Introduction
Fire is a major disturbance to net primary production in boreal forest
ecosystems . Determining how fire regimes and ecosystems have
changed in the past provides insight into how climate change may influence
fire and its impact on the carbon cycle in the future. Terrestrial and
lacustrine sedimentary charcoal records are the main source of information
about regional variations in past biomass burning. A synthesis of high-latitude Northern Hemisphere charcoal records indicates a gradual decline in
burning related to late Holocene cooling, followed by an increase from
1750 to 1870 CE and a decline after 1870 CE associated with anthropogenic
activity . Biomass burning increased during the first half of
the 20th century, declined during the second half of the 20th century, and
rose sharply after 2000 CE . Siberia is the largest forested
area in the Northern Hemisphere, and Siberian wildfires constituted 5 to
20 % of global biomass burning carbon emissions from 1998 to 2002 CE
. There are only 11 Siberian sedimentary charcoal records in
the Global Charcoal Database . These records cover very
different age ranges at varied temporal resolutions. As a result, it is not
yet possible to reconstruct Siberian biomass burning trends on centennial or
millennial timescales with confidence .
Biomass burning emissions histories also have been inferred from a variety of
different ice core proxies . Variations in the
stable isotopic composition of ice core methane have been used as a proxy for
global biomass burning emissions . Biomass
burning is not the primary source of atmospheric methane as more methane is
emitted from geologic and various microbial sources. The contribution from
burning is calculated from measurements of the stable isotopic ratio
(13C / 12C) of methane by assigning end-member isotopic
compositions to the various sources. Late Holocene methane isotopic records
show that global burning emissions were high from 1 to 1000 CE, declined from
1000 to 1700 CE, and increased again after 1700 CE .
Several other chemicals with shorter atmospheric lifetimes have been used as
regional, rather than global, fire proxies. Ammonium has been commonly used
as a tracer for biomass burning in several ice cores . Elevated levels of ammonium with the same timing as
elevated levels of formate, acetate, oxalate, glycolate, formaldehyde,
hydrogen peroxide, potassium, nitrate, or black carbon have been interpreted
as indications of elevated burning periods in ice cores . The concurrent timing of ammonium spikes with decreases in
electrical conductivity has also been used as an indication of biomass
burning in ice cores . The challenge of
using ammonium as a biomass burning tracer is that it has several other
sources, including animal excreta, synthetic fertilizers, oceanic sources,
crops, natural vegetation soils, lightning, industrial processes, fossil
fuels, and other anthropogenic sources .
Simultaneous timing of ammonium and formate peaks in ice cores have also been
used as a proxy for increases in biogenic emissions due to periods of
increased temperature . Ice core proxies that are
uniquely derived from burning are needed to confirm the interpretation of
ammonium as a biomass burning tracer.
Black carbon in ice cores has been used as a tracer for preindustrial biomass
burning .
Ammonium, formate, black carbon, and organic carbon (dissolved organic carbon
or total organic carbon) were enriched substantially relative to background
levels during fire events in Greenland ice . During
industrial times, black carbon in ice cores also originates from fossil fuel
combustion. Differences between black carbon and other biomass burning proxy
records in ice cores has been attributed to transport and combustion
conditions. Black carbon primarily is produced under flaming combustion
conditions, while ammonia is generated by smoldering fires .
Boreal fires, which are geographically closer to high-latitude ice core
sites, are often smoldering fires .
Wildfires generate a wide range of aerosol-borne organic compounds that are
derived from the partial combustion of plant matter. Levoglucosan is an
aerosol-borne anhydrous sugar exclusively produced by burning of cellulose
. Levoglucosan is a promising tracer because it makes up a
large fraction of the organic aerosol mass produced by biomass burning, is
emitted from burning of all types of cellulose-containing plant matter, and
has been detected in aerosols and ice in polar regions . However, the utility of
this compound as a quantitative tracer is somewhat controversial due to the
potential for rapid degradation in the atmosphere .
