Relative importance of climatic and autogenic controls on Holocene carbon accumulation in a temperate bog in southern Ontario,Canada

Chronologies

Peat core LB03 was 601 cm in length, and the full length of the core was analyzed for Holocene-scale interpretations. The chronology for the LB03 core was obtained through a combination of ²¹⁰Pb and ¹⁴C dating, along with a charcoal horizon at depths 18–20 cm corresponding to a local fire documented in historical records in ad 1870 (Sandilands, 1984), and the rise of Ambrosia pollen at 20 cm (Figure 2), taken to indicate the onset of land clearance and Euro-Canadian settlement, ad 1850 in this area (McAndrews and Boyko-Diakonow, 1989).

OPEN IN VIEWER

Figure 2.

Depth (in cm) of the Ambrosia pollen rise signaling Euro-Canadian settlement, ~ad 1850 in the vicinity of WLB, for cores LB02 and LB03.

The relationship between ln [excess ²¹⁰Pb] and the cumulative dry mass of sediment in the peat core LB03 is approximately linear for the middle five data points (Figure 3), suggesting that ages can be modeled with some degree of confidence from ²¹⁰Pb activities in the core. The upper three samples analyzed for dating show lower than expected activities, likely because of the fact that these samples included living vegetation at the surface of the peatland, and low-density acrotelm peat. Below 7 cm, the peat is more compact and similar in texture and porosity to the lower portions of the core. The break in the slope of the ²¹⁰Pb activity curve (Figure 3) at depth 18 cm may be related to the peat loss caused by the local fire of ad 1870. Thus, the accumulation rate for age inference is calculated from the middle five data points. We do not infer any ages below the charcoal horizon and use the depth of 17 cm as the lowest date confidently dated by ²¹⁰Pb activity, and based on constant rate of supply modeling, this age is approximately ad 1930. This age model suggests that the fire likely resulted in several centimeters of peat loss, which has been reported in other studies of peatlands subject to fire (Turetsky and Wieder, 2001). Therefore, we interpret the chronology with caution through this portion of the core. The lower, pre-European settlement portion of core LB03 is dated with six AMS radiocarbon dates (Table 1). The base of the most reliable portion of the ²¹⁰Pb chronology (ad 1930), the fire horizon (ad 1870), the Ambrosia pollen rise (ad 1850), and the six AMS radiocarbon dates were used to construct an age–depth model by linear interpolation for core LB03 (Figure 4a).

OPEN IN VIEWER

Figure 3.

Excess ²¹⁰Pb activity (ln scale), as a function of cumulative dry peat mass (g/cm²) and depth (cm) in core LB03.

OPEN IN VIEWER

Figure 4.

Age–depth models produced by linear interpolation between dating points for (a) LB03 and (b) LB02. Radiocarbon dates are shown with error ranges associated with the probability distributions resulting from the calibrations.

The basal date for core LB03 (10,250 cal. yr BP) confirms an early Holocene initiation for the peatland, and a mean Holocene peat accumulation rate of 0.06 cm/yr. Other studies of WLB returned similar basal dates and/or mean rates of peat accumulation (Givelet et al., 2003; Irwin, 1988; McAndrews, 1996). The LB03 age–depth model shows some variability in peat accumulation, with slower rates in the early Holocene, and again in the late Holocene, after 2400 cal. BP. It is acknowledged that the low number of radiocarbon dates for a whole Holocene sequence precludes any detailed discussion of rates of change in rates of peat accumulation. The most recent apparent increase in rate of peat accumulation in the age–depth model (Figure 4) reflects the lack of peat compaction in the upper acrotelm portion of the core; therefore, the recent rate is not directly comparable to Holocene rates.

Only the top 100 cm of the second core, LB02, was analyzed. An age–depth model for this section was constructed using the charcoal horizon corresponding to the ad 1870 fire (21–23 cm), the rise in Ambrosia pollen at 25 cm (Figure 2), and four AMS radiocarbon dates (Table 1). The resulting age model (Figure 4b) shows a similar mean rate of peat accumulation as recorded in the longer sequence (LB03), 0.06 cm/yr. There is also a similar apparent increase in rate of peat accumulation in the uppermost portion of the core, but as bulk density is lower in the acrotelm, these rates are not directly comparable to those of the Holocene.

