The Stoichiometry of Carbon, Hydrogen, and Oxygen in Peat

Carbon (C), hydrogen (H), and oxygen (O) form ~90% by mass of peat, a product of the input of plant tissues and litter and the output of decomposition under aerobic and anaerobic conditions. We examined patterns of these elements, as the O:C and H:C atomic ratios, in over 1,300 peat samples collected from over 400 profiles in Ontario, Canada, representing bogs, fens, and swamps. The overall O:C ratio decreased from the surface (0.6 to 0.7) to ~0.5 at a depth of 50 cm and showed little further change to a depth of 5 m. In contrast, the H:C ratio decreased only slightly (1.30 to 1.25) over the top 1 m and showed no further significant decline with depth. The C oxidative state (Cox) and oxidation ratio showed strong decreases and increases, respectively, with depth with most changes occurring in the top 0.5 m. The O:C ratio, and Cox and oxidation ratio values were significantly correlated with the von Post humification index, with most changes occurring in index values 1 through 4, the latter representing a slight degree of decomposition. Collation of the Ontario peats with other data sets revealed the very large range in O:C and H:C values, with a general decrease from temperate to tropical and subtropical peatlands. Estimation of the O:C and H:C ratios of input (litter) and output (mineralization to CO2, methanogenesis to CH4 and CO2, and loss as dissolved organic C) allowed an estimation of the degree of decomposition or C loss.


Introduction
Peat soils contain very large amounts of organic carbon (C), generally between 30 and 200 kg C m À2 , and it is estimated that northern peatland soils contain over 400 Gt globally (Loisel et al., 2014). This accumulation arises from the rates of plant production being faster than the rates of litter and soil organic matter decomposition, associated with a high water table and the development of anoxic conditions, as well as peat plants, such as Sphagnum moss, having a slow intrinsic rate of decomposition (Rydin & Jeglum, 2013). Organic C generally forms 45-50% of the mass of boreal peat, and another~35% is comprised of oxygen (O) and~5% hydrogen (H), totaling 85-90%, with smaller amounts of nutrients and metals.
During decomposition, organic molecules are broken down by microbes, resulting in the loss of decomposition products such as carbon dioxide (CO 2 ), methane (CH 4 ), and dissolved organic C (DOC). Patterns of organic matter decomposition in peat have been approached in many ways, ranging from the simple von Post humification index, where higher values indicate greater decomposition (Table S1 in the supporting information), through ash content and C:N ratio, to nuclear magnetic resonance spectroscopy, Fourier transform infrared spectroscopy, and isotopes of C and N (e.g., Biester et al., 2014;Tfaily et al., 2014). Given the differences in C, H, and O stoichiometry between the input of organic matter to peatlands in plant tissues and litter and that of the decomposition products, one could expect changes in the C, H, and O stoichiometry of peat as it undergoes decomposition. The H:C and O:C ratios of the decomposition products are larger than most of the inputs, so there should be a decrease in the H:C and O:C ratios of the peat as it decomposes. Peat is the precursor to coal deposits, which are often characterized by the atomic ratios of O to C, and H to C, commonly illustrated as a van Krevelen (1961) diagram. These diagrams (e.g., Large & Marshall, 2015) show decreases in both O:C and H:C ratios as material passes from vegetation through soil organic matter and peat into coal, though not much attention has been given to variations within peat types. These ratios have also been used to characterize the chemistry of DOC (e.g., D'Andrilli et al., 2010;Kim et al., 2003). Worrall and colleagues (e.g., Clay & Worrall, 2015;Worrall et al., 2016) have used the concentrations of C, H, O, and N to assess the oxidation state of a variety of soils, including peats, as part of a global stoichiometric approach to C cycling and sinks (e.g., Masiello et al., 2008;Worrall et al., 2013). Two metrics have been applied to soil organic matter, the C oxidation state (C ox ), which is based on the relative molar concentrations of C, H, O, N, and S, and the oxidative ratio (OR), which is a derivative of C ox and C and N concentrations (Masiello et al., 2008). Clay and Worrall (2015) concluded that, for a range of samples collected from the upper 1 m of peat soils in the UK, the average C ox and OR values were À0.33 and 1.10, respectively, compared to À0.05 and 1.03 for vegetation. They then applied these metrics to assess the oxidation state of a blanket peatland within the Moor House National Nature Reserve in the UK (Worrall et al., 2016).
