Microtopographic and depth controls on active layer chemistry in Arctic polygonal ground

Abstract

Polygonal ground is a signature characteristic of Arctic lowlands, and carbon release from permafrost thaw can alter feedbacks to Arctic ecosystems and climate. This study describes the first comprehensive spatial examination of active layer biogeochemistry that extends across high- and low-centered, ice wedge polygons, their features, and with depth. Water chemistry measurements of 54 analytes were made on surface and active layer pore waters collected near Barrow, Alaska, USA. Significant differences were observed between high- and low-centered polygons suggesting that polygon types may be useful for landscape-scale geochemical classification. However, differences were found for polygon features (centers and troughs) for analytes that were not significant for polygon type, suggesting that finer-scale features affect biogeochemistry differently from polygon types. Depth variations were also significant, demonstrating important multidimensional aspects of polygonal ground biogeochemistry. These results have major implications for understanding how polygonal ground ecosystems function, and how they may respond to future change.

Key Points

• Chemistry was measured in Arctic polygonal ground
• Differences were found for polygon types, features, and depth
• Topographic and depth controls on polygonal ground chemistry are significant

1 Introduction

Thawing permafrost is a major concern in Arctic landscapes [Hinzman et al., 2013; Jorgenson et al., 2006; Mack et al., 2004; Schuur et al., 2008]. Such degradation increases soil carbon availability, which could result in increased emissions of carbon dioxide and methane with potentially large feedbacks on the global climate [Shaver et al., 1992; Weintraub and Schimel, 2005; Zona et al., 2009]. Degradation is also associated with substantial geomorphological shifts which are coupled with changes in hydrology, vegetation, and biogeochemical cycling [Frey and McClelland, 2009; Hinzman et al., 2013; Jorgenson et al., 2006; Lara et al., 2014; Shaver et al., 1992]. Active layer geochemistry reflects redox states, nutrient cycling by plants and microbes, mineralogy, and cation exchange capacities. Therefore, it is important to understand the relationship between active layer geochemistry and the microtopography of ice wedge polygons that control water movement over much of the relatively flat Arctic Coastal Plain [Liljedahl et al., 2012].

A common feature of Arctic landscapes with low relief is the presence of polygonal or patterned ground (Figures 1 and S1 in the supporting information) [Hussey and Michelson, 1966]. This type of terrain is related to the distribution and depth of ice wedges that create and surround the characteristic polygonal elements. There are different polygonal ground types including low-centered polygons (the center is a topographic low) and high-centered polygons (the center is a topographic high). High-centered polygons can develop from permafrost degradation through topographic inversion of low-centered polygons [Gamon et al., 2012; Jorgenson and Osterkamp, 2005]. Inversion occurs when melting of ice wedges that bound low-centered polygons create trough-like lows, and formerly low centers become topographic highs. The spatial scale of polygon features is typically on the order of a few to tens of meters, resulting in a condition where there is significant lateral microtopographic variation over relatively short distances.

Figure 1

Types of polygonal ground at the Barrow Environmental Observatory observed using 0.5 m resolution lidar imagery. White circles indicate active layer sampling locations.

Existing studies (discussed below) have noted a strong link between some polygon types and features and biogeochemistry. This microtopographic relationship suggests that preferential distribution of water must play a major role in controlling biogeochemical variability. Existing studies highlight why microtopography is important in understanding spatial variability of carbon dioxide and methane emissions, how nutrients and other solutes are distributed, and how vegetation and Arctic energy balances may change with increasing permafrost degradation. Most studies have focused on low-centered polygon systems where there are strong microtopographic controls on thaw depths, pH, redox conditions, dissolved carbon dioxide, methane production, and iron cycling [e.g., Lipson et al., 2010, 2012, 2013; Zona et al., 2011]. Data on high-centered polygons are limited, but some biogeochemical differences have been noted between high- and low-centered polygons [Biasi et al., 2005; Brown, 1967; Gersper et al., 1980; Hubbard et al., 2013]. Previous studies of polygon types and features have largely focused on a limited suite of analytes, and few studies have examined how geochemistry varies with depth through the entire active layer (i.e., the seasonal thaw zone).

The objective of this study was to statistically examine variations in a large suite of analytes in a polygon-dominated landscape. Our hypothesis was that multiple significant differences would occur between low-centered and high-centered polygons (i.e., polygon type), between polygon centers and troughs (i.e., polygon features), and with active layer depth. Such microtopographic and depth controls on geochemistry have important consequences as polygonal tundra ecosystems evolve under a warming climate.