The charcoal horizon corresponding to the ad 1870 fire is more pronounced in LB03 than in LB02, with a larger number of charcoal pieces, and larger sized pieces Figure 5a and b). The amount of charcoal remaining in situ after a peatland fire is related to how much biomass was burned at that location (Turetsky and Wieder, 2001), suggesting greater combustion at the site of LB03. The break in slope in excess ²¹⁰Pb activity below the charcoal horizon in LB03 (Figure 3) and the depth of the Ambrosia pollen rise at a depth of 5–6 cm higher in LB03 relative to LB02 (Figure 2) suggest that at least 5–6 cm of peat was lost from core LB03 because of burning. The more abrupt increase in Ambrosia pollen in LB03 versus LB02 and the 400-year offset in radiocarbon dates for each core at dated depth of 49 cm (Table 1) suggest that more than 5–6 cm of peat was combusted in the LB03 sequence. Based on the rate of peat accumulation calculated for LB02, the 400-year age offset seen at depth 49 cm in core LB03 could correspond to up to 20 cm of peat lost. Differential burning depths may be related to variations in microtopography. Turetsky and Wieder (2001) found that fires occur irregularly within peatlands and that unburned sections can be as close as 50 cm to highly combusted peat. Peat losses in LB03 result in unrealistically low rates of apparent peat and C accumulation; as we infer significant peat loss in the LB03 sequence for the late Holocene, we focus our interpretation of the past two millennia on the LB02 sequence, from which there is little evidence for peat loss.

OPEN IN VIEWER

Figure 5.

Macrofossil and humification record for (a) core LB03 and (b) core LB02. Abundances of macrofossil are expressed in terms of five categories and a percent cover scale (see text for details). Charcoal and sclerotia are plotted as numbers of fragments. Round symbols denote samples where the number of charcoal fragments is below 2. Humification is plotted as absorbance values (ABS). Major transitions in macrofossil assemblages/wetland types, the onset of MCA and LIA climate changes based on regional pollen records, and the depths corresponding to the water table (WTD) and the construction of a nearby dam (ad 1954) are also shown.

Paleovegetation

Macrofossil analysis provides information on local paleovegetation during the period of record. At site LB03, peat accumulation begins at ~10,250 cal. BP (Figure 5a). Highly decomposed organic remains dominate the early Holocene, and this period coincides with low abundances of all other macrofossil materials; small numbers of charcoal fragments are present and peat humification is low. Around 9100 cal. BP, non-Sphagnum mosses and herbaceous peat begin to dominate macrofossil assemblages, while peat humification values increase markedly. Around 5400 cal. BP, non-Sphagnum mosses and herbaceous peat macrofossils slowly begin to be replaced by remains of Sphagnum mosses. By 4000 cal. BP, the transition from fen-like indicators to bog indicators is complete, and peat humification values increase somewhat relative to lower sections of the core (Figure 5a). The peat humification curve is variable through most of the core, and small changes in absorbance are often coincident with small shifts in the macrofossil record. The core contains some significant changes in peat-forming vegetation; thus, peat humification reflects mainly these vegetation changes, as opposed to actual variations in surface wetness (Bunbury et al., 2012; Yeloff and Mauquoy, 2006).

Sphagnum mosses are usually well preserved in the portion of the core above the fen to bog transition. This Sphagnum-dominated zone is also characterized by episodic fluctuations in the abundances of macrofossils derived from woody vegetation, while non-Sphagnum mosses are virtually absent. Charcoal fragments and sclerotia of the ectomycorrhizal fungus Cenococcum geophilum syn graniforme (Spatafora and Owensby, 2012) are occasionally observed (Figure 5a) and are interpreted as fire indicators (Benedict, 2011). The peak in charcoal fragments at 18–20 cm is associated with the ad 1870 fire. The uppermost portion of the LB03 core continues to be dominated by Sphagnum macrofossils, but peat humification values drop through the acrotelm.

