Volume 40, Issue 20 p. 5508-5513
Regular Article
Free Access

Surface vegetation patterning controls carbon accumulation in peatlands

Julie Loisel

Corresponding Author

Julie Loisel

Department of Earth and Environmental Sciences, Lehigh University, Bethlehem, Pennsylvania, USA

Corresponding author: J. Loisel, Department of Earth and Environmental Sciences, Lehigh University, 1 West Packer Ave., Bethlehem, PA 18015-3001, USA. ([email protected])Search for more papers by this author
Zicheng Yu

Zicheng Yu

Department of Earth and Environmental Sciences, Lehigh University, Bethlehem, Pennsylvania, USA

Search for more papers by this author
First published: 16 July 2013
Citations: 47


[1] Localized ecohydrological feedbacks have been proposed as important internal drivers of patterns and processes in terrestrial ecosystems. However, there is a significant gap between the prominence of theoretical models and the paucity of observations describing linkages between ecosystem structure and functioning. Here we report empirical evidence for strong interactions between surface vegetation patterning and long-term carbon sequestration rates in four peat bogs in Patagonia on the basis of high-resolution stratigraphic analyses. Peatland development across the study region was characterized by repeated switches between wet- and dry-adapted plant communities, depicting changes between hollows (wet assemblages) and lawns (dry assemblages) that could represent past surface patterning. We found a site-specific relationship between the frequency of the wet-to-dry cycles and carbon accumulation rates that followed a power function, with rapid vegetation shifts causing high peat accumulation rates. Our results show strong evidence for internal regulation of vegetation dynamics and peatland growth, implying that understanding peatland processes is a prerequisite for peat-based paleoclimate reconstructions.

Key Points

  • Temporal changes in spatial patterning promote carbon sequestration in bogs
  • A power-law relationship exists between vegetation patterns and C accumulation
  • Internal feedbacks are important in regulating peat bog growth

1 Introduction

[2] A mosaic of microtopographic irregularities characterizes most peatland surfaces [Sjörs, 1961; Foster et al., 1983]. The spatial organization of these microforms is easily observable across the peatland surface due to their distinctive plant communities: hummocks and lawns are dominated by dry-adapted Sphagnum mosses and shrubs, and wet-adapted rushes and brown mosses colonize wet hollows (Figures 1a and 1b). The biogeochemical functioning of these microenvironments also varies strongly, with wet hollows releasing significantly more methane to the atmosphere and accumulating peat at slower rates than their drier counterparts [Waddington and Roulet, 1996]. Hummocks and lawns are generally stronger carbon dioxide emitters but also stronger carbon sinks than hollows [Laine et al., 2007]. Understanding the spatial and temporal dynamics of these surface structures is important because alterations to the proportion or spatial arrangement of these microforms may have important effects on peatland carbon fluxes, with direct implication for the long-term peatland carbon-sink function [Baird et al., 2009].

Details are in the caption following the image
Peatland surface patterning and peat stratigraphy in southern Patagonia. (a) Aerial view of Upper Andorra Valley peat bog showing striped pattern of lawns and hollows [Loisel and Yu, 2013]. (b) Ground photo of Cerro Negro peat bog showing Sphagnum lawns (red moss) and wet hollows bordered by the rush Tetroncium magellanicum (green herbaceous plant). (c) Peat core (240–280 cm) from Cerro Negro peat bog showing the alternating fossilized lawn (Sphagnum) and hollow (herbaceous) plant assemblages. Herbaceous peat is dark brown and highly decomposed, whereas Sphagnum peat is reddish brown and poorly decomposed. (d) Peatland cross section showing lenticular structures that could be indicative of regeneration cycles as suggested by Osvald [1923]. Continuous light- and dark-colored horizontal beds that may represent a threshold response to internal or external forcing are also present [Barber, 1981; Couwenberg, 2005].

