In the Alaskan Arctic, rapid climate change is increasing the frequency of
disturbance including wildfire and permafrost collapse. These pulse
disturbances may influence the delivery of dissolved organic carbon (DOC) to
aquatic ecosystems, however the magnitude of these effects compared to the
natural background variability of DOC at the watershed scale is not well
known. We measured DOC quantity, composition, and biodegradability from 14
river and stream reaches (watershed sizes ranging from 1.5–167 km
As the Arctic warms, the biogeochemical signature of its rivers and streams
will likely be an indicator of the response of aquatic and adjacent
terrestrial ecosystems to climate change (Holmes et al., 2000; McClelland et
al., 2007; Frey and McClelland, 2009). Arctic freshwater ecosystems process
and transport substantial loads of dissolved organic carbon (DOC) delivering
34–38 Tg yr
Disturbance in arctic and boreal ecosystems is expected to escalate in response to future changes in climate. Examples of physical responses to climate change in northern Alaska include the deepening of the seasonally thawed surface soil or active layer (Shiklomanov et al., 2010), permafrost warming (Romanovsky et al., 2002, 2011), permafrost collapse (Jorgenson et al., 2006; Belshe et al., 2013; Balser et al., 2014), and wildfire (Randerson et al., 2006). There is evidence of recent increases in permafrost disturbance (Gooseff et al., 2009; Balser et al., 2014) on the North Slope of Alaska and wildfire has the potential to become a major disturbance factor in the tundra region (Higuera et al., 2011; Rocha et al., 2012).
Thaw of ice-rich permafrost results in soil collapse or subsidence, termed thermokarst (Kokelj and Jorgenson, 2013). Thermokarst can export substantial quantities of sediment, carbon, nitrogen, and phosphorus to Arctic streams, rivers and lakes (Kokelj et al., 2005, 2009, 2013; Bowden et al., 2008; Lamoureux and Lafrenière, 2009; Lewis et al., 2011; Dugan et al., 2012; Malone et al., 2013; Harms et al., 2014). The magnitude of exported material depends largely on thermokarst size, type, activity, and hydrologic connectivity (Lewis et al., 2011; Lafrenière and Lamoureux, 2013; Abbott et al., 2014). For example, thermokarst features can mobilize substantial amounts of sediments and nutrients that are not delivered to downslope aquatic ecosystems and instead retained along the hillslopes or in the riparian zone (Larouche, 2015). DOC in the outflow of thermokarst features is highly labile (Woods et al., 2011; Vonk et al., 2013; Abbott et al., 2014), particularly when exposed to light (Cory et al., 2013). While sediment and solute concentrations and the proportion of BDOC can be high in thermokarst outflow, the impact on the watershed depends on the total mass flux or load (Lewis et al., 2012). The effects of thermokarst disturbance on Arctic aquatic ecosystems are poorly understood at the watershed scale, limiting useful inferences about future system response to climate change.
The organic horizon of tundra soils insulates permafrost from warm summer air temperatures. The removal of surface soil carbon during fire promotes underlying permafrost degradation (Burn, 1998; Yoshikawa et al., 2002), increases thaw depth for decades post-fire (Rocha et al., 2012), and triggers thermokarst development (Osterkamp and Romanovsky, 1999). Wildfire disturbance in lower latitude ecosystems can increase concentrations of major ions and nutrients in soil and stream water (Bayley et al., 1992a, b; Chorover et al., 1994). In the boreal forest of Alaska, stream DOC concentration declined following a wildfire, presumably due to loss of microbial biomass (Schindler et al., 1997; Petrone et al., 2007; Betts and Jones, 2009) and bioavailable dissolved organic matter in streams decreased post-fire and during thermokarst formation (Balcarczyk et al., 2009).
