Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 Here we present a new data synthesis of global peatland ages, area changes, and carbon (C) pool changes since the Last Glacial Maximum, along with a new peatland map and total C pool estimates. The data show different controls of peatland expansion and C accumulation in different regions. We estimate that northern peatlands have accumulated 547 (473–621) GtC, showing maximum accumulation in the early Holocene in response to high summer insolation and strong summer – winter climate seasonality. Tropical peatlands have accumulated 50 (44–55) GtC, with rapid rates about 8000–4000 years ago affected by a high and more stable sea level, a strong summer monsoon, and before the intensification of El Niño. Southern peatlands, mostly in Patagonia, South America, have accumulated 15 (13–18) GtC, with rapid accumulation during the Antarctic Thermal Maximum in the late glacial, and during the mid-Holocene thermal maximum. This is the first comparison of peatland dynamics among these global regions. Our analysis shows that a diversity of drivers at different times have significantly impacted the global C cycle, through the contribution of peatlands to atmospheric CH4 budgets and the history of peatland CO2 exchange with the atmosphere.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Peatlands worldwide, in particular northern (boreal and subarctic) peatlands, have been shown to be important players in the global carbon (C) cycle in the recent past. Their possible trajectories in a changing climate have become a focus of global C cycle research, and a number of modeling groups have started to incorporate peatlands into global models [Frolking et al., 2009; Kleinen et al., 2010]. However, we still lack a fundamental understanding of broad-scale controls over peatland expansion and C accumulation in different regions. Also, basic estimates of the size of the peat C pool are variable. For example, for northern peatlands, by far the best-studied region, estimates range from 270 to 450 GtC [Gorham, 1991; Clymo et al., 1998; Turunen et al., 2002]. Further, although progress has been made for wetlands in general [Matthews and Fung, 1987] and for northern peatlands [MacDonald et al., 2006], an ecosystem-based digital map of global peatland regions is still lacking at a scale useful for modeling and synthesis. The need for synthesized data at a global scale, and the need for a better understanding of processes and controls, currently limit efforts to incorporate peatlands into global models to better constrain potential carbon-cycle – climate interactions and feedbacks [Joos et al., 2004; Friedlingstein et al., 2006].
 Here we present the first results of our synthesis of global peatland inception age and C accumulation data, and we discuss the broad-scale controls of peatland dynamics in different regions since the Last Glacial Maximum (LGM). The objectives of this paper are (1) to present a new global-scale map of major peatland regions; (2) to present the first global peat-based database of peatland initiation and area change since the LGM; (3) to document C accumulation variations and associated broad-scale controls in different regions (northern, tropical and southern); and (4) to discuss the roles and implications of peatlands for the global C cycle. The presented map and database will provide a valuable foundation for global C cycle modeling and synthesis activities.
2. Data Sources and Data Analysis
 Data sources for the peatland map (Figure 1) were based on the most up-to-date information available from individual countries or regions in major peatland regions of the world (see Table S1 of Text S1 in the auxiliary material). Some of these peatland data sets are available in shapefile or raster digital formats, including those from Canada, Tasmania and part of Russia. For other regions, we mapped peatlands either as histosols and/or gleysols layers as in the Harmonized World Soil Database V1.1 or from digitized paper sources. The peatland areas we used in the peat C pool calculations were derived from the literature (see Table 1 and the auxiliary material), rather than directly from the new peatland map presented. This is necessary because the peatland map shows peatland-abundant regions where peatlands cover at least 5% of the landmass, but accurate true peatland coverage and distribution is not available for many mapped regions.
Table 1. Summary Results From Northern, Tropical and Southern Peatlandsa
C Pool (GtC) (Range)
Holocene C Rate (gC m−2 yr−1)
References for peatland area data are in Text S1, available online.
Southern peatlands (Patagonia)
 The radiocarbon-dated (14C) ages of basal peat, indicating the onset of peat-accumulating conditions, were taken from original published sources for tropical peatlands and southern peatlands (see Table S2 of Text S1) and from a previous synthesis for northern peatlands (Figure 2) [MacDonald et al., 2006]. All 14C dates were calibrated to calendar years with the IntCal04 dataset [Reimer et al., 2004]. The frequency histograms were constructed by adding the number of dates within 2-σ range (95% probability) of calibrated ages at 10-year intervals. The frequencies were then added to calculate cumulative percentages. We interpret the frequency of basal dates as reflecting changes in peatland area over time, under the assumption that individual peatlands expanded linearly in their area since their initial formation. Detailed lateral expansion studies of individual peatlands, and especially a better process-level understanding of peatland expansion, would help us refine the temporal patterns of regional peatland area increase [Korhola et al., 2010]. For example, the higher rates of peatland initiation and peat carbon accumulation we observe during the early Holocene [Yu et al., 2009] would likely have also caused much higher rate of peatland area expansion of individual peatlands at that time, but it is clear that more basin-scale studies are needed.