[3] The physical characteristics of peatland microforms induce a series of ecohydrological feedbacks that allow for the development and persistence of these microforms within the ecosystem. For instance, water flows and associated nutrient transfers are both controlled by local variations in peat hydraulic conductivity and transmissivity, with hummocks acting as barriers to flow due to their low hydraulic conductivity [Foster et al., 1983]. These localized water flows control vegetation communities that, in turn, regulate the vertical rate of peat accumulation [Belyea and Baird, 2006]. In addition, the nonlinear relationship between the rate of peat accumulation and thickness of the unsaturated zone results in a differential rate of peat formation between hummocks, lawns, and hollows that amplifies over time and leads to sharply bounded microforms [Belyea and Clymo, 2001; Eppinga et al., 2009]. Furthermore, modeling studies have shown that these localized feedbacks might propagate to the ecosystem scale and could be the main driver for surface patterning in some peatlands [Rietkerk et al., 2004; Eppinga et al., 2009]. From this standpoint, it can be argued that spatial organization of hummocks and hollows in peatlands is primarily the consequence of nonlinear, self-regulated internal dynamics that are independent from climate. These feedbacks have profound consequences on long-term peatland development [Morris et al., 2011]. Peat profiles with sufficient temporal depth and resolution can provide a means to further document and understand the genesis and persistence of peatland surface microforms and help elucidate the importance of self-regulation mechanisms in these ecosystems.

[4] Our main objective is to discuss the relative roles of autogenic and allogenic processes in controlling peatland vegetation structure and ecosystem C sequestration. We present detailed plant macrofossil analyses combined with long-term carbon accumulation estimates for four peat bogs located in southern Patagonia (Table 1). Thick peat deposits and high soil carbon densities (~ 170 kg C m−2) characterize these southern peat bogs [Loisel and Yu, 2013]. These ecosystems also have a low vegetation biodiversity, with large lawns almost exclusively covered by a single moss species (Sphagnum magellanicum) and hollows dominated by the rush Tetroncium magellanicum and the brown moss Drepanocladus spp. (Figure 1b). The combination of rapid peat accumulation and simple vegetation communities make Patagonian peat bogs ideal candidates for studying spatial and temporal peatland dynamics. Overall, this study provides an explanation of the linkages between surface patterning dynamics and carbon accumulation in peat bogs on the basis of inferences drawn from empirical evidence.

Table 1. Study Site Informationa
Upper Andorra Valley Escondido Harberton Bog Cerro Negro
Latitude (°S) 54.75 54.62 54.87 52.07
Longitude (°W) 68.33 67.77 67.28 72.03
Elevation (m) 200 127 11 217
Mean annual temperature (°C) 4.2 4.8 3.4 6.4
Mean annual precipitation (mm) 542 469 507 651
Mean relative humidity (%) 77 77 77 70
Seasonality (°C) 8.3 9.0 8.5 9.7
Peatland type Sphagnum bog Sphagnum bog Sphagnum bog Sphagnum bog
Total peat depth (cm) 715 450 525 445
Peat inception age (ka) 10.823 6.929 1.866 9.205
Fen-to-bog transition depth (cm) 310 350 440 350
Fen-to-bog transition age (ka) 2.163 4.763 0.947 4.171
  • a Age (ka) is presented in thousands of calibrated years before present. Seasonality is defined as temperature difference between the warmest (January) and coldest months (July).

2 Study Sites and Methods

[5] All four study sites are raised bogs dominated by Sphagnum magellanicum. The central portion of these peatlands is flat and lies a few meters higher than the surrounding area. All sites are underlain by glacial or glaciolacustrine clay-rich deposits. Three of the four sites developed in valley bottoms, while the fourth site (Escondido) is found in a large depression of a Pleistocene morainic complex. Surface patterning is apparent in parts of these peatlands [Loisel and Yu, 2013]. Three of the four studied peatlands are located in the vicinities of Ushuaia, Argentina (54°47′S, 68°14′W). The climate in Ushuaia is cool temperate, with weak temperature seasonality (monthly means of 1.6°C in July and 10.3°C in January) and mean annual precipitation of 528 mm (Table 1). The fourth site (Cerro Negro) is located on the leeside of the Andes, about 150 km north of Punta Arenas in Chile (53°10′S, 70°56′W). The climate in Punta Arenas is similar to that in Ushuaia (Table 1).