Wildfires also generate phenolic breakdown products derived from the
pyrolysis of lignin during the smoldering stage of a fire . The chemistry of these emissions reflects the
composition of the precursor lignin and the rate, temperature, and oxidative
conditions under which burning occurs. Aromatic acids, such as vanillic acid
(VA), para-hydroxybenzoic acid (p-HBA), and syringic acid are molecules that
are diagnostic of biomass burning because they retain the basic aromatic
building block of the precursor lignin Fig. S1 in the
Supplement;. Laboratory burning studies show a range in the yield of
different aromatic acids from natural biomass fuels. For example, burning of
North American conifers produces both VA and p-HBA, with VA in greater
abundance, while North American tundra grass fires produce p-HBA with
essentially no VA . German peat fires have been shown to produce both VA
and p-HBA, with p-HBA in greater abundance . Surprisingly,
there are no published studies examining the burning products of Siberian
flora.
The abundance of aromatic acids, such as VA and p-HBA, in polar ice cores
reflects the combined effects of biomass burning emissions, atmospheric
transport and transformations, depositional processes, and post-depositional
processes. VA and p-HBA are semi-volatiles that may reside in either the gas
or condensed phase depending on temperature, aerosol water content, pH, and
cation concentrations. There is some debate regarding the atmospheric
lifetimes of these semi-volatile compounds because in the gas phase they can
react rapidly with hydroxyl radicals. The OH lifetime for gas-phase oxidation
of these compounds is on the order of a day. However, modeling suggests that
such compounds are shielded from oxidation inside aerosol particles, with
atmospheric lifetimes of several days . Such lifetimes are
supported by observations of long-distance atmospheric transport of biomass
burning aerosols. There are numerous observations of aromatic acids in
burning-derived coarse- and fine-mode atmospheric aerosols in terrestrial,
marine, Arctic, and Antarctic environments . The field observations and model estimates of
reactivity support the idea that long-distance transport is the likely source
of these compounds to the ice sheet. Near-surface postdepositional processes
such as revolatilization, photochemical reactivity, or meltwater
infiltration have not been studied for aromatic acids. Such processes could
potentially influence the ice core levels of these compounds, particularly at
low-accumulation sites .
FLEXPART model forward trajectories suggest that in the summer aerosol
transport to the Arctic from biomass burning sources is primarily from
Siberia 48–66∘ N,
60–140∘ E;. The FLEXPART model is a Lagrangian
transport and dispersion model that is used to simulate long-range
atmospheric transport. Twenty-five percent of FLEXPART modeled forward
trajectories from Siberia reached the Arctic in 3 days, and 50 % reached
the Arctic in 10 days . Russian fires in 2003 contributed
40–56 % of the mass of black carbon deposited in the spring and summer above
75∘ N . Lidar data from the Arctic Research of the
Composition of the Troposphere from Aircraft and Satellites (ARCTAS) mission
indicate that biomass burning plumes from Russian forest fires in 2008
contributed to aerosol loadings over the North American Arctic
.
The deposition of burning-derived aerosols to the polar ice sheets and ice
caps raises the possibility that ice cores contain well-dated biomass burning
records that integrate fire emissions over wide geographic regions. These
records are complementary to other historical records of biomass burning,
such as sedimentary charcoal and ice core gas records. Although more complex
in terms of integrating both emissions and transport, the ice cores bring
some advantages to the study of past biomass burning. They are regionally
integrated archives of climate and burning proxies that are stored in the
same well-dated record. The information stored in aromatic acid records is
distinctly different from that contained in ice core gas records of methane
stable isotopes, carbon monoxide, or ethane, which are influenced primarily
by tropical, rather than boreal or high-latitude emissions .
The distribution of these aerosols in polar ice cores has not been
systematically investigated. VA and p-HBA were measured in a 300-year ice
core from the Kamchatka Peninsula, northeastern Asia , and VA
was measured in a 200-year ice core from the D4 site in west-central
Greenland . VA, p-HBA, and several other organic biomass
burning tracers were measured in an ice core covering part of the 20th
century from the Swiss Alps .
