and, instead, the combined estimate of Duarte (2017) is used in our summary table. We also do not provide estimates of carbon stocks in the sediments of non‐vegetated, mainly Lacustrine, wetland subclasses that include rivers, lakes, and small ponds. Fig. 1.1 shows the global distribution of wetlands based on a combined remote sensing and inventory based integration, with key wetland complexes visible, such as the Hudson Bay Lowlands, the Western Siberian Lowlands, the Cuvette Central, Sudd wetlands, Okavango Delta in Africa, and the Pantanal wetlands and Amazonian lowlands in South America.
1.1.3. Overview of Chapter
The chapter is organized first by introducing methodologies used to estimate above‐ and belowground carbon stocks, describing field methods, remote sensing, and ecological modeling approaches. A combination of these three methods is used in most regional‐ to global‐scale accounting of carbon stocks, where field data constrain models that are used to interpret remote sensing observations of vegetation indices or soil moisture. The second section of the chapter provides a review of carbon stock estimates for boreal wetlands (permafrost and peatlands separately, and mineral soil wetlands), tropical peatlands, temperate wetlands (peatlands and mineral‐soil systems), and for coastal ecosystems (mangroves and tidal marshes). The last section of the chapter describes how land‐use change has affected wetlands, through drainage, degradation, and peat harvest, and more recently the conversion of tropical wetlands to oil palm plantations, and then presents how climate change is expected to affect wetland carbon stocks through increases in air temperature but also via changes in precipitation regimes.
Figure 1.1 Global maximum extent of vegetated wetland area using the Wetland Area Dataset for Methane Modeling (WAD2M, based on Zhang et al., 2020), which is the basis of the wetland methane budget for Saunois et al. (2020). The dataset combines surface inundation data from the Surface Water Microwave Product Series V3.2 (SWAMPS) with inventories of tropical wetlands (Gumbricht et al., 2017), temperate wetlands (Lehner and Doll, 2004) and high‐latitude wetlands (Hugelius et al., 2014). Inland waters are removed using the Landsat permanent water bodies dataset of Pekel et al. (2016) and rice cultivated areas removed using MIRCA2000.
1.2. PAST CHANGES IN WETLAND CARBON STOCKS
1.2.1. Holocene Timescale
The quantity of carbon stored in wetlands fluctuates over millennia due to climate, glacial retreat, and, more recently, from human activities that include peat extraction or drainage. Simulations of wetland extent at the Last Glacial Maximum (LGM) show wetlands were more expansive than at present, but these areal estimates remain uncertain (Kaplan, 2002; Kaplan et al., 2006). For example, larger areas of Amazonian wetlands during the mid‐Holocene have been invoked as drivers of CH4 flux to explain atmospheric CH4 over this period (Singarayer et al., 2011). The fate of carbon in coastal wetlands submerged by the simultaneous sea level rise is less understood.
Subsequently, Holocene expansion of boreal peatlands in previously glaciated areas has sequestered significant amounts of carbon. Currently, it is thought that the catotelm in the peatlands north of 40°N alone could have accumulated 330 PgC (240–490 PgC) over the past the past 8000 years (Kleinen et al., 2012). Globally, carbon stocks in peatlands estimated from peat cores is 103 ±9 PgC and 145 ±13 PgC for the periods 11–9 kyBP and 9–7 kyBP, respectively, while earth system models estimated stocks of 54 PgC and 76 PgC for these two time periods (Stocker et al., 2014).
Historic Time Period
Carbon storage in wetlands has declined due to anthropogenic land use and land cover change, primarily from conversion to cropland, forestry, urban area, and peat extraction over the past millennia and centuries (Joosten and Clarke, 2002; Asselen et al., 2013). Artificial soil drainage also exposes soil organic carbon, accumulated over millennia, to aerobic oxidation, leading to large carbon fluxes to the atmosphere (Erb et al., 2017; Armentano, 1980). The global area of drained wetlands is estimated to be as high as 71% since 1700 (Davidson, 2014), and 35% since 1970 according to recent meta‐analyses (Dixon et al., 2016; Darrah et al., 2019), while mapping approaches estimate cumulative wetland losses to be 33% (Hu et al., 2017). The uncertainty in wetland area loss presents a challenge to estimating losses in soil carbon storage.
Global inventories of land‐use related emissions have not considered the impact of wetland drainage outside of recent drainage in Southeast Asia (Pongratz et al., 2018). Drainage of peat swamps in Indonesia alone are estimated to have emitted 6 PgC from 1850–2015 (Houghton and Nassikas, 2017). Peat drainage in this region still occurs at a rapid pace, i.e., 14,500 km2 of peat swamp forest have been converted to oil palm and pulpwood plantations between 2000–2010 (Page and Hooijer, 2016). Separate accounting efforts using geospatial data and emission factors have estimated that >250,000 Mkm2 of organic soiled wetlands were drained for agriculture globally, leading to a CO2 release of 0.078 Pg/yr, more than one‐fourth of all land‐use CO2 emissions (Tubiello et al., 2016). Nearly 13% of these emissions have occurred since 1990 (Conchedda and Tubiello, 2020). A separate bookkeeping approach estimates peatland degradation and losses to 510,000 Mkm2 and a cumulative release of 80.8 PgC (Leifeld and Menichetti, 2018). Following drainage, carbon is likely also transported to the river network then to the ocean as dissolved carbon, though this pathway and emissions of carbon to the atmosphere is uncertain (Cole et al., 2007). The decline in global wetland area since 1850 is estimated to have reduced methane emissions by 56 Tg CH4/yr with most of the decline from the northern temperate zone (Paudel et al., 2016).
1.3. METHODOLOGIES
1.3.1. Field Sampling of Wetland Carbon Stocks
Monitoring wetland vegetation and carbon stocks remotely via satellites and airborne instruments is increasingly common, but field‐based monitoring remains fundamental to understanding and quantifying wetland characteristics. Several important biogeochemical pathways unique to wetlands form important links between wetland vegetation, water, and soil (Ardón et al., 2013; Herbert et al., 2015; Osland et al., 2016). Therefore, monitoring both vegetation and soil over time are requisite to more comprehensively understanding wetland responses to global change (Taillie et al., 2019).
Both the vegetation composition and ecosystem structure contribute to the ecological function and value of wetlands. Though wetlands are often characterized according to the dominant plant species (e.g., “pond pine pocosin”), threshold responses to stressors can make certain species indicators for given stressors (Dufrêne & Legendre, 1997). As such, inventory of canopy species, as well as herbaceous and woody groundcover, may be necessary to adequately describe wetlands, particularly in temperate regions (Bratton, 1976). Aside from species composition, variation in vegetation structure (e.g., height, density, heterogeneity) may be dramatically different within and among wetlands. In addition, some variation in vegetation structure, such as mid‐ and under‐story vegetation density, may be difficult to estimate via remote sensing and will be best quantified via field‐based inventory (Riegel et al., 2013). Given the effort required to survey plants within a plot, care should be taken to balance the number and size of experimental units that are appropriate for the research objectives and capturing landscape heterogeneity. In forested settings, 11‐m radius plots are often used (Henttonen & Kangas, 2015). Selecting in‐situ measurements to match the scale of observation of airborne and spaceborne remotely sensed data (e.g., canopy height) may allow for scaling up field‐collected measurements (Hudak et al., 2012; Riegel et al., 2013).
Because of the value of wetlands for carbon storage, researchers often aim to translate vegetation inventories to biomass or carbon stocks. Such calculations are often made with the use of allometric