Across various biomes the composition and biodegradability of riverine DOC changes seasonally due to a tight coupling between terrestrial and aquatic ecosystems (Holmes et al., 2008; Fellman et al., 2009; Wang et al., 2012). In the Arctic, DOC concentration and biodegradability is highest during snowmelt and early spring and decreases progressively through the summer (Holmes et al., 2008; Mann et al., 2012; Vonk et al., 2013). However, the majority of studies investigating Arctic BDOC have focused on downstream reaches in large alluvial systems leaving the seasonal and spatial variation of BDOC in headwater streams largely unknown.
The questions we address in this paper are, “Does BDOC and water chemistry differ at the watershed scale among landscape types?” and “Does BDOC and water chemistry differ in streams impacted by thermokarst and fire?” To answer these questions we measured the quantity, biodegradability, and aromaticity of DOC and background water chemistry from Arctic headwater streams and rivers. We sampled watersheds in three geographic regions affected by a combination of fire and thermokarst to evaluate controls on DOC quantity and biodegradability at the watershed scale. We hypothesized thermokarst would increase DOC concentrations and BDOC due to the delivery of labile carbon from thawed permafrost. Because wildfire in the Arctic can directly impact DOC export, as well as have secondary effects due to changes in active layer depth and extent of permafrost, we hypothesized that wildfire may decrease BDOC due to the combustion of soil carbon stocks during fire. However, if wildfire promotes extensive permafrost degradation and thermokarst production then BDOC concentrations might increase.
We took advantage of natural disturbance to test our hypotheses. We
collected stream water from 16 reaches, 11 of which were individual Arctic
rivers and streams on or near the North Slope of Alaska including the
regions around the Toolik Field Station, Feniak Lake, and the Anaktuvuk
River wildfire area (Fig. 1, Table 1). Seven of the stream sites were
apparently undisturbed (reference) reaches and nine sites were impacted by a
combination of wild fire and thermokarst of various types, including
retrogressive thaw slumps and active layer detachment slides, two of the
most common thermokarst morphologies in upland landscapes (Kokelj and
Jorgenson, 2013). The Toolik Field Station is located 254 km north of the
Arctic Circle and 180 km south of the Arctic Ocean. The average annual
temperature is
Watershed characteristics of sampling sites.
Continued.
In the summer of 2007 in the Anaktuvuk River area, above-normal
temperatures, below-normal precipitation, and extremely low soil moisture
conditions favored fire conditions when a lightning strike ignited the
tundra on 16 July. Air temperature in July to September of that year was the
warmest over a 129-year record, with a
Map of study areas. Map credit: J. Noguera and R. Fulweber, Toolik Field Station GIS and Remote Sensing Facility.
In 2011, we sampled reference streams and streams impacted by thermokarst
and wildfire near the Toolik Field Station, the Anaktuvuk burn scar, and
Feniak Lake. In the Toolik area we sampled the Kuparuk River (Site 1) and
Oksrukyuik Creek (Site 2), both of which have not been impacted by fire or
thermokarst. In the Anaktuvuk area, we sampled four reference rivers on 6
August 2011, two of which we analyzed for BDOC (Burn Reference 1
To quantify seasonal variability of BDOC we took repeat measurements 4-5
times over the 2011 summer season from the Toolik and Anaktuvuk stream
sites, except for the two Anaktuvuk reference sites that were sampled once
(Sites 3 and 4). Due to their remote locations, sites located in the Feniak
Lake area were sampled once during 2011. At each stream site, we collected
four replicate field samples, which we filtered (0.7
We followed the BDOC incubation protocol described in Abbott et al. (2014).