[6] Peat cores were collected in January 2010 using a box corer (top 100 cm) and a 50 cm long Russian-type peat corer (below 100 cm). The cores were sealed in split PVC pipes in the field and transported to Lehigh University where they were stored at 4°C in a cold room. For each core, chronology was constrained using at least 10 radiocarbon (14C) dates determined by the accelerated mass spectrometry (AMS) method. Dating material consisted of nonaquatic plant macrofossil remains that were handpicked and cleaned with distilled water. Samples were submitted to Keck AMS Carbon Cycle Lab at University of California, Irvine. Results were calibrated using Bacon [Blaauw and Christen, 2011] on the basis of the INTCAL09 calibration data set [Reimer et al., 2009]. Postmodern 14C dates were calibrated using CALIBomb and a concatenation of the Levin's Vermunt and Schauinsland calibration data sets [Levin and Kromer, 2004]. Radiocarbon dating results and the age-depth relationships can be found in Loisel and Yu [2013].

[7] A total of 762 peat samples were analyzed for high-resolution plant macrofossil analysis following a modified version of Mauquoy and van Geel's [2007] method. Peat subsamples (2 cm3) were rinsed with distilled water through a 150 µm sieve. For each sample, Sphagnaceae (peat moss family), Amblystegiaceae (brown moss), herbaceous, ligneous, and unidentifiable organic matter plant materials were quantified as percentages of the total sample by volume. Spectral analysis was performed on Sphagnum percentages using the SSA-MTM toolkit, and the multitaper method confidence levels were calculated with respect to the robust red noise [Mann and Lees, 1996; Ghil, 2002]. Prior to the analysis, each Sphagnum percentage time series was resampled at an even age interval using AnalySeries [Paillard et al., 1996]. Only spectra above the 99% confidence interval were considered statistically significant. The uncertainty associated with each frequency was established by calculating the minimum and maximum period of each significant peak.

[8] Bulk density and organic matter content were determined along the peat cores at 1 cm increments following standard loss-on-ignition procedures. Contiguous peat subsamples (1 cm3) were dried overnight at 105°C, weighed, and burned at 550°C for 2 h. Peat-carbon content was assumed to be 50% of the organic matter density. Rates of peat-carbon accumulation (in g C m−2 yr−1) were calculated by multiplying the carbon mass of each depth increment (organic matter density × carbon content) by the interpolated deposition rate of each sample as inferred from the age-depth models [Loisel and Yu, 2013].

3 Results

[9] High-resolution plant macrofossil analysis clearly shows that regular oscillations between light- and dark-colored peat layers have occurred at each site throughout the bog developmental phase (Figure 1c). While the light layers were composed of well-preserved Sphagnum macrofossils, the dark layers were composed of a highly decomposed organic matter matrix where monocotyledon plant remnants (likely the rush Tetroncium magellanicum) and Drepanocladus spp. were identified (Figure 2 and Figure S1 in the supporting information). These alternating macrofossil assemblages along the peat core could represent distinct switches between hummocks/lawns (light peat) and hollows (dark peat), as observed across the surface microtopographic gradient at present (Figures 1a and 1b). Testate amoebae analysis supports this interpretation, as wet-adapted species such as Amphitrema wrightianum and Trigonopyxis microstoma dominated the assemblages within the dark and humified peat layers, and intermediate species including Heleopera sphagni and Nebela martiali were identified within the Sphagnum-rich, light-colored layers (Figure S2 in the supporting information).

Details are in the caption following the image
Wet-dry cycles recorded in peatland stratigraphy. (a–d) Simplified plant macrofossil diagrams for Cerro Negro, Escondido, Upper Andorra Valley, and Harberton Bog, respectively. Age (ka) is presented in thousands of calibrated years before present. Sphagnum percentages are presented as “dry indicators” (red areas); the summed brown moss, monocots, and unidentifiable organic matter percentages are presented as “wet indicators” (blue areas). More detailed plant macrofossil diagrams are presented in Figure S1 in the supporting information. (e–h) Power spectrum analysis of Sphagnum percentages for Cerro Negro, Escondido, Upper Andorra Valley, and Harberton Bog, respectively. Orange and green lines represent the 95% and 99% confidence intervals, respectively. Significant spectral peaks and periods are also indicated.