In this study, VA and p-HBA were measured in an ice core from the Eurasian
Arctic (80∘31′ N, 94∘49′ E). The 724 m long ice core
was drilled on the Akademii Nauk ice cap (5575 km2, 800 m above sea
level) on the Severnaya Zemlya archipelago
Fig. ;. This is the longest ice core record of these aromatic
acids measured to date, with samples ranging in age from 650 BCE to 1988 CE.
Location of Akademii Nauk ice core drilling site
(80∘31′ N, 94∘49′ E).
Analysis of vanillic acid, para-hydroxybenzoic acid, and syringic
acid using ion chromatography with electrospray ionization and tandem mass
spectrometry. Left: standards; right: Akademii Nauk ice core sample (219 m,
1450 CE) containing 0.288 ppb vanillic acid and 0.545 ppb
para-hydroxybenzoic acid. Syringic acid was not detected in the sample.
Results
Analytical results and data processing
In this study, VA concentrations are reported for 3294 Akademii Nauk ice core
samples, and 2585 of these were also analyzed for p-HBA (Fig. ;
Grieman et al., 2016). The instrument was originally optimized to analyze VA.
Several samples were analyzed for VA before a method was developed to analyze
p-HBA. In addition, a subset of 1074 Akademii Nauk samples were analyzed as
replicates for vanillic acid using the older HPLC-ESI/MS/MS method described
by . The results indicate no bias between the techniques and
illustrate the improved sensitivity (∼ 10 times better signal-to-noise
ratio) of the IC-based method (Fig. S3). For the remainder of this paper,
only the IC-ESI/MS/MS results are discussed.
Akademii Nauk vanillic acid (top) and para-hydroxybenzoic acid
(bottom) ice core records.
VA and p-HBA were below detection in 56 and 76 % of the samples,
respectively, and the frequency distributions of both compounds were skewed
towards lower concentrations. Skewness is expected because burning is
episodic and spatially heterogeneous, and atmospheric levels of burning
aerosols are highly enriched during those episodes. The distribution was
successfully normalized using a logarithmic transformation. Outliers were
excluded from the entire dataset prior to transformation. Outliers were
defined as VA and p-HBA levels outside of ±2σ from the median (0.0075
and 0.021 ppb, respectively). This process excluded 0.43 % of the VA
data and 2.6 % of the p-HBA data. For the entire Akademii Nauk ice core,
the geometric means of VA and p-HBA were 0.0087-0.0071+0.037 and
0.019-0.014+0.048 ppb (±1σ), respectively.
A total of 960 samples were analyzed for syringic acid. Only 0.21 % of
these samples were above the detection limit, despite the fact that the
detection limit of the instrument for syringic acid was similar to that of
the other aromatic acids. Standard additions of syringic acid to ice core
samples at concentrations comparable to the ambient levels of VA and p-HBA
were recovered quantitatively, indicating that there was no suppression of
signal due to matrix effects. The low levels of syringic acid suggest that it
was either (1) not generated at the biomass burning sources that impact the
Akademii Nauk ice core, (2) chemically lost from the aerosols during
transport, or (3) degraded in the ice core after deposition. Syringic acid is
similar in molecular structure to the other two compounds, and it does not
differ greatly in terms of volatility or reactivity. It is therefore most
likely that syringic acid was not generated at the biomass burning sources.
Syringic acid is structurally related to the lignin commonly found in
grasses, including tundra grasses, and its absence in the ice core may simply
indicate that grasses were not a significant component of the parent fuels
. Laboratory studies of biomass burning indicate that syringic
acid is not a component of burning-derived aerosols from conifers
. We are not aware of any
laboratory combustion studies of plant species typical of Siberian forests or
tundra.
Akademii Nauk vanillic acid (top) and para-hydroxybenzoic
acid (bottom) ice core records. Individual measurements are shown as
gray dots. The color-filled lines
are exponentials of ±1 standard errors of 40-year bin averages of the
log-transformed data. The solid horizontal lines represent the 75th
percentile of each dataset. The vertical gray shaded areas are periods of
elevated vanillic acid or para-hydroxybenzoic acid, identified as
periods when the bin-averaged data are in the upper quartile of the
transformed dataset.