Data on C, H, and O concentrations in peat soils are sparse, but a detailed survey of peatlands in Ontario, Canada, by the Ontario Geological Survey (Riley, 1994a(Riley, , 1994bRiley & Michaud, 1989) resulted in the collection of cores from over 400 sites, representing bogs, fens, and swamps, and the analysis of over 1,300 samples for a variety of elements and properties. In this study, we start with these data to address three objectives: 1. We examine how O:C and H:C ratios and C ox and OR vary with peatland type, depth, and von Post humification index. 2. Combining these data with C, H, and O analyses of peat from temperate (UK, Latvia, northeastern United States) and subtropical and tropical (southeastern United States and Indonesia) regions, we compare how O:C and H:C ratios change, in relation to the input from vegetation/litter and the output from decomposition processes. 3. Assuming decomposition processes introduce changes in the O:C and H:C ratios of peat, we examine whether these stoichiometries can be used to estimate degree of decomposition of the peat, based on what would be the input from vegetation and litter and the stoichiometry of decomposition.

Materials and Methods
The sampling sites in Ontario were part of a large-scale survey of peat resources extending from 74°W to 94°W and 43°N to 51°N in northwestern, northeastern, and southeastern regions (Riley, 1994a(Riley, , 1994bRiley & Michaud, 1989). The sites were selected to represent a range of typical peat accumulation conditions, had a uniform vegetation cover, were >100 ha in size, and had at least 40 cm of peat accumulation. They were classified into bogs, fens, and swamps, depending on their pH, surface peat botanical composition, and tree cover (Riley & Michaud, 1994). Bogs had a pH lower than 5.2, and surface peat was dominated by Sphagnum remains, whereas fens and swamps had a higher pH and a graminoid, woody, or brownmoss-dominated surface peat. Swamps had a tree or tall shrub cover higher than 25%. Peat cores were sampled to a depth of up to 8 m using a mini-Macaulay or Hiller sampler and divided into four intervals or more for deeper profiles, based on their botanical composition and apparent degree of humification. The highest and lowest depth of each section was recorded, as well as ash content, botanical content of the peat (as percent of moss; sedge and other graminoids; wood; and "other materials") and the chemical composition of the samples (Riley & Michaud, 1994). The von Post humification index, created almost a century ago (von Post, 1924) was assessed. It is an index of the degree of decomposition based on the distinctness of the original plant structure, on the color of water squeezed from the peat, and on the tactile properties of the peat (Table S1).
For details on the C, H, O, ash, nitrogen (N), and sulfur (S) analyses, see Riley (1989). Briefly, organic C concentration was determined by treatment with HCl and then ignition in a Leco Induction Furnace at 500°C and trapping in KOH. The H concentration was determined by ignition in a Leco Induction Furnace under a stream of O and absorption of water by a magnesium perchlorate absorption tube. The O concentration was determined by difference: The ash content was determined by ignition at 750°C for 30 min, N concentration by a Kjeldahl method, S by ignition with tin and copper, absorption in HCl, and titration with potassium iodate (Riley, 1989). Analysis of replicate samples of peat shows coefficient of variation of between 4% and 15%, but as C mass dominates, the likely error in O mass by difference is <4%.
Atomic ratios of H:C and O:C were calculated from the individual sample mass values.
To generate the average von Post humification index and H:C and O:C ratios with depth in profiles representing bog, fen, and swamp peatlands, we used the average depth of each sample, and binned them by 10 cm depth intervals for the top 250 cm of the profile, and by 20-cm-depth intervals for the rest of the profile.