 To estimate regional averages of apparent C accumulation rates, we calculated time-weighted rates for each available site in 1000-year bins, either from raw data including multiple calibrated ages and bulk density measurements/estimates or from published C accumulation rates. We then averaged the rates in each 1000-year bin for each region (northern, tropical and southern peatlands). These reconstructed rates of peat C accumulation from peat cores are apparent rates in that they have been affected by total deep C decomposition since peat formation, often spanning thousands of years, and therefore underestimate by some degree the true rate of past C uptake (but see Z. C. Yu (Holocene carbon flux histories of the world's peatlands: Global carbon-cycle implications, submitted to Holocene, 2010)).
 Changes in peat C pools at 1000-year intervals in different regions were calculated as the product of the peatland area at that time, as inferred from cumulative basal age frequency and peatland total area at the present, and C accumulation rates for each 1000-year bin. These C-pool intervals were added to generate cumulative C pools. The C pool ranges were estimated on the basis of standard errors of the mean C accumulation rates, which represent a minimum estimate of error, as other uncertainties in basal ages, possible non-linear peatland expansion, and peatland areas are not included.
 The peatland initiation patterns for northern peatlands show a peak in the early Holocene around 11-9 ka (1 ka = 1000 cal year BP) (Figure 2b) [MacDonald et al., 2006] (n = 1516). The C accumulation rates in many northern peatlands also show a peak in the early Holocene (Figure 2c), with a rate of about 25 g C m−2 yr−1. The overall time-weighted average rate is 18.6 g C m−2 yr−1 during the Holocene based on 33 sites from northern peatlands [Yu et al., 2009].
 Tropical peatlands started to form before 20 ka, much earlier than northern peatlands. A peak in tropical peatland initiation occurred at 8-4 ka, especially in Southeast Asia, but with a gap at 5.5 ka (Figure 2d), and then initiation frequency decreased until the present. The peat C accumulation rates were based on 26 sites from tropical peatlands in SE Asia, South America and Africa (11 span the entire Holocene), and among these 9 sites were from Southeast Asia (3 span the entire Holocene). The C accumulation rates were very low before 13 ka but increased gradually from 13 to 5 ka, followed by an accumulation peak at 5-4 ka especially for all tropical peatlands. The C accumulation rates then increased abruptly over the last 2000 years to > 20–50 g C m−2 yr−1. The large fluctuations and large spread in average C accumulation rates from Southeast Asia are partly due to the low number of sites. The overall average rate is 12.8 g C m−2 yr−1 during the Holocene based on 26 sites used from tropical peatlands in Southeast Asia, South America and Africa.
 From available data in the published literature, southern peatlands mostly initiated before 10 ka, with initiation peaks around 17–14.5 ka and 13.5 ka. The C accumulation rates were available from 17 sites (13 span the entire Holocene) in southern South America (Patagonia). The C accumulation rates were highest around 16 ka, although this was based on only 2 sites. During the Holocene C accumulation rates show a gradual increase from ∼15 to about 28–40 g C m−2 yr−1. The overall average rate is 22.0 g C m−2 yr−1 during the Holocene based on 20 sites from all southern peatlands (Patagonia, New Zealand, and sub-Antarctic islands).
 The total area of northern peatlands has been estimated from 3.88 to 4.09 × 106 km2 by Maltby and Immirzi . We used 4 × 106 km2 in our calculations of peatland area and peat C pool changes in northern peatlands. Tropical peatland area is estimated to be 368,501 km2, with 68% (250,580 km2) in Southeast Asia, and peatlands in Patagonia are estimated to be 45,000 km2 (see auxiliary material data). The rate of peatland area change varies over three orders of magnitude for northern, tropical and southern peatlands (Figure 3b). Our newly estimated peat C pools, based on area change over time from basal age frequencies and average C accumulation rates from multiple dated sites, are 547 GtC (ranging from 473 to 621 GtC) for northern peatlands, 50 (44–55) GtC for tropical peatlands, and 15 (13–18) GtC for southern peatlands (Figure 3c and Table 1).
4.1. Patterns and Controls of Global Peatland Dynamics since the LGM
 The highest rates of peatland expansion and C accumulation in the early Holocene in northern peatlands (Figures 2b and 2c) [Yu et al., 2009] are most likely caused by the maximum summer insolation and the greatest seasonality in insolation and climate at that time (Figure 2a). This temporal pattern is evident in West Siberia (Figure 2b) [Smith et al., 2004], the largest peatland region in the world, and also in Alaska [Jones and Yu, 2010]; in both regions the early Holocene corresponds with the Holocene thermal maximum (HTM). As with the record from Alaska [Jones and Yu, 2010], we suggest that the peak in C accumulation is controlled by increased plant production during warm summers and reduced peat C respiration during cold winters. The general decrease in peatland expansion and C accumulation in the mid- and late Holocene are likely in response to neoglacial climate cooling after the HTM [Kaufman et al., 2004; MacDonald et al., 2000], although the effect of cooling on northern peatland expansion may be more complex [Korhola et al., 2010] owing to land availability in peatland basins and regional moisture conditions.