In brief, we amended all samples with nutrients (increasing ambient
concentrations by 80
We characterized DOC composition by specific ultraviolet absorbance at 254
nm (SUVA
We analyzed water samples for total suspended sediment (TSS, mg L
The variance values around all mean values reported below are standard
errors (SE). We tested for differences in BDOC metrics and background water
chemistry variables among streams within groups defined a priori by analysis of
variance (ANOVA). Significant differences between streams (
Anaktuvuk reference sites had higher initial DOC concentrations (977
Assessing the impact of thermokarst on stream DOC quantity
The absolute BDOC concentrations in Reference Feniak streams (125.2
The values of SUVA
Assessing the impact of region (regardless of treatment) on stream
DOC quantity
Most background water chemistry variables differed significantly among
regions (Fig. 4). Stream alkalinity was approximately five-fold higher in
the Feniak streams (1734
We compared background water chemistry between the Anaktuvuk reference sites
(from the opportunistic sampling of sites 3 and 4 on 06 August 2011) and
burned sites using data only from that date (data not shown). We found that
NH
Biodegradability of DOC did not change significantly over time in five of the eight streams from which repeat measurements were taken (Fig. 5a). The pattern in DOC biodegradability across the season differed among the three alluvial streams. BDOC % from samples obtained from the Kuparuk River (Site 1) and South River (Site 5) increased (Fig. 5b). In contrast, BDOC % from samples obtained from Oksrukyuik Creek (Site 2) decreased as the season progressed (Fig. 5c).
Contrary to our hypothesis, we found that streams disturbed by thermokarst and fire did not contain significantly altered labile DOC fractions compared to adjacent reference waters. The quantity, composition and biodegradability of riverine DOC sampled in this study differed primarily by region, likely driven by unique landscape and watershed characteristics (e.g. lithology; soil and vegetation type; elevation; and glacial age). Watershed characteristics influence ecological patterns by controlling the chemistry of soils (Jenny, 1980); plants (Stohlgren et al., 1998); water (Hynes, 1975); and microbial community composition (Larouche et al., 2012). Thus, it is not surprising to observe differences in DOC quantity and character across the three different regions sampled. A circumboreal study across diverse watersheds found that DOC loadings also varied by region (i.e. extent of permafrost and runoff; Tank et al., 2012). The range of BDOC % from streams and rivers measured in this study (4–46 %) is similar to other studies of Arctic riverine BDOC (< 10–40 %; Wickland et al., 2007; Holmes et al., 2008; Mann et al., 2012).
Our study tested for differences in DOC quantity and biodegradability across three geographic regions for headwater stream reaches disturbed by fire and thermokarst. DOC in thermokarst outflow is highly biodegradable (Woods et al., 2011; Cory et al., 2013; Vonk et al., 2013), though biodegradability returns to pre-disturbance levels once features stabilized (Abbott et al., 2014). Two potential explanations for the lack of thermokarst impact in this study are the relatively small portion of the watersheds occupied by thermokarst and the fact that the receiving streams were relatively large (2nd and 3rd order, in the case of Twin 1 and 3 in the Feniak region), diluting highly labile DOC exported from thermokarst at the watershed scale. The two comparisons of the Valley of Thermokarst Reference watershed vs. the Impacted in the burned landscape also did not show an expected impact attributed to the presence of stabilized active layer detachment slides. In this case, the lack of physical and hydrologic connectivity between the slides on the south-facing hillslope and the stream valley bottom, and the rapid stabilization of the features may explain the lack of a watershed-scale influence. Approximately 2–3 years had passed since active layer detachment slide initiation when we sampled for BDOC. Moreover, 2011 was a particularly dry summer season with few storm events resulting in limited hydrologic connectivity between disturbed surfaces and the stream. A study in the High Canadian Arctic also concluded that seasonal solute export from watersheds disturbed by thermokarst (disturbed watershed areas range from 6–46 %) were more sensitive to increased soil temperatures and rainfall events than to the presence of active layer detachments (Lafrenière and Lamoureux, 2013).
Cory et al. (2014) concluded that DOC in thermokarst outflow, with little
prior exposure to light is > 40 % more susceptible to microbial
conversion to CO
The typical post-burn biogeochemical signal that many have found in lower latitude ecosystems may not manifest in burned Arctic watersheds due to the added complexity of permafrost dynamics that also change due to fire. Monitoring and modeling efforts in the terrestrial system of the Anaktuvuk River Fire scar suggest that tundra surface properties (e.g. greenness, albedo, thaw depth) appear to recover rapidly post-fire (Rocha et al., 2012). DOC quantity and biodegradability may have been altered immediately after the tundra burned but our sampling 4 years post-fire may have missed the initial response to fire.