We calculated the C ox , eliminating S because of its small concentration in peat (Masiello et al., 2008) and OR values of each peat sample, as defined in Worrall et al. (2016) from atomic concentrations: (from equation (6) in Worrall et al., 2016) We accessed other peat data from UK blanket bogs (Clay & Worrall, 2015, Figure 2) and Latvia-raised bogs (Klavins et al., 2008, Table 1). We accessed the United States Geological Survey (USGS, 2012) peat database (https://energy.usgs.gov/Coal/Peat.aspx#378847-data) for the northeastern United States mainly bogs and fens (Maine and New Hampshire), southeastern United States with a wide variety of peatland types (Florida and Louisiana), and Indonesia likely forested swamps (Borneo and Sumatra).
To interpret the H:C and O:C ratios in the peat, we simplified the decomposition pathways to aerobic respiration resulting in CO 2 and H 2 O and anaerobic methanogenesis to CH 4 and CO 2 , resulting in CH 2 O (Schlesinger & Bernhardt, 2013, Figure 7.9). We used fulvic and humic acids (Clay & Worrall, 2015;Rice & McCarthy, 1991) to establish the stoichiometry of DOC loss. We did not apply stoichiometric changes associated with fire.
There was a pronounced decrease in the von Post humification index in the upper layers of the peat, from average values of 1 to 2 at 5 cm depth, to 3 to 5 at 100 cm, with the decline fastest in the swamp peatlands ( Figure 1a). Below 100 cm, average values ranged from 4 to 5, with little variation with depth or difference among bog, fen, and swamp peatlands. There was a significant logarithmic relationship with depth in all three peatland types (Table 1).

Journal of Geophysical Research: Biogeosciences
There was also a pronounced decrease in the H:C ratios from typical peatland vegetation and litter to the uppermost layers of the peat profiles ( Figure 1b), but there was neither a significant relationship between H:C ratio and depth in the peat types nor a significant difference between bog, fen, and swamp (Table 1). There was also a decrease in the O:C ratio from typical peatland vegetation and litter to the uppermost layers of the peat profiles and then a further decline to a depth of 50 cm ( Figure 1c). Below 50 cm, the change was small, and there was a significant logarithmic relationship between O:C ratio and depth in the bog, fen, and swamp (Table 1).
There was a decrease in the C ox value from the surface layers of the peat (average 0 to 0.2) to a value of À0.2 to À0.3 at a depth of 50 cm and little further decline with depth ( Figure 1d), with a significant logarithmic relationship with depth in the bog and fen profiles, but not the swamp (Table 1). The OR value showed an increase from 1.05 to 1.10 at the peat surface to averages of 1.10 to 1.13 at a depth of about 50 cm, with no further increase deeper in the profile (Figure 1e). There was a significant logarithmic relationship between OR and depth in the bog and fen profiles, but not the swamp, associated with the greater degree of decomposition and larger OR near the surface of the swamp ( Table 1).
Comparison of the ratios with the von Post humification index showed no significant change in H:C ratio with increasing degree of decomposition, whereas there was a significant decrease in O:C ratio, though most of this change occurred between index values 1 and 4 ( Figure 2a). Combination of the two ratios resulted in an increase in the H:O ratio from 2-2.5 to 3.0 as the index rises from 1-2 to >4 (Figure 2b). Both C ox and OR showed a significant relationship with the index, C ox showing a decline from 0.10 to À0.29 and OR an increase from 1.03 to 1.12, with the change being most pronounced from index values 1 to 4 (Figure 2c).
There was no significant relationship between the H:C and O:C ratios and the origin of the peat, when peat was binned into the von Post humification index (Figure 3). Samples that were considered uniform in terms of botanical composition (>80% of their content dominated by one peat type) were grouped into moss, sedge, and wood types. It should be noted that the least decomposed samples (von Post humidification index of 1) did not have enough uniform samples of all three types to allow a comparison.

Relationship Between H:C and O:C Ratio of Ontario and Other Peats
The relationship between the H:C and O:C ratio among the 1,313 peat samples from Ontario is depicted in Figure 4a, showing the very wide range of ratios recorded, with a mean of 1.28 H:C and 0.62 O:C and standard deviation of 0.13 and 0.12, respectively ( Figure 4b). Overall, there was a weak (R 2 adj = 0.040) but statistically significant (p = 0.001) relationship between H:C and O:C ratios with a shallow slope (0.22) of H:C on O:C (Table 2).