The changes documented in the LB03 Holocene macrofossil record are well supported by available pollen data from the WLB1 site (Figure 6). The Wylde Bog (WLB1) pollen record shows the presence of a number of local fen indicators from peatland initiation until about 5000 cal. BP, including Cyperaceae, Equisetum, and Larix; Sphagnum spores become abundant also after 5000 cal. BP, confirming this timing for the fen to bog transition (McAndrews, 1996).

OPEN IN VIEWER

Figure 6.

Percent pollen diagram for the WLB1 site (diagram drawn from data of McAndrews (1996)). The age model was determined from two available radiocarbon dates, and standard pollen sums were used for calculating percentages (see text for details). The upper panel shows tree and shrub pollen types; the lower panel shows aquatic, wetland, and herbaceous pollen types.

The macrofossil record from the shorter, late Holocene sequence (core LB02; Figure 5b) reflects the trends shown in the LB03 record, but the shorter sequence provides greater temporal resolution with an apparently more complete record through the most recent 1500 years (Figure 3). LB02 comprises mostly well-preserved peat-forming vegetation. Sphagnum is the dominant macrofossil type; however, the record shows three short episodes in which Sphagnum abundance decreases to <10%, while the proportions of macrofossils of herbaceous peat, ligneous peat, or roots increases. As in LB03, sclerotia of C. geophilum are recorded periodically. The significant peak in charcoal abundance at 21 cm is associated with the ad 1870 fire, but it is less prominent than in LB03 (see above). Peat humification varies little through the catotelm portion of core LB02, but, as in LB03, values shift somewhat in association with the changing frequencies of Sphagnum versus other macrofossil types (Figure 5b). As in LB03, absorbance values decrease through the acrotelm, as living vegetation and litter are less humified.

C accumulation

Mean bulk density for both WLB cores is 0.07 g/cm³, mean C content for the two cores is 48.6%, and mean N content is 1.5%. Measurements of bulk density and C content were combined with the age–depth models to calculate apparent LORCA (Figure 7). In the LB03 Holocene macrofossil sequence, the oldest section of the core is dominated by highly decomposed organic matter (Figure 5a); higher values of bulk density and lower C:N ratio at that time (Figure 7a) suggest limnic sediments, associated with low values for LORCAs. By 8100 cal. BP, bulk density decreases and the macrofossil record becomes dominated by herbaceous peat; C:N ratio and LORCA begin to increase. At the fen to bog transition, beginning around 4100 cal. BP, C:N ratio increases again, leading to marginally higher values for LORCA, particularly between 2400 and 3000 cal. BP. After 2400 cal. BP, bulk density increases, and C:N ratio begins to drop as ligneous and herbaceous peat become more important relative to Sphagnum peat in the macrofossil record. Decreases in bulk density and the inferred decrease in the rate of peat accumulation since 2400 cal. BP (Figure 4) result in very low values for LORCA in the most recent two millennia in the LB03 core; this decrease, however, is related to the age model in which radiocarbon dates were likely ‘too old’ relative to the depths dated because of up to 20 cm in lost peat following the ad 1870 fire (see above). Mean LORCA (±standard deviation (SD)) for core LB03 is 20 ± 11 g C/m²/yr; when only the portion of the core below the fire of ad 1870 is considered, mean LORCA (±SD) is 18.6 ± 6.4 g C/m²/yr. There was no apparent difference in LORCA for the fen versus bog portions of the LB03 core (Mann–Whitney U statistic = 5886, p > 0.05).

OPEN IN VIEWER

Figure 7.

C:N (mass) ratios, bulk density and calculated LORCA for (a) core LB03 (whole Holocene) and (b) core LB02 (most recent 2000 years). The shaded bar on the LB03 curves indicates the section of the core affected by combustion and overly low rates of calculated peat accumulation because of peat losses resulting from the ad 1870 fire. C:N and LORCA are plotted using a log₁₀ scale, while bulk density is plotted on a linear scale.