The Akademii Nauk vanillic and para-hydroxybenzoic acid time series
The raw time series of VA and p-HBA exhibit broadly similar patterns, each
showing several multi-century periods of elevated levels (Fig. ).
Throughout the preindustrial late Holocene (prior to 1700 CE), the levels of
VA generally are higher than those of p-HBA. The median and mean of the ratio
of VA to p-HBA prior to 1700 CE were 1.4 and 4.4, respectively. Elevated
levels of both compounds also occur during the industrial period (after
1700 CE), but here the levels of p-HBA exceed those of VA. There are also
numerous smaller multi-decadal features in both records, as well as higher
frequency (sub-annual and inter-annual) variability throughout the ice core.
The Akademii Nauk site experiences summertime surface melting and
infiltration, which results in redistribution of soluble compounds of over
1 m of surface snow roughly 2–3 annual layers;. The distribution of melt layers in the core over the last 500 years
does not correlate with aromatic acid levels and is not likely to be
responsible for the major features in the record (Fig. S4). Due to this
disturbance in annual layering, we focus exclusively on multi-decadal and
longer timescales. To remove short-term variability, smoothed records were
constructed using log-transformed 40-year bin-averaged VA and p-HBA records.
The exponentials of the smoothed log-transformed records were used to present
the records in concentration units (Fig. ). Smoothing was also
carried out using locally weighted polynomial regression
LOESS;. Bin-averaged and LOESS smoothing of the
Akademii Nauk organic acid records show essentially the same centennial- to
millennial-scale features that are clearly visible in the raw data (Fig. S5).
As evident from visual inspection of the bin-averaged record, VA and p-HBA
are correlated (r2=0.47, p<10-6, n=80; Fig. ). The
similarity in VA and p-HBA suggests that the two compounds are derived from a
common source and/or are modulated by similar depositional/post-depositional
processes.
Comparison of vanillic acid and para-hydroxybenzoic acid records.
Linear fit is the 40-year bin-averaged log transform of vanillic acid against
the 40-year averaged log transform of para-hydroxybenzoic acid.
For the purposes of this study, we define elevated periods as including data
for which the standard error bands for 40-year binned averages reach or
exceed the upper quartile of the entire dataset (Fig. ). The
start/stop dates for these time ranges each have an uncertainty of
±20 years due to the bin averaging of the data. These time ranges are
highly uncertain prior to 743 CE due to the high uncertainty of the age
scale. Three periods of elevated VA are identified (650–300 BCE,
340–660 CE, and 1460–1660 CE), all of which are shared by p-HBA. p-HBA
has an additional period of elevated levels from 1780 to 1860 CE that is not
shared by VA. The levels of aromatic organic acids during this elevated
period are enriched manyfold over the intervening “quiet” periods. This
large dynamic range is quite different from the enrichment patterns typically
found for inorganic ice core burning tracers such as ammonium and nitrate,
where relatively small burning signals are superimposed on a large background
from other natural sources such as biogenic emissions or lightning
. We interpret these
large pulses of aromatic acids in the Akademii Nauk ice core as evidence of
multi-century periods of enhanced deposition of biomass-burning-derived
aerosols.
A period of elevated p-HBA levels is identified from 1780 to 1860 CE, using
40-year bin averaging of the entire dataset. This period is qualitatively
different from the other elevated intervals in that p-HBA is more abundant
than VA. The log-transformed dataset (after 1700 CE) was 10-year bin-averaged to determine shorter-term elevated periods during the industrial
period (1750–2000 CE). Elevated intervals during the industrial period are
defined as periods during which the standard error bands of the 10-year
binned averages reach or exceed the upper quartile of the dataset after
1700 CE (Fig. ). VA and p-HBA are both elevated from
1860 to 1900 CE, using this definition. p-HBA is additionally elevated from
1770 to 1790 CE. The start and end dates of these periods each have an
uncertainty of ±5 years due to the application of bin averaging.
