oxygen (Seabra et al., 2019; Abrahams et al., 2021; Gallego et al., 2021; Varela et al., 2021) cannot be robustly attributed to anthropogenic climate change, and projected future changes in upwelling are expected to be relatively small and variable among and within EBUS (Section 3.2.2.3; WGI AR6 Chapter 9; Fox-Kemper et al., 2021). We therefore have few updates to assessments provided by AR5 and SROCC (Table 3.13) and restrict our brief assessment to the limited amount of new evidence (Figure 3.12).

Table 3.13 | Summary of previous IPCC assessments of eastern boundary upwelling systems (EBUS)

The California EBUS is arguably the best-studied of the four ecosystems in terms of robust projections of climate change, although even here, there is limited evidence and low agreement among projections. For example, trends in outputs from high-resolution, downscaled models in the California EBUS generally reflect those from underlying coarser-scale ESMs, but projections for physical variables are more convergent among modelling approaches than are those for biogeochemical variables (high confidence) (Howard et al., 2020a; Pozo Buil et al., 2021). Models agree on general warming in the California EBUS, with concomitant declines in oxygen content (medium confidence) (Howard et al., 2020b; Fiechter et al., 2021; Pozo Buil et al., 2021). But implications for the future spatial distribution of species, including for some fisheries resources (Howard et al., 2020b; Fiechter et al., 2021), are confounded by local-scale oceanographic processes (Siedlecki et al., 2021) and by lateral input of anthropogenic land-based nutrients (Kessouri et al., 2021), suggesting that such projections should be accorded low confidence.

More generally, changes in upwelling intensity are observed to affect organismal metabolism, population productivity and recruitment, and food-web structure (medium confidence) (van der Sleen et al., 2018; Brodeur et al., 2019; Ramajo et al., 2020). But low confidence in projected trends in upwelling make it difficult to extrapolate these results to understand potential changes in the ecology of EBUS. Projected changes in fish biomass within EBUS (Carozza et al., 2019) are therefore accorded low confidence. Finally, although MHWs are an important emerging hazard in the global ocean, with intensity, frequency and duration increasing strongly (Section 3.2.2.1), the number of MHW days yr –1 within EBUS has been increasing more slowly (or decreasing faster, in the case of the Peru-Humboldt system) than in surrounding waters (Varela et al., 2021). Notwithstanding these trends, EBUS remain vulnerable both to MHWs (high confidence) (Sen Gupta et al., 2020) and to their long-lasting impacts (high confidence) (Arafeh-Dalmau et al., 2019; Harvell et al., 2019; McPherson et al., 2021). On this basis, the suggestion that EBUS may represent refugia from MHWs is accorded low confidence.

Despite low confidence in detailed projections for ecological changes in EBUS, the WGI assessment (WGI AR6 Chapter 9; Fox-Kemper et al., 2021) that upwelling-favourable winds will weaken (or be present for shorter durations) at low latitude but intensify at high latitude (high confidence), albeit by no more than 20% in either case (medium confidence), presents some key risks to associated EBUS ecosystems. These risks include potential decreases in provisioning services, including fisheries and marine aquaculture (Bertrand et al., 2018; Kifani et al., 2018; Lluch-Cota et al., 2018; van der Lingen and Hampton, 2018), and cultural services such as nature-based tourism (Section 3.5).

The polar seas cover ~20% of the global ocean and include the deep Arctic Ocean and surrounding shelf seas as well as the Southern Ocean south of the polar front. They play a significant role in ocean circulation and absorption of anthropogenic CO2 (Meredith et al., 2019). The Arctic is characterised by polar seas surrounded by land, while the Antarctic comprises continental Antarctica surrounded by the Southern Ocean. These high-latitude ecosystems share key properties, including strong seasonality in solar radiation and sea ice coverage. Sea ice regulates water-column physics, chemistry and biology, air–sea exchange and is a critical habitat for many species. In spring, when solar radiation returns and sea ice melts, intense phytoplankton blooms fuel food webs that include rich communities of both resident and summer-migrant species, with typically high dependency on a few key species for trophic transfer (Meredith et al., 2019; Rogers et al., 2020). Over the past two decades, Arctic Ocean surface temperature has increased in line with the global average, while there has been no uniform warming across the Antarctic (high confidence) (WGI AR6 Chapter 9; Fox-Kemper et al., 2021). Thus, the rate of change due to warming, and associated sea ice loss, is greater in the Arctic than in the Antarctic (high confidence) (Section 3.2; Table 3.14; WGI AR6 Chapter 9; Fox-Kemper et al., 2021). Both Arctic and Antarctic regions have a long history of living resource extraction, including some of the largest fisheries on the globe in terms of catches. However, only the Arctic hosts human populations, holding a rich Indigenous knowledge and local knowledge (IKLK) on these social–ecological systems (Cross-Chapter Paper 6; Meredith et al., 2019).