Peat samples from other temperate regions showed similar patterns with H:C and O:C means, respectively, of 1.36 and 0.66 (Latvia) and 1.34 and 0.49 (UK), but smaller values of 1.05 and 0.46 for the northeastern United States ( Table 2). The two tropical/subtropical regions showed generally smaller mean H:C and O:C ratios (1.12 and 0.45 for Indonesia and 1.06 and 0.37 for southeastern United States, respectively). In all cases, however, there is a great variation among peat samples from each location. There was also a statistically significant (p < 0.05) relationship between H:C and O:C in all regions, except for northeastern United States (p = 0.063), with variable regression slopes ( Table 2). The slopes were shallow (0.22 to 0.47) for the

10.1029/2018JG004574
Journal of Geophysical Research: Biogeosciences temperate Ontario, Latvia, and northeastern United States peats, but steeper for the UK peat (1.21) and the samples from southeastern United States and Indonesia (0.74 and 1.62), suggesting a greater loss in H accompanying the O loss in decomposition in the latter.

Estimating the Degree of Decomposition of Peat From the H:C and O:C Stoichiometry of Peat and Decomposition Processes
The H:C and O:C characteristics of the peat result from the effect of decomposition processes acting upon the plant tissues and litter from which the peat formed and released C, H, and O. The H:C and O:C ratios of sedge, shrub, Sphagnum, and wood, common contributors to peatlands, are illustrated in Figure 5a, along with the range of ratios input for bog, fen, and swamp peatlands, based on a variable contribution of the four plant categories. This suggests that most input to peatlands may fall within a H:C ratio of 1.5 to 1.6 and a O:C ratio of 0.6 to 0.8. Output represents losses as respiration and methanogenesis with ratios of H:C of 2 and O:C of 1, respectively, DOC (as fulvic and humic acids) of H:C~1.25 and O:C of 0.7, and combustion, which can result in biochar with a H:C ratio of 0.7 and a O:C ratio of 0.3, close to the range of values encountered in coal (see Figure 5b).
Based on the range of original O:C and H:C ratios for bog, fen, and swamp peatlands, the trajectories of ratios associated with increasing decomposition and mode of decomposition can be plotted. As an example, the trajectories of the bog and swamp peat associated entirely with respiration and methanogenesis are plotted in Figure 5b, as well as the trajectories if decomposition was 75% respiration and methanogenesis and 25% DOC (combining fulvic and humic acids). DOC loss, with decomposition output lower in H:C and O:C ratio than gas losses, results in peat with larger H:C and O:C ratios. These trajectories encompass much of the variation in H:C and O:C ratios found in peats and suggest that about 50-75% of the original C has been lost. Fire, important in some peatlands, would result in smaller H:C and O:C ratios; Conedera et al. (2009) cite average O:C ratios of 0.5 and 0.3 for slightly charred biomass and charcoal, respectively, and the equivalent values for H:C are 1.0 and 0.6. However, differences in original composition of litter   Table 2).

10.1029/2018JG004574
Journal of Geophysical Research: Biogeosciences input and the type and degree of decomposition may explain the very large variability in H:C and O:C ratios observed in peat.