Akademii Nauk vanillic acid (top) and para-hydroxybenzoic
acid (bottom) ice core records. Individual measurements are shown as
gray points. The color-filled lines
are exponentials of ±1 standard errors of 10-year bin averages of the
log-transformed data. The solid horizontal lines represent the 75th
percentile of each dataset (after 1700 CE). The vertical
gray shaded areas are periods of
elevated vanillic acid or para-hydroxybenzoic acid, identified as
periods when the bin-averaged data are in the upper quartile of the
transformed dataset.
VA and p-HBA may be subject to post-depositional revolatilization or
photochemical destruction . If post-depositional processes
were a significant factor in determining the trends in aromatic acid levels
in the ice core, one might expect to see a strong relationship with
accumulation rate or with ice chemistry. There is no evidence of large
changes in accumulation that can explain the large concentration changes
apparent in the VA and p-HBA measurements. The age-depth relationship for
Akademii Nauk indicates constant average accumulation rate from
1700 to 1999 CE (140–0 m depth; Fig. S2). Below 140 m, the age-depth curve
varies smoothly in a manner consistent with thinning due to ice flow at
relatively constant accumulation . Similarly, there are no
obvious correlations between major ion chemistry (sea-salt-derived Na+,
terrestrial-derived Ca2+, or volcanic S) and the aromatic acid levels
(Fig. S6). One could perhaps argue that acidification of aerosols and/or ice
during the past century was responsible for the decline in p-HBA levels
around 1900 CE due to revolatilization (Fig. ). However, large
volcanic sulfate peaks throughout the record do not exhibit evidence of loss
of aromatic acids. Given the absence of indication of any postdeposition
artifacts, we interpret ice core VA and p-HBA as tracers for biomass burning
variability. Further investigation of such effects is warranted, particularly
for low-accumulation ice core sites. Future studies should examine the
relationship between levels of aromatic acids in air and snow in order to
develop transfer functions following the method of .
Discussion
Potential source regions and vegetation types – air mass back trajectories
The possible source locations for biomass burning impacting the Akademii Nauk
ice core site were identified by calculating the fraction of air mass back
trajectories from the ice core site originating in or passing over Siberia
(defined as east of 42∘ E), Europe (defined as west of
42∘ E), or North America. This analysis assumes present-day
meteorological conditions. Changes in atmospheric circulation patterns over
the past millennia may have occurred, but these are not considered here. Each
region was subdivided into ecofloristic zones defined by the Food and
Agriculture Organization Fig. S7;
http://cdiac.ornl.gov/epubs/ndp/global_carbon/carbon_documentation.html;.
Air mass back trajectories were computed using the HYSPLIT model
. Ten-day back trajectories
from the ice core site (80∘ N, 94∘ E) were started at
100 m above ground level, at 00:00 and 12:00 local time (UTC +7 h) each day for spring (trajectories beginning 1 March–31 May), summer
(trajectories beginning 1 June–31 August), and fall (trajectories beginning
1 September–30 November). NCEP/NCAR Reanalysis data from 2006 to 2015 CE
were used
ftp://arlftp.arlhq.noaa.gov/pub/archives/reanalysis;.
The results indicate that Siberia is the most likely source region to the
Akademii Nauk ice core site, with most of the trajectories either originating
in or transecting this region (Table ; spring 61 %, summer
28 %, and fall 60 %). Siberian boreal tundra woodlands, boreal
coniferous forests, and boreal mountain systems all contributed significantly
(Fig. S8). There were some seasonal differences, with more trajectories
originating from or passing over these areas in the spring and fall than in
the summer. All of the other ecofloristic zones in Siberia intersected fewer
than 5 % of the trajectories. Similarly, for these 10-day back
trajectories, all of the ecofloristic zones in Europe and North America
contributed less than 3 % for all three seasons. This analysis does not
prove that burning emissions from Europe and North America could not
contribute to the ice core signals, but that they would require significantly
longer atmospheric transport times.
