Microbiogeochemical Traits to Identify Nitrogen Hotspots in Permafrost Regions

Abstract

Permafrost-affected tundra soils are large carbon (C) and nitrogen (N) reservoirs. However, N is largely bound in soil organic matter (SOM), and ecosystems generally have low N availability. Therefore, microbial induced N-cycling processes and N losses were considered negligible. Recent studies show that microbial N processing rates, inorganic N availability, and lateral N losses from thawing permafrost increase when vegetation cover is disturbed, resulting in reduced N uptake or increased N input from thawing permafrost. In this review, we describe currently known N hotspots, particularly bare patches in permafrost peatland or permafrost soils affected by thermokarst, and their microbiogeochemical characteristics, and present evidence for previously unrecorded N hotspots in the tundra. We summarize the current understanding of microbial N cycling processes that promote the release of the potent greenhouse gas (GHG) nitrous oxide (N₂O) and the translocation of inorganic N from terrestrial into aquatic ecosystems. We suggest that certain soil characteristics and microbial traits can be used as indicators of N availability and N losses. Identifying N hotspots in permafrost soils is key to assessing the potential for N release from permafrost-affected soils under global warming, as well as the impact of increased N availability on emissions of carbon-containing GHGs.

1. Introduction

Permafrost-ground that remains frozen continuously for at least two years, lies mainly under about 13–18 million km² (9–14%) of the land surface in the Northern Hemisphere and smaller portions in the alpine and ice-free regions of Antarctica [1,2,3,4,5]. Based on the spatial extent of permafrost, regions are divided into continuous (90–100%), discontinuous (50–90%), sporadic (10–50%), or isolated (0–10%) permafrost zones [6] (Figure 1). A variety of Arctic terrestrial ecosystem types is subdivided according to the composition of the vegetation and the depth of organic layer [1,7,8,9]. These ecosystems are dry and wet tundra with a thin organic layer (<40 cm); two peatland ecosystems with a thick organic layer (>40 cm): ombrothrophic bogs (including uplifted palsas and peat plateaus) and minerotrophic fens; and upland forests and littoral ecosystems where permafrost is not always present [9,10,11]. Permafrost-affected soils are characterized by low temperatures (mean annual temperature < 0 °C) and often high soil moisture [8]. Soils are usually divided into three layers: (1) the active layer (AL), the upper part of the soil, that is subject to seasonal freezing-thawing cycles, (2) the permafrost layer (PL), which remains constantly frozen throughout the year, and (3) a transition zone (TZ, [11,12]). The permafrost table acts as a water barrier, significantly affecting redox conditions [13]. Due to the heterogeneity of topography and hydrological conditions, vegetation types, and soil physicochemical properties (including SOM quantity and quality), (sub)Arctic tundra environments exhibit high small-scale variability [8,11,14,15,16,17,18,19,20,21,22]. In terms of seasonal dynamics, permafrost soil ecosystems differ from those at lower latitudes due to long winters, a strong influence of snow cover on soil temperature and moisture, repeated freeze-thaw cycles in spring and fall, and short growing seasons. Therefore, theses soils are subject to strong temporal and spatial dynamics that affect the availability of nutrients [23,24,25,26,27,28,29,30].