In the LB02 late Holocene sequence, where we do not suspect significant peat loss, C:N ratio fluctuates with shifts in the relative abundance of Sphagnum versus ligneous macrofossils, with higher C:N ratios occurring when Sphagnum is more dominant. Lower C:N values since 700 cal. BP contribute to the lower LORCA recorded for that time (Figure 7b). The highest values for bulk density in both sequences are recorded at the charcoal horizons associated with the fire of ad 1870 (Figure 7). Mean LORCA (± SD) for core LB02 is 24 ± 16 g C/m²/yr; when only the portion of the core below the 1870 fire is considered, mean ± SD is 18 ± 5.4 g C/m²/yr. LORCA in core LB02 decreases as the abundance of Sphagnum spp. macrofossils decreases, around 1400 cal. BP and even more so, around 550 cal. BP. LORCA reaches maximum values near the depth of the ad 1870 fire charcoal layer and the approximate depth of the water table. After the construction of Luther Lake (ad 1954), LORCA values fall and remain constant until the surface.

Explaining variability in Holocene C accumulation

Low C:N ratios and high sediment bulk densities, coupled with highly decomposed organic remains in the macrofossil record (Figure 5a and b) and abundant Cyperaceae pollen (Figure 6; McAndrews, 1996), suggest marsh-like conditions in the early Holocene at the site of core LB03. While some higher latitude peatlands show elevated rates of apparent C accumulation in the early Holocene in response to orbitally mediated increases in summer insolation and warmer temperatures associated with the HTM (Jones and Yu, 2010; Van Bellen et al., 2011a), LORCA in the early Holocene is low at LB03. Early peatland development at this site is controlled more by post-glacial geomorphic setting than climate. Furthermore, peatlands at lower latitudes experienced insolation anomalies of lower magnitude and less pronounced thermal maxima in the early Holocene. C accumulation in temperate peatlands is therefore likely less sensitive to orbitally mediated climate changes. Other studies of temperate peatlands similarly show lower values for LORCA in early successional marsh-like phases (Cai and Yu, 2011).

Around 8100 cal. BP, C:N ratio and peat humification increase in the LB03 record, coincident with a decrease in bulk density, and a shift from highly decomposed organic remains to herbaceous peat, ligneous remains, and brown mosses in the macrofossil record, indicating a transition to a peat-accumulating fen ecosystem. LORCA values increase at this point. The WLB1 pollen record reveals an increase in fern taxa and Larix at this time (Figure 6; McAndrews, 1996), further supporting the interpretation of a fen-like environment. A shift from a shallow water marsh to fen is frequently linked to terrestrialization because of infilling by organic matter (Cai and Yu, 2011; Muller et al., 2003).

By about 4000 cal. BP, C:N ratio had increased markedly, along with Sphagnum remains in the macrofossil and pollen records, suggesting the completion of the fen to bog transition. The bog zone in the core is associated with variable LORCA; maximum LORCA values for the catotelm portion of the LB03 record occur between 2500 and 3000 cal. BP. Marginally higher values within a portion of bog zone for LORCA suggest some possibility of climatic controls, since once the bog becomes established, directional autogenic processes cannot explain episodic changes, unless nonlinear dynamics are invoked (Belyea and Baird, 2006). The pollen record shows the decline in Tsuga pollen between 5000 and 3000 cal. BP. Among other factors, drought is considered important in explaining the Tsuga decline (Haas and McAndrews, 1999). The re-initiation of wetter conditions after 3000 cal. BP, coincident with the onset of ‘Neoglacial climates’, as shown here through the re-establishment of Tsuga at that time, suggests a wetter climate which may be promoting higher values for LORCA, either by enhancing production or by reducing rates of decomposition. There is a notable decrease in LORCA following ~2270 cal. BP, which reflects the reduction in the rate of peat accumulation inferred from the age model (Figure 4). Because of peat losses of perhaps up to 20 cm related to a local fire in ad 1870 in the LB03 core, and resulting anomalously old radiocarbon dates (Table 1), the decrease in LORCA is interpreted as reflecting peat loss.