Fractions of air mass back trajectories originating from or
intersecting various ecofloristic zones and geographic regions (% rounded
to the nearest integer). Ecofloristic zones are defined by the Food and Agriculture
Organization Fig. S7;
http://cdiac.ornl.gov/epubs/ndp/global_carbon/carbon_documentation.html;.
Season
Geographic region
Ecofloristic zone
Spring (%)
Summer (%)
Fall (%)
Siberia
Boreal tundra woodland
36
7
40
Boreal coniferous forest
23
3
32
Boreal mountain system
17
2
22
Total
61
28
60
Europe
Total
<1
2
<1
North America
Total
11
18
8
Both the trajectory analysis and the VA / p-HBA ratio in the ice core are
consistent with Siberian conifer forests and tundra woodlands as main sources
for the three major preindustrial burning peaks in the Akademii Nauk record.
Laboratory studies show that burning conifers results in higher or similar
yields of VA to p-HBA .
Pine wood burned in a fire in northern Alberta, Canada, yielded a VA / p-HBA
ratio of 27:4 µg g-1 carbon . European pine
combustion yielded a VA / p-HBA ratio of 14:1.6 mg kg-1 of fuel burned
. Burning of ponderosa pine, Sitka spruce, and Douglas fir
yielded VA / p-HBA ratios of 790:40, 2194:2968, and
3441:4345 µg/kg of fuel burned, respectively .
Timeline of elevated burning periods in Northern Hemisphere ice core
studies. From top: Akademii Nauk vanillic acid and
para-hydroxybenzoic acid (this study); Belukha glacier nitrate,
potassium, and charcoal ; NEEM levoglucosan and black carbon
; NEEM and Summit ammonium ; and
Kamchatka Peninsula para-hydroxybenzoic acid, vanillic acid,
dehydroabietic acid, and levoglucosan . Lines indicate
periods of elevated burning. Gray
bars mark the time range analyzed in each core.
In the burning peak after 1700 CE, the levels of p-HBA are higher than VA.
This likely indicates a change in the type of vegetation burned. This could
reflect either a change in ecosystem at the source region, a shift in the
location of burning, or a change in atmospheric transport pattern. Laboratory
studies indicate that tundra grass fires yield high p-HBA-to-VA ratios. We
therefore speculate that Siberian tundra fires or boreal peat fires may have
contributed to high levels of aromatic acids during this period
.
The Akademii Nauk aromatic acid record suggests that high biomass burning
emissions were sustained for multi-century periods during the last 2600 years
of the Holocene. This result perhaps indicates that the fires were
widespread, but of relatively low intensity, consistent with the fact that
low-intensity ground fires are the principle mode of burning in Eurasian
boreal forests today. Model results suggest that ground fires dominated the
eastern Siberian region over the past 1200 years . By contrast,
North American boreal forest burning occurs predominantly by high intensity,
stand-replacing crown fires .
Aromatic acids should be measured in ice cores from North America to examine
whether the fire conditions are reflected in the pattern of centennial-scale
variability.
Comparison to other biomass burning proxy records
There are few records of Siberian biomass burning covering the past
2600 years, and none of the existing records show all three of the prominent
periods of increased fire activity present in the Akademii Nauk record
(Fig. ). Measurements of nitrate, potassium, and charcoal in a
750-year ice core from southern Siberia suggest elevated burning from
1600 to 1680 CE . This period overlaps the most recent of the
major preindustrial peaks in the Akademii Nauk VA record (1460–1660 CE). A
number of organic biomass burning tracers were measured in a 300-year ice
core from the Kamchatka Peninsula, northeastern Asia . In that
study, elevated p-HBA, VA, dehydroabietic acid, and levoglucosan were
observed during the periods from 1700 to 1800 and 1880 to 2000 CE. Akademii Nauk
p-HBA and VA are similarly elevated from 1770 to 1790 (p-HBA only) and
1860–1900 CE. VA and p-HBA remain elevated late in the 20th century in the
ice core from the Kamchatka Peninsula. It is not possible to determine
whether
Akademii Nauk VA and p-HBA also increase during this period due to limited
Akademii Nauk sample availability after 1970.