Low temperatures, cryoturbation, i.e., the mixing of soil by repeated freeze-thaw cycles, and poor soil drainage result in slow decomposition, and thus the net sequestration of organic carbon in permafrost-affected soils [12,32]. Consequently, permafrost soils store a large amount of organic nitrogen [33] and carbon (1014 Pg C, depth 0 to 3 m, northern circumpolar and Tibetan permafrost regions [34]). The actual N pool is much larger than current estimates of 0.9–2.2 (6.0) kg N m⁻² in the first meter of soil [35,36,37,38,39], and 67 Pg N within in the first 3 m for the northern permafrost region [37]. Because of the highly variable C/N ratios in northern circumpolar soils [35,38,39,40] with the majority of soils ranging from 10–50 [21], the actual N stock is uncertain and could be twice as high as calculated [37] especially for the thermokarst-influenced Yedoma—ice-rich Pleistocene sediments—landscapes [39,41,42]. Recent studies on Yedoma permafrost estimated total N stocks in permafrost to be as high as 97 Pg N at depths to more than 3 m [42]. However, most of the nitrogen is firmly bound in soil organic matter (SOM) by tight N mineralization-immobilization-turnover (MIT) or in the perennially frozen ground, therefore tundra ecosystems are generally described as severely N-limited [30,43,44,45,46,47,48,49,50,51,52,53,54,55,56]. Nitrogen limitation is enhanced by low internal N inputs due to low or even negative net N mineralization rates, including ammonification and nitrification under cold and wet conditions (summarized in [53,57,58,59]) and low external N inputs due to biological N fixation (20–2000 mg m⁻² y⁻¹, [53,59,60,61]) combined with low atmospheric N deposition (<200–300 mg m⁻² y⁻¹, [62,63]), and no external N fertilization. Recycled N accounts for over 95% in the Arctic tundra [47] and inhibited microbial inorganic N turnover [64,65]. Bioavailable dissolved inorganic N (DIN) concentrations are low and N processes are dominated by N immobilization indicating a tight N cycle in organic rich soils [47,64,66,67,68,69] due to strong competition between plants and microorganisms [65,70,71,72]. The inorganic N demand of plants could be compensated by using bioavailable dissolved organic N (DON) forms [23,65,73], which are mostly available in higher concentrations than DIN [23,71,74,75,76,77]. Although the N cycling of permafrost-affected soils is less open and N output is lower than in other ecosystems, the organic-only N cycling paradigm has recently been refuted [59] as gross N mineralization rates are higher than previously expected [15,51,78,79,80,81] and soils may be susceptible to lateral N losses [56,65,82].

In addition, advances in microbial gene abundance and activity surveys over the last two decades have shown that N cycling microbes can be very abundant in both the seasonally thawing active layer and the long-term frozen permafrost, and that microorganisms actively drive N cycling (recently reviewed in [58,59]). At the same time, there is increasing evidence that permafrost-affected soils represent an important source of the potent greenhouse gas (GHG) nitrous oxide (N₂O) [82], a strong ozone-depleting gas and the third most important GHG with a global warming potential of 298 times higher than that of CO₂ on a 100-year timescale [83]. Permafrost-affected soils release N₂O, especially when N uptake by plants is low and soil water content is intermediate. According to a recent first-order estimate, permafrost-affected soils emit up to 20% of the global N₂O emissions from natural soils [82]. In addition, hydrologic export of inorganic N from permafrost-affected soils may be high, which likely has strong impacts on aquatic ecosystems [84,85]. Nitrogen losses occur primarily in permafrost-affected soils disturbed by wind erosion, cryoturbation, or thermokarst, reducing N uptake by plants and increasing N inputs creating hotspots of high N availability [86,87,88]. Furthermore, the microbial N cycle and its effect on C cycle and GHG release are still subject to large uncertainties, although the contribution of permafrost soils to global N₂O emission from pristine soils is high, and the effects of inorganic N availability on the C cycle in soils and adjacent aquatic ecosystems could have implications for the Earth’s GHG budget [47,50,52,55,89,90,91,92,93,94,95,96]. In the future, permafrost regions will be strongly affected by climate change, rising temperatures and changing soil moisture conditions [11,97,98], deepening of the active layer and thermokarst [5,99,100], which will affect microbial processes in these soils [101,102]. Increased N availability is expected due to thawing of permafrost containing large amount of inorganic N [42,75,76,103], increased microbial N mineralization [75,102,104], and expansion of disturbed areas known for high N losses [82]. Therefore, it is of great importance to better understand (1) the current pattern of N availability in permafrost-affected soils, (2) the underlying microbial driven N cycling, (3) the environmental controls that regulate the microbial formed N₂O emission and inorganic N export to aquatic ecosystems.