Discussion
There are major changes in the H:C and O:C ratios as plant tissues senesce into litter and then as that litter decomposes in the surface layers of the peat. Plant compounds vary considerably in their H:C and O:C ratios, such as 2 and 1 for glucose, 1.7 and 0.8 for cellulose, 1.2 and 0.3 for protein, and 1.2 and 0.35 for lignin Note. Values in bold indicate p < 0.05. Figure 5. (a) Estimated input of H:C and O:C ratio from plant tissues and resulting input into bog, fen, and swamp peatlands, along with ratios of decomposition to CO 2 , methanogenesis and creation of dissolved organic C (fulvic and humic), and burning to biochar. Estimated ratios in plant tissues are derived from Clay and Worrall (2015) and Klavina et al. (2012) and assuming wood is 60% cellulose and 40% lignin. The estimated input into peatlands is based on the following: bog 10% sedge, 20% shrub, 50% Sphagnum, and 20% wood; fen 60% sedge, 20% shrub, 20% Sphagnum, and 0% wood; and swamp 35% sedge, 35% sedge, 0% Sphagnum, and 30% wood. The outputs are based on respiration (to CO 2 ), methanogenesis (to CH 4 and CO 2 ), creation of fulvic and humic acids from Rice and McCarthy (1991) and Clay and Worrall (2015), and biochar from Domingues et al. (2017). (b) Trajectory of changes in H:C and O:C ratios passing through the initial bog and swamp ratios as a function of loss of C, H, and O through respiration and methanogenesis (solid lines) and through a mixture of respiration and methanogenesis (75%) and dissolved organic C (25%; using the average fulvic and humic acid ratios, dashed lines). The estimated ratios representing a C loss of 50% in swamp and bog peatlands are indicated. The central distribution of ratios in coal is indicated (from Large & Marshall, 2015).
( Figures 1b and 1c). Aboveground inputs from plants also vary from Sphagnum moss (1.7 and 1.0), shrubs (1.6 and 0.6), and sedges (1.6 and 0.7) to woody plants (1.5 and 0.7; e.g., Clay & Worrall, 2015;Kļaviņa et al., 2012). In peatlands, the belowground to aboveground biomass ratio is relatively large (e.g., Murphy et al., 2009) and fine root production is often tied to aboveground biomass and water table position (e.g., Murphy & Moore, 2010). Less is known about the composition and amounts of belowground inputs to soil organic matter (Kögel-Knabner, 2002, so it is difficult to assess their contribution to the observed peat H:C and O:C ratios. There is a rapid change in H:C ratio from some plant tissues to uppermost layer of peat, but little change with depth in the peat column: H is lost at the same rate as C. In contrast, the O:C ratio shows a decline from most plant tissues and continues to decline in the uppermost layers (0-50 cm) of the bog, fen, and swamp profiles, with few changes lower in the profile. This depth coincides with the position of the water table in many peatlands and the conversion from aerobic to anaerobic decomposition.
Assuming that northern peatlands store 436 Pg of C (Loisel et al., 2014) and that the average atomic H:C and O:C ratios of 1.3 and 0.5, respectively, from the Ontario data ( Figure 1) are applicable, then northern peatlands, which have stored C primarily in the last 8,000 years (Yu et al., 2010), contain~45 Pg H and 290 Pg O, compared to~19 Pg N (Wang et al., 2015).
The Ontario peatlands, with a large number of samples analyzed from varying peat types and depths, showed that there was a strong variation in C ox and OR with depth in the upper part of the three peatland types, C ox decreasing and OR increasing, but that there was little change beneath 50 cm (Figures 1d and 1e). Given average growth rates of peat in bog profiles in Canada (see Talbot et al., 2017), the upper 50 cm represents material mainly accumulated in the last 200 years and under mainly aerobic conditions.
While there is a great variation among samples, the data from the Ontario peatlands suggest that C ox values of 0.079, 0.038, and 0.017 would be applicable to the top 30 cm of bog, fen, and swamp soils, respectively, falling to À0.214, À0.109, and À0.226 for peat deeper than 100 cm (Figure 1d). Similarly, the bog, fen, and swamp values of OR would be 1.031, 1.048, and 1.077 the top 30 cm, rising to 1.120, 1.132, and 1.104 below 100 cm (Figure 1e). These values contrast with a median C ox value of À0.33 for UK peat and median OR values of 1.03 for histosols globally and 1.10 for UK peat, again in the top 30 cm (Clay & Worrall, 2015;Worrall et al., 2013). As the OR value has been used to estimate the C flux to the terrestrial biosphere (Worrall et al., 2013), the Ontario peats tend to support these OR numbers. However, if the large amount of peat stored deeper than 30 cm were to play a greater role in atmospheric exchange (as might occur if peatlands became warmer and drier), then the increase in OR with peat depth to~1.12 would need to be taken into consideration, potentially decreasing the magnitude of the C flux to oceans and land.