Historical changes in biomass burning have also been reconstructed from
ammonium spikes in the NEEM and Summit Greenland ice core records. Variations
in the frequency of ammonium peaks for the past 1000 years suggest elevated
burning from 1200 to 1500 CE and low burning from 1600 to 1800 CE
. This pattern of preindustrial burning is
different from the Akademii Nauk record, which is not surprising given that
Greenland is primarily influenced by transport from North America, rather
than Eurasia. These trends are generally consistent with charcoal records
from northeastern Canada . presented a
2000-year NEEM record of levoglucosan and black carbon. They show four
preindustrial maxima in levoglucosan around 100 BCE–100 CE, 200–600 CE,
1000–1200 CE, and 1500–1700 CE. The last of these maxima is strongest and
coincident with the 1400–1600 peak in Akademii Nauk aromatic acids.
Interestingly, the same feature is not the largest peak for NEEM black
carbon. There are clearly unresolved differences between various Greenland
ice core proxy records, particularly for the period around 1500 CE.
As noted earlier, there are only 11 charcoal records from the Siberian region
in the Global Charcoal Database . Of these records, only
the record from Bolshoe Bog in the Lake Baikal region of southern Siberia has
sufficient temporal resolution that allows comparison with the Akademii Nauk
record. The Bolshoe Bog record does exhibit similar timing of elevated levels
to Akademii Nauk (Fig. ), suggesting a common biomass burning
source region between the two records. Further charcoal studies throughout
Siberia are needed to assess the full range of biomass burning source regions
contributing to aromatic acids in the Akademii Nauk ice core.
Comparison to climate proxy records
Variability in biomass burning can be caused by human activity. The effect of
humans on Siberian wildfire activity is not well established. Human
civilizations were predominantly nomadic in Siberia prior to 16th century.
Comparison between pollen records, civilization development, and climate in
the Lake Baikal region suggests that vegetation changes were more likely
linked to climate than human-induced land use change throughout the Holocene
.
Variability in regional biomass burning generally is driven by changes in
temperature and precipitation, which are linked to atmospheric circulation
patterns. Over recent decades, Siberian wildfire burned area correlates with
changes in the Arctic Oscillation, with increased biomass burning during the
positive phase of the Arctic Oscillation when Siberian summers are warmest
. An 8000-year Holocene proxy record of
Arctic Oscillation shows a 1500-year cycle , but it is not
synchronous with increased biomass burning in the Akademii Nauk ice core.
This sedimentary proxy evidence does not support the Arctic Oscillation as
the primary mode of climate variability controlling Siberian burning on
millennial timescales.
concluded that the period of elevated burning recorded in
the Belukha glacier was preceded by a drought event that was possibly related
to the positive phase of the Pacific Decadal Oscillation (PDO). The period of
elevated VA and p-HBA from 1460 to 1660 CE also overlaps a positive phase of
the PDO reconstructed using tree ring chronologies . A
longer record of the PDO is needed to determine whether the other peaks in the
Akademii Nauk VA and p-HBA records follow this pattern.
also relate the period of elevated burning in the NEEM Levoglucosan record
from 1500 to 1700 CE to drought conditions. They link Asian drought conditions
to monsoon failures during the 16th and 17th centuries. This variability may
also be related to the PDO, given that the PDO can modulate the summer
monsoon .
Climate reconstructions based on Northern Hemisphere proxy records show a
long-term cooling trend over the past 2000 years
Fig. ;. Centennial-scale climate variability, most notably the
Medieval Climate Anomaly (830–1100 CE) and Little Ice Age (1580–1880), is
superimposed on this trend .