Previous assessments of polar seas (Table 3.14) concluded that climate change has already profoundly influenced polar ecosystems, through changing species distributions and abundances from primary producers to top predators, including both ecologically and economically important species (high confidence), and that it will continue to do so (Table 3.14).

Table 3.14 | Summary of previous IPCC assessments for polar seas

Since SROCC, evidence demonstrates that warmer oceans, less sea ice and increased advection results in increasing primary production in the Arctic, albeit with regional variation (high confidence), while trends remain spatially heterogeneous and less clear in the Antarctic (medium confidence) (Cross-Chapter Paper 6; Del Castillo et al., 2019; Lewis et al., 2020; Pinkerton et al., 2021; Song et al., 2021a). Furthermore, climate warming influences key mechanisms determining energy transfer between trophic levels including (a) altered size spectra, (b) shifts in trophic pathways, (c) phenological mismatches and (d) increased top-down trophic regulation (Table 3.15); however, the scale of impacts from changes in these mechanisms on ecosystem productivity in warming polar oceans remains unresolved and is hence assigned low confidence.

Table 3.15 | Examples of mechanisms influencing the transfer of energy between lower trophic levels in warmer polar oceans

Major community shifts, both gradual and abrupt, are observed in polar oceans in response to warming trends and MHWs (Arctic only) (high confidence) (Figure 3.12; Cross-Chapter Paper 6; Beaugrand et al., 2019; Meredith et al., 2019; Huntington et al., 2020). In general, abundances and ranges of Arctic fish species are declining and contracting, while ranges of boreal fish species are expanding, both geographically and in terms of feeding interactions and ecological roles (high confidence) (Huserbråten et al., 2019; Meredith et al., 2019; Huntington et al., 2020; Pecuchet et al., 2020a), with variable outcomes for large commercial fish stocks (Cross-Chapter Paper 6; Kjesbu et al., 2014; Holsman et al., 2018; Free et al., 2019). The extreme seasonal solar radiation cycles of these high latitudes may both act as a barrier for species immigration and change predator–prey dynamics in previously ice-covered areas, factors not currently considered in projections (limited evidence) (Kaartvedt and Titelman, 2018; Ljungström et al., 2021). Responses by marine mammals and birds to the ongoing changes in polar ecosystems are both positive and negative (Meredith et al., 2019; Bestley et al., 2020). Phenological, behavioural, physiological and distributional changes are observed in marine mammals and birds in response to altered ecological interactions and habitat degradation, especially to loss of sea ice (high confidence) (see Box 3.2; Cross-Chapter Paper 6; Beltran et al., 2019; Cusset et al., 2019; Descamps et al., 2019; Meredith et al., 2019; Huntington et al., 2020). Reproductive failures and declining abundances attributed to warmer polar oceans and less sea ice cover are observed in populations of polar bears, Ursus maritimus, seals, whales and marine birds (high confidence) (see Box 3.2; Duffy-Anderson et al., 2019; Ropert-Coudert et al., 2019; Bestley et al., 2020; Chambault et al., 2020; Molnár et al., 2020; Stenson et al., 2020). The ongoing changes in polar marine ecosystems can lead to temporary increases in biodiversity and functional diversity (e.g., due to immigration of boreal species in the Arctic, high confidence), but reduced trophic-transfer efficiencies and functional redundancy, with uncertain consequences for ecosystem resilience and vulnerability (limited evidence, low agreement ) (Griffith et al., 2019b; Alabia et al., 2020; du Pontavice et al., 2020; Alabia et al., 2021; Frainer et al., 2021).