 Carbon accumulation dynamics in tropical peatlands were likely affected by summer monsoon intensity, sea-level change and El Niño intensity. The observed peak in tropical peatland expansion in the mid-Holocene at 8-4 ka, especially in Southeast Asia at 7-4 ka (Figure 2d), is in response to the high and stabilized sea level and the resultant monsoon maximum [Griffiths et al., 2009]. In particular, many peatswamps in Southeast Asia developed in coastal regions (Figure 1) on mangrove substrates after sea-level stabilization. The slowdown in peatland expansion and in C accumulation rates after 4 ka may have been affected by intensified El Niño in the latter half of the Holocene [Moy et al., 2002]. Earlier initiation of tropical peatlands before 10 ka in Southeast Asia was mostly in inland peatlands at higher elevations [e.g., Page et al., 2004]. The highest C accumulation rates in the last 2 ka may be an artifact of the limited decomposition of recently deposited peat in undisturbed sites.
 The rapid expansion and C accumulation of peatlands in the Southern Hemisphere at 17-14 ka (Figures 2f and 2g) were likely induced by a warm climate during the Antarctic Thermal Maximum in the South Ocean region [Barker et al., 2009]. This warm interval, even warmer than the early Holocene, was caused by the bipolar seesaw effect resulting from the slowdown in the Atlantic meridional ocean circulation in the North Atlantic Ocean [Barker et al., 2009]. Also, the climate was wet as documented by speleothem paleoclimate records in South America [Wang et al., 2006]. It is interesting to note that the gap in peatland initiation in Patagonia and in southern peatlands at 13 ka corresponds to the Antarctic Cold Reversal [Barker et al., 2009]. The recurrence of peatland expansion and slight increase in C accumulation of existing peatlands around 4 ka might have been in response to the HTM in the Southern Hemisphere and/or to a change in the average position and the seasonal migration patterns of the Southern Hemisphere westerlies [Pendall et al., 2001].
4.2. Roles of Peatlands in the Global Carbon Cycle
 Our estimate of peat C pools of 547 GtC in northern peatlands using a new approach based on peatland initiation ages and time-variant C accumulation rates is larger than previous estimates that are based on peatland area, mean peat depth and mean bulk density (270 GtC [Turunen et al., 2002]; 450 GtC [Gorham, 1991]). Our estimate is conservative in that peatland area increases were potentially much more rapid than at the assumed linear expansion rate when the C accumulation was highest in the early Holocene (Figures 2b and 2c). Our peat C pool estimate of 50 GtC for tropical peatlands is lower than the previous estimate of 65–70 GtC [Page et al., 2004]. One possible reason is that the previous estimates were likely biased toward the much higher C accumulation during the last 2000 years. Our synthesis shows that the mean rate of apparent C accumulation for tropical peatlands is relatively low during portions of the Holocene (Figure 2e), likely reflecting high decomposition as well as high primary production in tropical climates. To our knowledge, our estimate of peat C pool of 15 GtC is the first for southern peatlands (mostly Patagonia).
 During the Holocene global peatlands have contributed significantly to atmospheric CH4 budgets. In particular, peatland area change should be a primary index for peatland CH4 emission potential under constant climate conditions and assuming consistent vegetation and hydrology. Our global peat data synthesis suggests that northern peatlands played a dominant role in the early Holocene CH4 budget, but tropical peatlands likely became more important in the mid-Holocene around 8-4 ka (Figure 3b). These findings are in agreement with the inter-polar gradient of CH4 concentrations in Greenland and Antarctica (Figure 3a) [Brook et al., 2000; Chappellaz et al., 1997]. However, the inter-polar gradient also indicates an increasing tropical CH4 contribution sometime after 4 ka (Figure 3a). As our tropical peatland data suggest a slowdown after 4 ka in the rate of expansion of natural peatlands, and presumably of non-peat forming wetlands and lakes under weakened monsoons and dry climates, our synthesis supports the proposal of Ruddiman  that the late-Holocene CH4 rise was strongly contributed to by human activities.
 The large total C pool of >600 GtC that we estimate for these global peatlands is large enough to have had significant impact on the global C budget and atmospheric CO2 change (Figure 3d) [Joos et al., 2004; Elsig et al., 2009]. However, the observed cumulative C pools as shown in Figure 3c do not allow us to make direct assessments of peat C burden on CO2 concentrations. A further analysis of the peat C data would need to partition the time-dependent net C uptake and C release terms in order to assess the role of global peatlands in the global C cycle since the LGM, especially during the Holocene (Yu, submitted manuscript, 2010). Although global peatland C sink intensity has varied greatly over time, our analysis here shows that peatlands have accumulated >600 GtC over the Holocene, serving as a long-term persistent C sink of >5 GtC per century on average.
 We thank Andy Baird and another reviewer for useful comments. This research was supported by National Science Foundation grant ATM 0628455.