Our study area was an interlake polygonal ground area on the Arctic Coastal Plain near Barrow, Alaska, USA (71.3°N, 156.5°W). Soils are underlain by the Gubik Formation of late Quaternary (mid-Wisconsinan) sediments deposited during marine transgressions [Sellmann and Brown, 1973]. Since then, the area has developed continuous permafrost, and the thaw lake cycle, ice wedge growth, plant organic matter deposition, and cryogenic processes have formed the soils, which thaw annually to a depth of approximately 40 cm [Hinkel et al., 2003, 2005]. Quartz and chert dominate the mineral fraction of the soils [Black, 1964]. Radiocarbon dating indicates a slow accumulation of soil organic matter in the top 1–2 m during most of the Holocene period [Brown, 1965; Hinkel et al., 2003; Meyer et al., 2010]. Finer grain sizes correlate with higher salt content [Brown, 1969], and conductivity generally increases with depth [Meyer et al., 2010; O'Sullivan, 1966]. Ion leaching, brine migration, and redistribution due to water freezing have reduced the ion concentrations in soil water relative to marine sediments and increased the proportion of Ca²⁺ and Mg²⁺ compared to Na⁺ [O'Sullivan, 1966]. The current study area is part of the Barrow Environmental Observatory (BEO) and is an ideal location to study fine-scale variation in geochemistry across polygonal ground, because it contains high-centered and low-centered polygon areas in close proximity.

2 Methods

2.1 Field and Laboratory

Surface waters were collected as grab samples. Collection of subsurface water samples was often challenging, and multiple methods were used for sampling (see supporting information). Samples were classified by three factors: polygon feature (center or trough of a polygon), polygon type (high-centered or low-centered polygon), and depth. Depths were classified as surface water, shallow subsurface water (<10 cm), and deep subsurface water (20–55 cm, collected just above the frost line). A few low-center rims were sampled, but not enough for robust statistics. Troughs form the surface drainage network in both high- and low-centered polygons. One hundred and twenty-one samples were collected from early July through mid-September 2012 (Figure 1). Standard water chemistry preservation and analytical techniques were used, and determinations were made for 54 different analytes (see supporting information).

2.2 Statistical Methods

Statistical analyses were conducted using the R environment (v. 2.14.0; R Core Team, 2013). Principle component analysis (PCA) was conducted using the 54 analytes (see supporting information Table S2 for full listing). Data were log transformed and converted to Z scores prior to PCA analysis. Z scores reduce the effect of high-concentration analytes having undue influence on PCA results. PCA was used to identify key analytes for subsequent statistical comparisons.

For microtopographic (polygon types and features) and depth comparisons, we used a one-way Spearman permutation test with 9999 Monte Carlo simulations to account for the unbalanced statistical design (i.e., an unequal number of samples from different polygonal topographic units or depths being tested). We used a hierarchical approach that compared low-centered polygon areas to high-centered polygon areas (i.e., polygon type), and polygon features (troughs versus centers regardless of polygon type). We additionally made pairwise comparisons of the centers of high- and low-centered polygons, troughs of high- and low-centered polygons, centers and troughs within high-centered polygon areas, and centers and troughs within low-centered polygon areas. Finally, we tested the effect of depth on selected analytes with ANOVAs comparing surface water samples, shallow subsurface pore waters (0–10 cm), and deep pore waters (30–55 cm). Any p value < 0.1 was considered significant, but we also explored tighter criteria up to p < 0.001 to highlight analytes that were highly significant.

3 Results and Discussion

As an initial characterization of the site-wide polygonal ground geochemistry, we plotted a selected set of analytes using a fingerprint diagram [see Mazor, 2004; Schoeller, 1959] (Figure 2). Fingerprint diagrams are a useful way of evaluating differences in water chemistry at a site and not only show the concentrations (meq/L) of different ions for the various samples but also graphically show how the various ions relate to each other within a sample via the linear connections between the concentrations of the different ions. For the BEO waters, the parts of the fingerprint traces that include sodium, magnesium, calcium, and chloride show relatively small concentration differences and the relationships between these ions are fairly consistent. This observation suggests that physical geochemical influences (e.g., mineral weathering) on water chemistry do not vary greatly across the study area. However, there are substantial concentration and ion ratio differences between samples for ions strongly influenced by biological/redox processes. Manganese, potassium, iron, sulfate, nitrate, and phosphate concentrations drive wide variations in patterns. Iron has another notable feature, in that iron concentrations can be exceptionally high. This result suggests that the significance of iron biogeochemistry discussed by [Lipson et al., 2010, 2012, 2013] in an adjacent drained thaw lake basin also extends into interlake polygonal areas of the landscape. A key question that we address in this study is whether variations reflected in the fingerprint diagram are spatially controlled.