Calcareous polar organisms are among the groups most sensitive to ocean acidification (high confidence) (Section 3.3.2). Niemi et al. (2021) reports that >80% of sampled sea snail, Limacina helicina, a key species in pelagic food webs, displayed signs of shell dissolution in the Amundsen Gulf. However, bacteria, phytoplankton, zooplankton and benthic communities are found to be detrimentally impacted, resilient or even positively influenced by ocean acidification in observational and experimental studies (Section 3.3; Hildebrandt et al., 2016; Thor et al., 2018; Ericson et al., 2019; McLaskey et al., 2019; Meredith et al., 2019; Petrou et al., 2019; Renaud et al., 2019; Brown et al., 2020; Hancock et al., 2020; Henley et al., 2020; Johnson and Hofmann, 2020; Torstensson et al., 2021). While fish larval stages may be sensitive, adult fish are expected to have low vulnerability to projected acidification levels (Section 3.3.3; Hancock et al., 2020), although reduced swimming capacity in polar cod in an ocean acidification experiment has been observed (Kunz et al., 2018). Polar organisms’ sensitivity to ocean acidification may increase with increasing light levels due to the loss of sea ice (algae; Donahue et al., 2019; Kvernvik et al., 2020), temperature stress (pteropods; Johnson and Hofmann, 2020) or indirectly via increased heterotrophic bacterial productivity (limited evidence) (Vaqué et al., 2019). Due to limited mechanistic understanding of observed effects, and mixed responses among Arctic species, future impacts of ocean acidification are assigned medium confidence for polar species and low confidence for outcomes for polar ecosystems (Meredith et al., 2019; Green et al., 2021b).

While levels of pollutants in biota (e.g., persistent organic pollutants, mercury) have generally declined over the past decades, recent increasing levels are associated with release from reservoirs in ice, snow and permafrost, and through changing food webs and pathways for trophic amplification (medium confidence) (see Box 3.2; Ma et al., 2016; Amélineau et al., 2019; Foster et al., 2019; Bourque et al., 2020; Kobusińska et al., 2020). Also, a warmer climate, altered ocean currents and increased human activities elevate the risk of invasive species in the Arctic (medium confidence), potentially changing ecosystems in this region (high confidence) (Chan et al., 2019; Goldsmit et al., 2020). In the remote Antarctic, there is a lower risk of invasive species (limited evidence) (McCarthy et al., 2019; Holland et al., 2021).

Fisheries are largely sustainably managed yet are expanding polewards following sea ice melt in the Arctic (high confidence) (Fauchald et al., 2021) and possibly in the Antarctic (limited evidence) (Santa Cruz et al., 2018). Tourism is increasing and expanding in both polar regions, while shipping and hydrocarbon exploration are growing in the Arctic, increasing the risks of compound effects on vulnerable and already stressed populations and ecosystems (high confidence) (Sections 3.6.3.1.3, 3.6.3.1.4; Cross-Chapter Paper 6; Hauser et al., 2018; Meredith et al., 2019; Helle et al., 2020; Rogers et al., 2020; Cavanagh et al., 2021).

Ensemble global model projections indicate future increases in primary production and total animal biomass towards 2100 under RCP2.6 (~5 and 50%, respectively) and RCP8.5 (~10 and 70%, respectively), in the Arctic (Bryndum-Buchholz et al., 2019; Lotze et al., 2019; Nakamura and Oka, 2019), highlighting opportunities for, and possibly conflicts over, new ecosystem services (Section 3.5). For the Southern Ocean, no overall trends are apparent, but greater variability in both primary production and total animal biomass are projected under RCP2.6, with an ~5 and 15% increase in primary production and total animal biomass under RCP8.5, respectively (Bryndum-Buchholz et al., 2019; Lotze et al., 2019; Nakamura and Oka, 2019). All projections presented exhibit high inter-model variability and hence uncertainty (Heneghan et al., 2021). Furthermore, regional models project significant distributional shifts and wide-ranging trends (i.e., relatively stable, increasing and declining) in productivity for key ecological and commercial species, and functional groups, with weak to strong dependence on emission scenarios, indicating low confidence in future outcomes for polar marine ecosystems and associated ecosystem services (Section 3.5; Piñones and Fedorov, 2016; Griffiths et al., 2017; Klein et al., 2018; Hansen et al., 2019; Meredith et al., 2019; Steiner et al., 2019; Tai et al., 2019; Alabia et al., 2020; Holsman et al., 2020; Reum et al., 2020; Veytia et al., 2020; Sandø et al., 2021). Potentially highly influential tipping points associated with Arctic sea ice melt and Antarctic ocean circulation change adds to this uncertainty (Cross-Chapter Paper 6; Heinze et al., 2021). Nevertheless, increasing evidence supports that sustainable and adaptive ecosystem-based fisheries practices can reduce detrimental impacts of climate change on harvested populations (medium confidence) (Section 3.6.3.1.2; Klein et al., 2018; Free et al., 2019; Hansen et al., 2019; Holsman et al., 2020).