In this review, we aim to highlight the hotspots of high N availability in permafrost-affected ecosystems, which can be identified by certain traits that correlate with high lateral and gaseous N losses. First, we provide an overview of the main microbial N transformation processes that are affected by N availability and largely determine N losses, in particular autotrophic nitrification and heterotrophic denitrification. In the main section, we describe the microbial N cycling and associated N-loss processes, particularly in the two best-described N hotspots in cryoturbation-affected soils, namely bare organic lowland soils in permafrost peatlands and disturbed mineral upland soils in hillslope thermokarst landscapes. In this section, we also present preliminary evidence of additional N hotspots in the tundra. Finally, we analyze how N hotspots interact with climate change and carbon-relevant GHGs, and provide an outlook for future research. Hypotheses are:

• Permafrost-affected soils are not generally N-limited, but there are N hotspots of N availability with more open N cycling and increased potential for N losses in the form of inorganic N leaching and N₂O emissions.
• N hotspots can be identified by certain soil characteristics and microbial traits: different N forms, microbial net N turnover rates, especially N mineralization, nitrification and denitrification, and abundance of key functional N cycling genes.

2. Regulation of N Availability by Microbial N Processes in Permafrost-Affected Soils

2.1. Levels and Traits of N Availability

The N availability of an ecosystem can be described as a gradient from N limitation to N saturation, which is indicated by several traits and determines the potential of N losses due to the excess of inorganic N. Nitrogen availability is defined as high when N supply is greater than N demand [105]. The N limitation of primary production in tundra soils has been demonstrated by numerous fertilizer experiments in which mainly inorganic nitrogen was supplied (recently reviewed by), indicating that N is mainly limited as DIN. In permafrost-affected soils, the following parameters were mainly used individually or together as indicators for N availability: (1) N forms in the bulk soil, such as high total N content (TN) (as well as associated low C/N ratio), detectable dissolved organic and inorganic N forms (DON, DIN), a high ratio of DIN/DON and nitrate/ammonium [28,30,55,68,75,86,95,106,107,108,109,110,111] and high δ¹⁵N content [27,95,111,112,113,114,115], (2) vegetation parameters such as high total N and low C/N ratio, nitrate or δ¹⁵N content in foliage or litter [55,57,95,107,108,115,116,117], (3) high rates of microbial driven N processes: N mineralization (ammonification, and nitrification) [17,51,55,68,75,81,93,95,107,109,111,115,118] or (4) a high relative abundance of key functional genes encoding the enzymes responsible as an indirect indicator of these processes [20,87,95,109,118,119,120,121].

Using these N availability traits, we propose to divide the N availability of permafrost-affected soils into three levels based on the model of Schimel and Bennett [65] (Figure 2). These levels mainly characterize the ratio of N immobilization to N mobilization indicated by the content of different N forms and rates of net microbial N cycling processes. Unlike gross N cycling rates, these net N cycling rates do not reflect the total amount of N cycling, but rather how much N remains after N uptake by soil organisms and plants and incorporation into their biomass by anabolic processes for inorganic catabolic N cycling [122,123]. Arctic ecosystems are generally at the first level of N availability, the N limitation level where N immobilization dominates [65,122,124] (Figure 2I). Plants and microbes compete strongly for organic and inorganic N forms and therefore the N cycle is tightly closed. This level is characterized by mainly organic N forms as high DON content, high C/N ratio of bulk soil (>25), low detectable DIN mainly in form of ammonium and therefore a low DIN/DON or DIN/TN ratio. In addition, net depolymerization rates may be high, but N mineralization and nitrification rates are zero or even negative. Since the bulk C/N ratio of SOM poorly reflect the N availability to microorganisms (especially in mineral soils where SOM is mineral bound), the imbalance between bulk soil C/N and microbial biomass C/N could be better used as indicator [68,125]. For boreal peat soils, the C/N ratio was used as a predictor of N₂O emissions, above a threshold of 25, emissions rapidly increase, and a similar threshold was demonstrated for permafrost peatlands [86,126]. Due to the high N immobilization rates, there is no excess of N surplus for net inorganic N turnover and therefore no N loss in form of DON, inorganic and N₂O.