Figure 2

Fingerprint diagram of key active layer ions. Each line on the cation or anion side of the diagram represents the “fingerprint” of a single sample. Linear connections between concentrations of each ion illustrate the relationships between one ion and another within a sample. Note that total iron values were used and assumed to be ferric iron for milliequivalent conversion. Lipson et al. [2010] studied an adjacent drained thaw lake basin and found that ferric iron was often the dominant dissolved iron species.

Principal components analysis (PCA) was used to reduce the dimensionality of the data and to identify which analytes were most critical to focus on (see supporting information Table S3 for PCA summary). The first two components (PC1 and PC2) accounted for 46% of the total variance, and the first four principal components accounted for 62%. These results suggest that the first four components have substantial explanatory power and could be used further to identify important aspects of polygonal ground chemistry. Each of the four components was examined to determine which parameters were most important based on the magnitude of their eigenvalues. Eigenvalues are calculated for each parameter and the larger the magnitude, the more important that parameter is within a particular component. This approach provides a way of identifying key parameters for additional analyses without having to investigate all parameters in detail. The highest positive eigenvalues for PC1 included strontium, magnesium, iron, calcium, manganese, and total organic carbon; and the highest negative eigenvalue was for dissolved oxygen. The highest positive eigenvalues for PC2 included chloride and sodium, and the highest negative eigenvalues were for sulfate, nitrate, phosphate, and aluminum. The highest positive eigenvalues for PC3 included silica and pH, and the highest negative eigenvalues included manganese, potassium, and dissolved oxygen. The highest positive eigenvalues for PC4 included phosphate and aluminum, and the highest negative eigenvalues included nitrate and potassium. Aluminum and silica appear with high eigenvalues for the first time in PC3 and PC4. Subsequent analysis focuses on the above set of 16 high (absolute value) eigenvalue analytes, and ANOVA results are grouped by principal component.

3.1 Microtopographic Effects

We first present ANOVA results and follow with a discussion of the importance and implications of biogeochemical variability in polygonal ground. When high-centered polygons versus low-centered polygons (including their centers and associated troughs) were compared, significant differences were found for nine analytes, and magnesium, dissolved oxygen, chloride, phosphate, nitrate, and sulfate had highly significant (p < 0.01) differences (Table 1, “Polygonal Type”). Box-and-whisker plots (Figure 3) demonstrate the magnitude of these differences and often show clear increasing or decreasing trends from high- to low-centered polygons. For polygon features (centers versus troughs, regardless of polygon type), significant differences occurred for five analytes, and highly significant (p < 0.01) differences occurred for manganese and total organic carbon (Table 1, “Features”).

Table 1. ANOVA Results (p Values) for Biogeochemical Differences by Polygon Type, Feature, Depth, and by Polygon Type Specific Featuresa