We originally planned for the Toolik river sites (Kuparuk and Oksrukyuik) to
be the reference sites for the burned streams. Had we not opportunistically
sampled the two sites north of the burn boundary or the sites in the Feniak
region, we may have attributed differences in water chemistry to fire
disturbance rather than watershed characteristics. Even though we detected
no effect of fire and thermokarst on BDOC, we had a limited sample size and
therefore low power in making this statistical conclusion. We conclude that
water chemistry differs significantly by region (Fig. 4), regardless of
disturbance. However, when we compare the Anaktuvuk reference sites to the
east of the burn boundary with the sites within the burned area from a
single sampling date on 06 August 2011 (the only date we were able to sample
reference sites outside of the burned boundary) we found significant
differences in water chemistry (i.e. higher DOC, NH
Biogeochemical characteristics of streams within each region
(includes all available data, not just from BDOC sampling sites/dates). Box
plots represent median, quartiles, minimum and maximum within 1.5 times the
interquartile range, and outliers beyond 1.5 IQR. Different letters
represent significant differences between regions,
Landscape age and associated ecosystem differences may explain the
differences in BDOC we observed. The Anaktuvuk landscape is substantially
older (> 700 ka) than the younger surfaces of Toolik (10–400 ka)
and Feniak (50–80 ka). Older landscapes can have deeper soil organic layers
with more decomposed soil organic layers (Hobbie and Gough, 2004),
potentially resulting in lower DOC biodegradability in streamwater.
Elevation could also play a role with warmer air and soil temperatures
accelerating decomposition in the Anaktuvuk (285
The three study areas had distinct vegetation communities (Table 1) and the concentration and characteristics of streamwater DOC differ according to its source (McDowell and Likens, 1988). We found that Anaktuvuk stream samples contained high DOC concentration of low biodegradability and that the area sampled (i.e. in the southern area of the burn scar) likely receives allochthonous inputs from moist acidic tundra (MAT) communities (Jorgenson et al., 2010). Conversely, Feniak streams, which receive allochthonous inputs from moist non-acidic tundra (MNAT; Jorgenson et al., 2010), contained low DOC concentration of high biodegradability. In general, the rivers in the Toolik area contained low DOC concentration of a relatively recalcitrant form. Thermokarst features draining MNAT have higher BDOC compared to MAT, perhaps due to accelerated decomposition of dissolved organic matter from higher N availability in acidic tundra before reaching the stream (Abbott et al., 2014). Thus, the MNAT vegetation type in the Feniak area may explain its high BDOC %.
Arctic rivers and streams are generally high in dissolved organic matter and
low in inorganic nutrients (Dittmar and Kattner, 2003). Although there is
little evidence for nutrient limitation of DOC degradation, background
dissolved inorganic N concentrations were positively correlated with BDOC %
in thermokarst outflow (Abbott et al., 2014). Feniak streams also tend
to have higher concentrations of NH
Another important consideration is DOC adsorption reactions and the role of suspended sediment and exposed mineral soils in permafrost areas disturbed by thermokarst and/or fire. In this study we found some of the streams in the Feniak region (those that were impacted by thermokarst) contained high concentrations of total suspended sediment. Feniak streams had lower DOC concentration, but DOC was more biodegradable. This pattern is consistent with observations of low DOC concentration following thermokarst and fire in lakes, likely due to adsorption of DOC to exposed mineral soil and suspended mineral soil particles (Kokelj et al., 2005). The input of suspended mineral soil from thermokarst can result in greater transparency in the water column compared to undisturbed lakes (Kokelj et al., 2009) due to adsorption of organic material by mineral sediment from slumps that adsorbs DOC and then settles to the lake bottom (Thompson et al., 2008).