Although the von Post humification index is a qualitative measure of the peat and subject to operator variability, it does appear to capture variations in the chemistry of the peat. Here we show that the index is strongly correlated, over a wide range of peat samples, with the O:C and H:O ratios and the C ox and OR values, but not H:C ratio ( Figure 2). Most of the changes occur from 1 to 4 in the index, with little further change with increasing degree of humification. However, based on a broad classification of samples, botanical origin of the peat did not appear to be influential on the ratios. Analysis of the same set of Ontario peat samples showed that the nutrient content, expressed as the C:nutrient ratio, was also related to the index for nitrogen, phosphorus, calcium, magnesium, and potassium, with most of the change observed from 1 to 4 (Wang et al., 2015, Figure 4). The incubation of peat samples collected from natural, harvested, and restored bogs in eastern Quebec also showed a correlation between aerobic CO 2 and anaerobic CH 4 production with von Post humification index, with much of the change occurring between 1 and 4 (Glatzel et al., 2004; Figure 5).
Although there was a great variability among samples, the H:C and O:C ratios for the peat samples from northeastern and southeastern United States and Indonesia were substantially smaller than those from Ontario, UK, and Latvia. In part, this may reflect differences in the analytical methods employed by the USGS for coal, (the "ultimate analysis" ASTM D3176, expressed as the elemental composition of the organic material (ASTM, 2011)) and the other groups. This would imply a larger C content than the other methods or a smaller H and O content, or a combination. Differences in moisture content of the sample when analyzed may account for this, with the USGS method using "proximate analysis" to provide an independent correction for the equilibrium moisture content, thereby reducing the H:C and O:C ratios to a truly dry basis. Alternatively, this may reflect a greater degree of decomposition, which could be expected in the tropical and subtropical samples, though unlikely in most of the samples collected from northeastern United States. Moreover, the Ontario samples suggest that there are few changes in H:C and O:C ratio from slightly to well decomposed peat.
There do, however, appear to be differences in the overall relationship between the H:C and O:C ratio in three of the four temperate peat sets (Ontario, Latvia, and northeastern United States) compared to the two tropical/subtropical sets (southeastern United States and Indonesia). The steeper regression in the latter pair ( Figure 4 and Table 2) suggests that H is being lost at a faster rate relative to O, compared to the temperate peatlands, which may result from differences in decomposition pathways or may indicate that decomposition has progressed further in regions with warmer soil temperatures. Fire will accelerate the reduction in H:C and O:C ratios, particularly the former.
We suggest that the stoichiometric relationship between the peat and the major decomposition pathways can be used to estimate how much of the original plant material has been decomposed, although there is a great variability in individual peat stoichiometry as well as potential decomposition pathways. The H:C and O:C ratios vary among vegetation, which is the input to the peatlands. The ratios in decomposition pathways also vary, such as through aerobic respiration, anaerobic methanogenesis by acetoclastic or hydrogenotrophic pathways, DOC with a range of compounds, and finally, the effect of combustion. Nevertheless, there appears to be evidence that a simple combination of inputs for bog, fen, and swamp peatlands, combined with outputs through respiration and methanogenesis with or without DOC can be used to estimate how much of the original C has been lost. For example, our estimate of 50% C loss for a bog peatland (Figure 5b) occurs at an average O:C ratio of 0.62 (combining the two decomposition pathways), which is at a depth of about 20 cm based on the Ontario bog profiles (Figure 1c and Figure S1). The Peat Decomposition Model (Frolking et?al., 2001), based on cohort inputs and estimated decomposition rates, estimates that about half the annual cohort mass of a bog would be lost at a depth of about 15 cm.
To test the decomposition history of the continuum from plant tissue to litter to peat, further work could include sites at which the C, H, and O stoichiometry of inputs, decomposition pathways, and peat is known, along with dating of the peat profile.