The LB02 record provides an opportunity to examine changes in LORCA over the past two millennia in a sequence in which peat was likely not lost to combustion (Figure 7b). Calculated values for LORCA are variable through this period. There has been considerable interest in linking C accumulation in peatlands to the climate changes associated with the MCA and the LIA (Charman et al., 2013; Finkelstein and Cowling, 2011), observed to have been time transgressive across the northern hemisphere. To evaluate those relationships in the LB02 record, we use local pollen datasets to assign timing for MCA and LIA initiation at WLB. Campbell and McAndrews (1991) summarize pollen data from a series of records in Ontario and show using cluster analysis that pollen assemblages begin to change in response to LIA cooling after ad 1250 (700 cal. BP), intensifying after ad 1450 (500 cal. BP). In the region of WLB, these changes typically include decreases in the abundance of Fagus pollen and increases in Pinus pollen percentages.

Because of the importance of wetland pollen types in the WLB1 pollen record (McAndrews, 1996) and the relatively low sampling resolution of that pollen record (samples are on average ~150 years apart; Figure 6), it is not possible to extract detailed information from the record on the timing of the LIA and the MCA. Despite these limitations, small changes in the abundances of Fagus and Pinus pollen are noted in the WLB1 pollen data (McAndrews, 1996). The dip in Pinus pollen and concomitant increase in Fagus pollen between 1220 and 1470 cal. BP (equivalent to ad 490–730) may be indicative of the MCA. Similarly, the small decreases in Fagus and increases in Pinus pollen which had begun to take place by the sample dated to 730 cal. BP (ad 1220) may be indicative of the onset of the LIA. These ages are consistent with the summaries reported on by Campbell and McAndrews (1991, 1993) and with the varve dated record for nearby Crawford Lake (McAndrews and Boyko-Diakonow, 1989). On the basis of these nearby records and the WLB1 pollen record (McAndrews, 1996) located within 1 km of our coring site, we date the onset of LIA cooling in the region to around 730 cal. BP, with intensification after 500 cal. BP. The MCA tends to be more subtly recorded in records from southern Ontario, but on the same bases, we place its onset around 1450 cal. BP.

LORCA as reconstructed from the LB02 core is variable over the past two millennia, and a strong contrast between values during the MCA versus LIA is not observed (Figure 7b). The MCA, ~1450 to ~750 cal. BP at the study site, however, contains periods of elevated LORCA and somewhat higher C:N ratios. The LIA phase at WLB is characterized by minimum LORCA values for the record, and lower C:N ratios. Lower values for LORCA during the LIA in the LB02 core are associated with a vegetation shift from Sphagnum spp. to more N-rich ligneous and some herbaceous remains in the macrofossil record, suggesting decreases in surface wetness during this time (Loisel and Garneau, 2010), which were likely climatically controlled (Cai and Yu, 2011). This interpretation suggests that LIA climates in the region may have been cool and dry, whereas other studies have suggested cool/wet LIA climates (Finkelstein and Davis, 2006; Pederson et al., 2005). Syntheses of multiple multi-proxy records from the region are required to resolve the LIA climate history for this region.

Role of disturbance

Similar to other studies on temperate peatland fire regimes (Lavoie and Pellerin, 2007), fires are infrequent in the LB02 and LB03 Holocene macrofossil records. Lavoie and Pellerin (2007) suggest that temperate peatlands are typically surrounded by deciduous trees or mixed forest which are not fire-prone and thus burn less frequently than boreal peatlands. In the humid climate of southwestern Ontario, natural fires are rare and, thus, do not play a large role in peatland vegetation dynamics or carbon accumulation. Charcoal fragments > 400 µm in size and sclerotia of C. geophilum are only evident in high-enough abundance to confirm the presence of one local fire for the Holocene at WLB. This is the fire of ad 1870, and is of anthropogenic origin. This single fire is important to the interpretation of the late Holocene record from this core, as it apparently resulted in ~20 cm of peat loss in the LB03 profile, complicating the interpretation of LORCA in the late Holocene in this core.