	Polygon Type (High Versus Low Center)	Feature (Center Versus Trough)	Depth (Surface, Shallow, and Deep)	Center of High Center–Center of Low Center	Center of High Center–Trough of High Center	Center of Low Center–Trough of Low Center	Trough of High Center–Trough of Low Center
PC1
Sr^2+	0.013**	0.419 ns	0.001***	0.002***	0.005***	0.753 ns	0.452 ns
Mg^2+	0.002***	0.785 ns	<0.001****	0.004***	0.028**	0.73 ns	0.126 ns
Fe^Total	0.136 ns	0.065*	0.002***	0.022**	0.008***	0.127 ns	0.37 ns
Ca^2+	0.988 ns	0.563 ns	0.0022***	0.681 ns	0.373 ns	0.81 ns	0.836 ns
Mn^2+	0.826 ns	0.001***	0.325 ns	0.151 ns	0.003***	0.023**	0.94 ns
Total organic carbon	0.35 ns	<0.001****	0.009***	0.142 ns	0.005***	<0.001****	0.034**
Dissolved oxygen	0.004***	0.889 ns	<0.001****	nd	nd	0.051*	0.01***
PC2
Cl^−	<0.001****	0.923 ns	0.261 ns	<0.001****	0.006***	0.776 ns	0.02**
Na^+	0.028**	0.014**	0.047**	0.133 ns	0.384 ns	<0.001****	0.007***
PO4^3−	<0.001****	0.569 ns	0.607 ns	<0.001****	0.029**	0.031**	0.149 ns
NO3^−	0.001***	0.072*	0.247 ns	<0.001****	<0.001****	0.984 ns	0.709 ns
SO4^2−	<0.001****	0.245 ns	0.032**	<0.001****	<0.001****	0.995 ns	0.056*
Al^3+	0.811 ns	0.897 ns	0.042**	0.312 ns	0.114 ns	0.769 ns	0.621 ns
PC3 and PC4
Si	0.013**	0.648 ns	0.039**	0.012**	0.947 ns	0.008***	0.026**
pH	1 ns	0.195 ns	<0.001****	nd	nd	0.137 ns	0.612 ns
K^+	0.081 ns	0.941 ns	0.009***	0.397 ns	0.69 ns	0.571 ns	0.033**

• ^a Analytes are separated based on their first appearance in principal components 1–4. Asterisks indicate level of significance: p < 0.1 (*); p < 0.05 (**); p < 0.01 (***); p < 0.001 (****); not significant (ns); and not determined (insufficient data, nd). Depth variables include surface water, shallow subsurface (<10 cm), and deep subsurface (typically > 25 cm, at or near the frost line).

Figure 3

Box plots for the 16 high eigenvalue analytes from PC1 to PC4 for location*polygonal type, with p values included in figure legends when significantly different. All units are in mg/L, except for pH (unitless). For each category, the bold line indicates the median, and the box indicates the second and third quartiles. Whiskers represent 1.5X the interquartile distance. Circles represent values outside the whisker range.

The effects of polygonal microtopography can be parsed further using pairwise comparisons of the same features across polygon types, and between centers and troughs within a type. At least six significant differences were found for each comparison: between the two types of centers, between the two types of troughs, and between centers and troughs within high- or low-center types (Table 1 and Figure 3). Highly significant differences were observed for strontium, magnesium, chloride, phosphate, nitrate, and sulfate between the centers of high-centered polygons and the centers of low-centered polygons. Comparing centers and troughs within high-centered polygons showed significant differences for eight analytes, and highly significant differences for strontium, iron, total organic carbon, chloride, nitrate, and sulfate. Within low-centered polygons, comparison of centers and troughs showed significant differences for manganese, dissolved oxygen, and phosphate, and highly significant differences occurred for total organic carbon, sodium, and silica

Results show significant geochemical variability at both the polygon type and polygon feature scales. The high number of significant differences for polygon type (nine analytes, including six highly significant differences) suggests that polygon type should be useful for larger-scale geochemical representations, consistent with Hubbard et al. [2013] who found that a variety of geophysical properties and other parameters (e.g., active layer depth and moisture content) also varied by polygon type at the BEO. Differences observed for divalent cations may reflect variations in the degree of organic complexation across polygon types, which are supported by the similar relationships between these ions and total organic carbon in Figure 3. Some of the differences in polygon types appear to be related to more oxic conditions in high-centered polygons (e.g., dissolved oxygen, sulfate, nitrate, and phosphate). However, differences in phosphate, for example, may also be related to bird behavior where centers of high-centered polygons are used as perching sites, and associated waste products are concentrated. Heterogeneity in dissolved oxygen and oxyanion concentrations also suggests that production of carbon dioxide and methane should vary with polygon type. For example, the more oxic conditions in high-centered polygons would tend to limit methane production and possibly promote methane oxidation [Morrissey and Livingston, 1992]. Lower carbon dioxide and methane fluxes from high-elevation polygon features such as high centers and rims have been reported [Morrissey and Livingston, 1992; Olivas et al., 2011; Sachs et al., 2010; Sturtevant et al., 2012; Zona et al., 2011]. Overall, these results emphasize that biogeochemical differences are an important characteristic of different types of polygons.

Results based on features (Table 1) suggest, even with mixed polygon types, that centers and troughs also play an important role in active layer biogeochemistry. Aside from sodium and nitrate, the types of analytes with significant differences were not the same between features and polygon types (i.e., of the 12 analytes with significant differences for either features or type, 10 of them were only found in one or the other treatment). This observation suggests that different types or rates of biogeochemical processes are operating at finer, feature-based microtopographic scales. These differences may also be a reflection of cryogenetic variations related to feature type [see Ping et al., 2008].