Seasonal trends in BDOC (%):
Contrary to several studies showing highest BDOC during snowmelt, followed by a decrease through the growing season (Holmes et al., 2008; Spencer et al., 2008; Mann et al., 2012; Raymond et al., 2007), we found variable seasonal patterns of BDOC. The majority of these studies are in larger, arctic river systems whereas our study sampled 1st and 2nd order headwater streams. Stream morphology may also play a role since beaded streams are made up of ice-rich polygons that may contain older forms of DOC and are typically colder compared to alluvial systems (Brosten et al., 2006). Thermo-erosion gullies, a common upland thermokarst type, often form from the thaw of ice-rich polygons and the outflow from gullies contained the least biodegradable DOC compared to other feature types, although still elevated compared to reference waters (Abbott et al., 2014). Thus, although polygonal areas are susceptible to thaw via gully formation or beaded stream formation, it is possible that the ice wedges contain low BDOC %. We observed an increase in BDOC % in the Kuparuk River (Site 1) and South River (Site 5), both of which are alluvial systems without any lake influence upstream of the river network (Fig. 5b), whereas we observed a decreasing trend in BDOC % in Oksrukyuik Creek, an alluvial system with a series of lakes upstream of our sampling point (Fig. 5c). In the alluvial streams without lakes, it is likely that after the pulse of labile terrestrial DOC during the freshet (which our study did not sample), tundra plant and in-stream algal productivity increases as the growing season progresses and in-turn increases stream DOC biodegradability as sources shift from allochthonous to autochthonous. We suggest that the lake effect in the Oksrukyuik Creek watershed serves as a reservoir for a pulse of highly labile, aquatic-derived BDOC in the beginning of the growing season, following the flush from the terrestrial ecosystem during the spring freshet. The BDOC in general from the alluvial stream with the lake influence is more labile (BDOC % range 15.7–24.6) compared to the alluvial systems without lakes (BDOC % range 0.75–13.9) as it leaks from the rich lake environment down the watershed, likely seeding the stream with rich material from the lake across the season.
Although active thermokarst outflow contains highly biodegradable DOC (Woods
et al., 2011; Cory et al., 2013; Vonk et al., 2013; Abbott et al., 2014) and
dissolved organic matter biodegradability from boreal soil leachate is lower
from burned than unburned soils (Olefeldt et al., 2013), we found no
significant effect of fire or thermokarst in the streams we sampled. Our
study indicates strong variation of stream water chemistry and DOC quantity,
biodegradability, and aromaticity based on landscape characteristics.
Although elevated concentrations and export of sediment and nutrients from
thermokarst have been documented (Bowden et al., 2008; Kokelj et al., 2009;
Lamoureux and Lafrenière, 2009; Lamoureux and Lafrenière, 2014), the
impact on hydrologic export depends largely on the magnitude and type of
thermokarst disturbance, the time from initial disturbance to stabilization,
and the hydrologic connectivity between the feature and downslope aquatic
ecosystems (Lewis et al., 2011; Shirokova et al., 2013; Thienpont et al.,
2013). Although thermokarst gullies and active layer detachment slides are
the dominant thermokarst types in the area we sampled (e.g.
J. R. Larouche and B. W. Abbott designed the experiment, collected and analyzed samples and collaborated closely on the manuscript written by J. R. Larouche. W. B. Bowden and J. B. Jones advised on the design of the experiment, assisted with the data analysis, and edited the final manuscript.
We thank the many individuals and organizations that assisted with this study. S. Godsey, A. Olsson, L. Koenig, and P. Tobin assisted with laboratory and field work. R. Cory and G. Kling provided technical assistance and advice with DOC analysis. A. Balser and J. Stuckey provided assistance with landscape classification and watershed characteristics and J. Noguera with the Toolik Field Station GIS and Remote Sensing Facility provided the map for this manuscript. We thank the staff of Toolik Field Station and of CH2M Hill Polar Services logistical services and support. Staff from the Arctic Network of the National Park Service and Bureau of Land Management facilitated research permits. This work was supported by the National Science Foundation's Arctic Systems Science Program under grant number ARC-0806394. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. Edited by: W. F. Vincent