Other recent disturbances, such as local deforestation, drainage, and damming, may also have impacted LORCA in the LB02 record; however, values for LORCA in the more aerated acrotelm portion of the peat cores should not be compared directly with results for the waterlogged catotelm, owing to different bulk densities and C composition. Additionally, the acrotelm–catotelm boundary, the location of the charcoal horizon deposited in the ad 1870 fire, and the onset of European settlement associated with draining and deforestation of the surrounding area are all recorded within <5 cm in both cores. Thus, their respective impacts on C accumulation are not easily deciphered.

While considering the above caveats, it is still clear that the LORCAs following the ad 1870 fire are high. These high values may be primarily related to elevated bulk density resulting from charcoal deposition. The acrotelm portion of the core is situated immediately above the fire horizon, and is characterized by large decreases in bulk density as porosity increases and humification of the peat decreases. C:N ratio also increases in the acrotelm relative to underlying peat, as expected for surface peats that have only been subject to limited decomposition. While these changes are expected across the acrotelm–catotelm transition in an undisturbed core (Rydin and Jeglum, 2006), anthropogenic disturbance may also have been a factor (Ireland and Booth, 2012). The deforestation and draining associated with European settlement, as well as the tree cover removed during the ad 1870 fire likely resulted in higher surface wetness (Lavoie and Pellerin, 2007) and more frequent flooding of the Grand River (Chapman and Putnam, 1984; Sandilands, 1984), as fewer trees were present to lock moisture in the soils and more water was directed toward the river. This interpretation is also supported by the macrofossil record, which reveals low ligneous cover during the time after the ad 1870 fire, and before the damming. Thus, increased surface wetness may have contributed to the higher LORCA recorded between the ad 1870 fire and the damming of Luther Lake in ad 1954.

Allogenic and autogenic controls on peatland development

The Holocene record of peatland development shows that this temperate wetland site in southern Ontario underwent a series of ecological changes in transitioning from a marsh to a fen to a bog; these changes mirror those reported from numerous other peatland sites. These changes in some cases were promoted by directional autogenic processes, namely, the accumulation of organic matter and the eventual segregation of the peatland surface from the water table. However, there is also evidence for climatic impacts on peatland ecology and C accumulation. For example, the drier climate inferred during the Tsuga decline may explain the periodic increases in woody vegetation on the bog surface and fluctuations in LORCA in the latter portion of the Holocene. During the past millennium, climate changes associated with the MCA were at times associated with somewhat higher values of LORCA, along with changes in peatland vegetation indicative of a wetter bog surface. The LIA was characterized by the lowest values of LORCA for the whole record, which may have been related to drier conditions, as suggested by increases in woody vegetation. Nevertheless, the difference in LORCA recorded for the MCA compared with the LIA is subtle. The climate of the study region is unlikely to have been moisture-limiting at any time during the Holocene from the perspective of peatland plant communities, thus autogenic processes are likely major factors at this site, as was also reported for other temperate wetlands (Muller et al., 2003). While in humid climates allogenic controls are less important in explaining peatland dynamics, there is evidence from the LB02 and LB03 records to suggest that periodically drier conditions did change peatland vegetation and consequently rates of apparent C accumulation.

Whereas during the Holocene there is little evidence for effects on peatland structure or function of fire or other disturbances, the LB02 and LB03 peat cores record major changes associated with a suite of disturbances since Euro-Canadian settlement in the area. The comparison of two cores, LB02 and LB03, taken in close proximity also shows the heterogeneous nature of peat accumulation across small spatial gradients and emphasizes the caution which must be used in making whole-peatland inferences from a single core. Despite this, studies such as ours indicate the potential role that temperate peatlands could have held in Holocene C dynamics. In some temperate regions, >80% of peatlands have been destroyed because of human land use change. An estimation of the total area once covered by temperate wetlands prior to European colonization might reveal that although LORCA are generally lower in temperate relative to boreal wetlands, the areal size of temperate peatlands may have been so great that they were an important sink of C for much of the Holocene. Continued research in reconstructions of C balance in Holocene peatlands of the temperate, boreal, and high-latitude regions will likely show that peatlands may have exceeded Holocene vegetation as C sinks and that their modern-day representation severely under-represents their potential for C storage over most of the past 10,000 years.