[10] Spectral analysis indicates that wet-dry plant community shifts followed site-specific periodicities that ranged from 472 ± 103 years at Cerro Negro to 35 ± 5 years at Harberton Bog (Figure 2). In the case of Upper Andorra Valley (Figure 2g), we observed a spectral peak above the 99 % confidence interval at 136 ± 16 years and a second one above the 95% confidence interval at 358 ± 150 years. The 136 year cycle was visible by inspection of the data series but the 358 year cycle was not, and was therefore not considered further in the interpretation.

[11] Mean time-weighted carbon accumulation rates were also site specific and varied from 19 ± 8.7 g C m−2 yr−1 at Cerro Negro to 92 ± 6.3 g C m−2 yr−1 at Harberton. Remarkably, peat-carbon accumulation rates were positively correlated with the frequency of wet-to-dry vegetation switches following a power function (R2 = 0.93, p < 0.0001), such that rapid plant vegetation shifts were associated with high peat accumulation rates (Figure 3). These site-specific associations suggest that local-scale conditions and autogenic processes have exerted major controls on surface vegetation dynamics and peat accumulation over centennial and millennial timescales in these four systems.

Details are in the caption following the image
Carbon sequestration controlled by surface vegetation patterning dynamics. Mean time-weighted carbon accumulation rate in relation to periodicities of wet-dry cycles from four peat cores in southern Patagonia. Error bars represent standard errors associated with peat-carbon accumulation rates and minimum and maximum periods of each significant peak. Power function: y = 13,035x−1.204 (p < 0.0001).

4 Discussion

[12] Localized ecohydrological feedbacks have been proposed as important internal drivers of patterns and processes in terrestrial ecosystems [Scheffer et al., 2001; Belyea and Baird, 2006; Eppinga et al., 2009]. In the present study, the transitions between wet- and dry-adapted plant communities can be explained by the internally driven differential rate of peat formation between lawn-forming Sphagnum (rapid) and hollow-colonizing Tetroncium (slow) that is the result of a positive feedback between acrotelm thickness and plant production [Belyea and Clymo, 2001]. For example, when Sphagnum growth and associated peat formation in lawns occur faster than water is stored, Sphagnum plants grow laterally and colonize hollows (Figure 1b) [Osvald, 1923]. Conversely, when Sphagnum peat accumulates slower than water storage in lawns, the resulting thinner unsaturated zone favors the establishment and growth of wet-adapted species, which results in a shift from lawn to hollow. These localized positive feedbacks between water storage, vegetation communities, and the rate of peat formation result in a mosaic of surface patterns at the ecosystem scale [Belyea and Malmer, 2004; Rietkerk et al., 2004].

[13] There have been debates over the past century as to whether the surface features of bogs can be explained by internally controlled cycling regeneration [Osvald, 1923] or by climatically controlled wetter and drier phases [Barber, 1981]. Under Osvald's [1923] regeneration cycle theory, a differential rate of peat formation between hummocks and hollows is assumed, which leads to stratigraphic changes that are self-regulated. Alternatively, Barber's [1981] phasic theory suggests that wet hollows expand across the peatland surface at the expense of dry hummocks during wetter episodes, and vice versa. We will briefly discuss our results in light of these alternative theories.

[14] The phasic theory is inconsistent with our observations because the wet and dry episodes were found to oscilate at different periodicities amongst sites (Figures 2 and 3). In addition, it is unlikely that climate change alone has caused the inferred plant community shifts, as three of the four sites are located within 50 km of one another and are most likely influenced by similar or same climatic conditions (Table 1). Finally, these wet and dry cycles could not be corroborated with published paleoclimatic reconstructions for this region [e.g., Huber and Markgraf, 2003; Mauquoy et al., 2004], suggesting that these cyclic vegetation changes are the result of site-specific ecological dynamics. The regeneration cycle theory is also partly inconsistent with our observations. The original hypothesis states that “degenerating old hummocks” can be identified along the peat cores in the form of highly decomposed Sphagnum remains preceding each hollow stage. But such features were not found along our records. Peatland cross sections in southern Patagonia do show series of lenticular structures that would be indicative of regeneration cycles (Figure 1d). Interestingly, the same cross section also shows continuous light- and dark-colored horizontal beds. The latter have been reported in the literature previously and have been linked to (1) climate change [Barber, 1981], (2) hummock migration across the peatland surface [Couwenberg, 2005; Kettridge et al., 2012], or (3) threshold response to internal or external forcing [Clymo, 1991]. Dating along these horizontal beds could be used to test the second hypothesis. If the ages of these beds get progressively younger towards the outer edges of the peatland, then Couwenberg's [2005] hypothesis would be supported.