Given the apparent importance of features, we made additional comparisons of specific features linked to polygon type (i.e., between centers and troughs within a particular polygon type, and across polygon types). For centers of high-centered polygons versus centers of low-centered polygons, significant differences were found for seven different analytes, and differences in the oxyanions and chloride were highly significant, paralleling differences observed by polygon type. Box-and-whisker plots (Figure 3) clearly demonstrate a variety of biogeochemical differences (especially for nutrients) between the two types of centers suggesting that permafrost degradation driven shifts from low- to high-centered polygons (which can occur over decadal time scales) would likely have substantial consequences on polygonal ground ecosystems. Biogeochemical changes will likely be coupled with other processes because microtopography also affects distributions of vascular and nonvascular plants, soil moisture, and evapotranspiration in polygonal ground [e.g., Engstrom et al., 2006; Gamon et al., 2013; Huemmrich et al., 2013]. Recent changes in the plant community composition and diversity have been documented for the Barrow area [Villarreal et al., 2012], and future vegetation shifts will likely be coupled with changes in nutrients [Bret-Harte et al., 2001; Clemmensen et al., 2006; Marion and Everett, 1989; Shaver et al., 2001].

Multiple significant differences were observed between centers and troughs within high-centered polygon areas (Table 1) where oxyanions tended to have higher concentrations in centers, while troughs had higher total organic carbon (Figure 3). Differences in the oxyanions are likely related to the fact that troughs are frequently inundated, while the centers are not. Biasi et al. [2005] also found higher total organic carbon in the troughs of high-centered polygons compared to centers.

Within low-centered polygons, significant differences occurred between troughs and centers for total organic carbon, phosphate, dissolved oxygen, and sodium (Table 1). Gersper et al. [1980] and Zona et al. [2011] also noted geochemical variations between low-centered features. Zona et al. [2011] noted that organic material tends to accumulate in low-elevation areas of drained thaw lake basins, and the total organic carbon data from this study suggest that accumulation occurs in troughs of both high- and low-centered polygon systems.

Significant differences also occurred between troughs in low- and high-centered polygons for total organic carbon, sulfate, dissolved oxygen, sodium, and chloride (Table 1). For these analytes, concentrations are typically higher in low-center troughs, except for dissolved oxygen which can be substantially higher in high-center troughs. The comparisons of the different combinations of features show that even fine-scale microtopographic features within a particular type of polygon or between high- and low-centered polygons result in substantial variations in biogeochemistry. Because some analytes significantly differed across features, but not for polygon type, features may also be important predictors of landscape-scale biogeochemical processes.

3.2 Depth Relations

The majority of analytes varied significantly with depth (see Table 1 (Depth) and supporting information Figure S2) with the exception of manganese, nitrate, phosphate, and chloride. Most ion concentrations increased with depth, as did pH. However, dissolved oxygen tended to decrease with depth. Calcium, pH, potassium, and aluminum were significantly or highly significantly affected by depth but were not significantly affected by polygon types or features suggesting that controls on chemistry differ between vertical and lateral dimensions. Differences in soil types with depth (i.e., between the shallow organic horizons and deeper mineral horizons, supporting information Figure S1) likely drive a substantial amount of the depth variability, but other factors discussed below may also be important.

Given the importance of depth on active layer chemistry, it is worthwhile to consider other depth-based studies of the active layer. Some investigations reported increases in organic carbon, selected cations, pH, or decreases in dissolved oxygen with depth [Brown, 1967, 1969; Gersper et al., 1980; Lacelle et al., 2007, 2008; Pecher, 1994; Ping et al., 1998; Stutter and Billett, 2003]. Other investigations did not observe concentration increases or obvious trends with active layer depth [Kokelj and Burn, 2003, 2005], which may be related the effects of cryoturbation. The literature is limited, and only small suites of analytes are typically described, so it is not clear how common the strong depth relations we observed are in other permafrost landscapes. Lacelle et al. [2007] suggest that increases in ion concentrations with depth should be typical in seasonally frozen soil as solutes are concentrated in residual water at the bottom of the active layer during freeze back. Our speculation is that depth trends are probably more common than the literature implies and more thorough fine-scale investigations of active zone concentration/depth relationships in different Arctic landscapes are warranted. It is also not clear what controls depth trends. Downward transport was implied by Brown [1967] and Pecher [1994], and while this might explain why concentration trends develop, it does not readily explain how analytes become concentrated in the first place, or why pH and dissolved oxygen also have depth relationships. Clearly, hydrological, microbial, and rhizosphere processes are going to affect dissolved oxygen, organic carbon, and some ion depth distributions. For example, Lipson et al. [2012] and Zona et al. [2011] suggest that biological controls are substantial for pH and iron distributions. An explanation of higher ion concentrations at the frost line was proposed by Lacelle et al. [2008] who suggested that the base of the active layer is dominated by chemical weathering which promotes enrichment. Precipitation must also play a role because it would tend to reduce surface and near-surface concentrations.