Akademii Nauk VA and p-HBA levels do not exhibit a trend following the
long-term cooling trend in temperature, but they do appear to correlate with
some centennial-scale climate variability. Akademii Nauk VA and p-HBA
levels are elevated from 340 to 660 CE, prior to and during the Late Antique
Little Ice Age 536–660 CE;. The Akademii Nauk
aromatic acids are low during the early part of the Medieval Climate Anomaly
(prior to 1050 CE). They are slightly elevated during the latter part of the
Medieval Climate Anomaly. Tree ring reconstructions suggest that the Medieval
Climate Anomaly was humid in northern Siberia . Akademii
Nauk VA and p-HBA are elevated from 1460 to 1660 CE during the Little Ice
Age. The Akademii Nauk aromatic acid trends are different from those in
composite Northern Hemisphere sedimentary charcoal records, which show an
overall decline over the past 2000 years, with a maximum during the Medieval
Climate Anomaly and minimum during the Little Ice Age .
Comparison of the timing of aromatic acid signals in the Akademii
Nauk ice core over the past 3000 years compared to other climate-related
proxy records. From top: 40-year bin-averaged (violet fill is 1 standard
error of log transform) Akademii Nauk ice core vanillic acid measurements
from this study, 40-year bin-averaged (green fill is 1 standard error of log
transform) Akademii Nauk ice core para-hydroxybenzoic acid
measurements from this study, Bolshoe bog charcoal record ,
30-year medians of domain areas of PAGES 2k temperature reconstructions
blue; and 20-year means of zonal 30–90∘ N
stacked temperature reconstruction black;, North
Atlantic ice-rafted debris indicating Bond events (blue < mean;
red > mean) , and smoothed Dongge Cave climate
record from Asia showing changes in the monsoon using a moving average
(blue < 0, red > 0; window size = 15)
. The vertical gray
shaded areas are periods of elevated vanillic acid and
para-hydroxybenzoic acid, identified as periods when the
bin-averaged data are in the upper quartile of the transformed dataset. The
red lines are the Medieval Climate Anomaly (MCA) and the Little Ice Age
(LIA).
The Little Ice Age is the most recent in a series of Holocene cooling events
known as “Bond events” . Bond events are episodes of
increased ice rafted debris in North Atlantic sediment cores throughout the
Holocene at intervals of 1470 ± 500 years with durations from
200 to 500 years. Bond events may be the result of a combination of
∼ 1000-year and ∼ 2000-year cycles of climate variability in the
Holocene . The three most recent Bond events were centered
at 2800, 1400, and 500 years before present 850 BCE, 550 CE, and
1450 CE; Fig. ;. The two most recent major periods of
increased Siberian burning found in the Akademii Nauk ice core are similar in
timing to the two most recent Bond events. The earliest major peak in the
Akademii Nauk ice core is later than the Bond event centered at 850 CE. This
difference in timing could be due to the uncertainty in the Akademii Nauk ice
core age scale. Simultaneous changes in climate also are observed in the
Chinese speleothem record from Dongge Cave, centered at 2700, 1600, and
500 years before present 750 BCE, 350 CE, and 1450 CE; Fig. 8,
bottom plot;. The speleothem record shows three 100–500 year
periods of increased δ18O, indicating decreased East Asian summer
monsoon intensity .
The East Asian winter monsoon is affected by changes in the intensity of the
Siberian High . Comparisons between temperature and
precipitation records between 1922 and 1999 CE show that when the Siberian
High became stronger, temperatures and precipitation over the Eurasian
continent decreased . Drier conditions resulting from
variability in the Icelandic Low and the Siberian High could have altered
biomass burning. The long-term variability in the Siberian High is not
well established , but GISP2 ice core sea salt Na (ssNa) and
non-sea-salt K (nssK) records indicate that the Icelandic low (ssNa) and the
Siberian high (nssK) were stronger than mean levels when VA and p-HBA were
elevated between 1460 and 1660 CE . The Akademii Nauk data
support the suggestion by of a link between Northern
Hemisphere boreal fires and monsoon weakening. The similarity in timing
between the Siberian biomass burning pulses, the Bond events, and the
monsoonal changes likely suggests a link in this region between fires and
large-scale climate variability on millennial timescales.