[15] Overall, our findings lend support to internal regulation of surface vegetation patterning and peat-carbon-sequestration rates. As previously mentioned, we speculate that feedbacks between peat growth, peatland hydrology, and vegetation communities are responsible for the observed changes in surface patterning and carbon sequestration. It is possible that peatland surface topography and gradient exerts some control on these processes as well [e.g., Couwenberg, 2005; Kettridge et al., 2012], potentially by triggering threshold responses in bog surface vegetation [Clymo, 1991]. However, we stress that climate most certainly influences peat growth and carbon sequestration over large temporal and spatial scales [e.g., Clymo et al., 1998; Yu et al., 2010]. Likewise, surface patterning may be initiated or amplified by climate change [e.g., Walker and Walker, 1961; Karofeld, 1998]. We must therefore conclude that no single hypothesis is likely to explain all these observations but that caution must be applied when interpreting the peat stratigraphy as an archive of climate change.

5 Conclusions and Implication

[16] Our observations provide strong empirical evidence for linking surface vegetation patterning dynamics with carbon sequestration in peat bogs. We show that the carbon-sequestration capacity of peatlands is linked to ecosystem structuring dynamics over centennial to millennial timescales, with rapid changes between wet- and dry-adapted plant species resulting in high carbon-sequestration rates. This relationship can be explained by a simple self-reinforcing behavior where the presence of Sphagnum lawns promotes rapid peat formation, accelerating localized bifurcations between new lawns (hollow colonization) and new hollows (lawn colonization), which further promote differential rates of peat formation and bifurcation [Osvald, 1923]. While the importance of the pattern structure (proportion, arrangement, and spatial organization of microforms) is thought to be of major importance for estimating peatland carbon balance [Baird et al., 2009; Kettridge et al., 2012], our results indicate that their temporal dynamics, i.e., the speed at which surface patterning transits over the peatland surface, are also very important for understanding the peatland carbon-sink function. These findings imply that, in systems characterized by a mosaic of alternative stable states, localized short-term instability may lead to long-term maintenance of ecosystem functioning such as carbon sequestration at the ecosystem scale [Levin, 2000]. Our results also show that internal peatland processes such as those responsible for surface patterning can disconnect, or at least mediate, ecosystem behavior from external climatic influence, which may lead to ecosystem behavior that is site specific rather than climate sensitive [Morris et al., 2011; Swindles et al., 2012]. These findings have implications for peat-based paleoclimatic reconstructions, as they challenge the assumption that peatland development is primarily sensitive to external forcing mechanisms. As both external and internal processes interact on similar timescales, they are particularly difficult to tease apart along the peat record. Therefore, how peat bogs respond to, and record, paleoenvironmental changes still remains a matter of debate. Overall, a better understanding of peatland accumulation processes is essential for robust peat-based paleoclimate reconstructions.


[17] We thank Daniel Brosseau and Greg Sills for their assistance in the field and laboratory; Daniel Minguez for his assistance with spectral analysis; Robert Booth, Nigel Roulet, Frank Pazzaglia, and two anonymous reviewers for their valuable comments on an earlier version of the manuscript; and Liliana Kusanovic Marusic from Estancia Cerro Negro near Rio Rubens (Chile) and R. Natalie P. Goodall from Estancia Harberton near Ushuaia (Argentina) for access to their properties. The research was funded by a Faculty Innovation Grant from Lehigh University (2009), a US-NSF Doctoral Dissertation Improvement Grant (DEB-1110665), and an NSERC Canada Postgraduate Scholarship (BESC-D3-362645-2008).

[18] The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.