However, there is another factor that may be important. Bockheim and Hinkel [2005] and Kokelj and Burn [2005] observed substantial increases in ion and organic carbon concentrations in the ice-rich uppermost part of permafrost known as the transition zone. For example, Kokelj and Burn [2005] found that mean total soluble cation concentrations were 1.4 to 4.3 times greater in the transition zone than in the active layer, and Bockheim and Hinkel [2005] found that organic material was also concentrated in the transition zone. Because the transitional layer periodically thaws, it could contribute to the increase of organic carbon and ions in the deepest part of the active layer. In summary, different hypotheses have been proposed about what causes biogeochemical depth variations in the active layer. These hypotheses imply different temporal modes where some assume that the controlling processes happen every year during the thaw (or freeze/thaw) period, while the transition zone hypothesis is controlled by a punctuated multiyear process. These differences in mechanisms and temporal dynamics suggest that more focused studies about the controls on chemical variations with depth are needed if our understanding is to go beyond empirical observations.

4 Summary and Conclusions

How sources and sinks of greenhouse gases vary laterally and vertically within the active layer and how the energy balance and vegetation relate to variations in microtopography are two major uncertainties related to feedbacks of Arctic polygonal ground systems to climate. Active layer geochemistry is both a reflection of and a control on microbial and plant processes relating to these uncertainties. Therefore, it is valuable to understand how geochemistry varies with microtopography and depth in polygonal ground. This study examined such variations in an interlake area in the Arctic Coastal Plain that contains both high- and low-centered polygons (Figure 1) during the July–September warm period. Significant differences occurred between polygon types for multiple analytes, with fine-scale differences also occurring between troughs and centers within and between particular polygonal types. Multiple observations of analytes with significant differences for polygon types (i.e., high- versus low-centered), but not for polygon features (centers versus troughs), and vice versa, suggest that these two scales are influenced by different biogeochemical processes and rates. Thus, while polygon type may be useful for representing some biogeochemical aspects of polygonal ground at the landscape scale, features may be more important for representing other aspects. It is also important to recognize that active layer depth is as important as microtopography in terms of biogeochemical variability. Depth significantly affected the concentrations of most analytes, some of which were not significantly different for polygon type or feature, which indicates that different mechanisms may be controlling variations with depth compared to microtopography. Because permafrost degradation may cause the landscape to evolve to one with more high-centered polygons (or other kinds of geomorphic changes), microtopographic and depth controls on localized geochemistry will likely have major impacts on polygonal ground ecosystems. These impacts would be coupled with changes in other ecosystem processes and characteristics because of the strong relationships between microtopography and vegetation types, microbial processes including greenhouse gas fluxes, hydrology, and the energy balance [e.g., Engstrom et al., 2006; Gamon et al., 2012; Huemmrich et al., 2013; Lara et al., 2014; Lipson et al., 2012; Zona et al., 2011]. Given the importance of microtopography and depth variations, we suggest additional studies to better understand spatial controls on active layer biogeochemistry in a broader Arctic context.

Acknowledgments

The Next-Generation Ecosystem Experiments (NGEE-Arctic) project is supported by the Office of Biological and Environmental Research in the DOE Office of Science. Review comments from John Gamon and an anonymous reviewer were much appreciated. We would also like to acknowledge Lily Cohen, Marvin Gard, Emily Kluk, Mike Rearick, George Perkins, and Xiangping Yin for their technical assistance and UMIAQ, LLC for logistical support in Barrow. Data used in this study can be obtained by contacting the corresponding author.

The Editor thanks John Gamon and two anonymous reviewers for their assistance in evaluating this paper.