At the second level (intermediate N availability, Figure 2II), in addition to the high importance of organic N forms, the N cycle is more open than at the first level. At this N level, net N mineralization rates, particular ammonification rates and thus higher ammonium contents are detectable. Since low or no net nitrification rates occur, the inorganic N is only lost as ammonium at this level.

The third level, also referred as N hotspot (Figure 2III) is mainly characterized by low C/N ratio and high δ¹⁵N of bulk soil, low DOC and DON content, and high DIN/DON or DIN/TN ratio, and high inorganic N content with high nitrate/ammonium ratio. In addition, net inorganic N cycling process rates of nitrification under oxic and denitrification rates under anoxic conditions can be considered as indicator of N availability, as they are limited by low inorganic N availability in soils with tight N cycle. An intermediate soil moisture with a water-filled pore space (WFPS) of about 60% allows aerobic and anaerobic microbial N processes to occur simultaneously [127]. Besides process rates related to microbial activities, a high relative abundance of functional genes encoding the key enzymes of these processes can serve as an indicator. At the third level, N mobilization predominates over the N immobilization because N demand is lower, resulting in a more open N cycle that favors N leaching, mainly in the form of nitrate, and N losses in gaseous form. Both nitrification and denitrification produce nitric oxide (NO) and nitric acid (HONO), both important for atmospheric chemistry, the GHG N₂O, and the atmospheric inert gas N₂ through the complete denitrification process (reviewed in [128,129,130]).

2.2. Key Microbial N Processes Determining N Availability in Permafrost-Affected Soils

Because of the large stores of organic matter, permafrost-affected soils have a great abundance and diversity of distinct soil microbial communities in both the active layer and permafrost [131,132,133]. Although these organisms must cope with extreme cold and fluctuating temperatures, frequent freeze and thaw cycles, and frozen ground combined with low nutrient and liquid water availability, microbial communities are well adapted or highly resistant to these adverse conditions [101,134,135,136,137,138]. The abundance, richness, and evenness of communities generally decrease with soil depth [101,131,132,139,140,141,142,143], but there is still microbial activity at subzero temperatures and in the permafrost layer [101,144,145,146,147,148]. Microbial community composition varies geographically and at smaller scales, and microbes can survive and be activated even after thousands of years [132,134,149,150,151]. Despite these cold conditions, the enriched bacteria are mostly cold-adapted (psychrotolerant) rather than cold-favoring (psychrophilic) organisms [119,152,153,154,155]. In addition to spatial heterogeneity, microbial communities also exhibit a high degree of seasonal heterogeneity [28,156,157,158,159,160,161].

A complex network of metabolically versatile microorganism catalyzes various N processes in soils in general (reviewed, e.g., by [128]) as well as in permafrost-affected soils (reviewed, e.g., by [53,58,59]). The microbial composition and activities of N cycling microorganisms, together with organic matter quantity and quality, determine the N availability of soils. Although ammonification, the conversion of organic N to ammonium, and biological N fixation (BNF) are the major sources of internal and external nitrogen input to the soil system, we focus in this review on the major processes of inorganic nitrogen cycle. In particular, these are autotrophic nitrification and heterotrophic denitrification as key processes for the display and regulation of N availability, since both processes depend on reactive N, but are also responsible for N loss at the final level of N availability (Figure 2III). Aerobic autotrophic nitrification can be considered a bottleneck for N losses in permafrost-affected soils, as it is the main process that forms nitrate, which is important for plant nutrition [57] and allows lateral inorganic N loss in form of nitrate leaching and vertical losses through the formation of N gases. The role of heterotrophic nitrification, recently reviewed in soils in general [162], is rarely analyzed in permafrost-affected soils [163]. The nitrification process produces NO and N₂O as a byproduct under oxygen-limited conditions [128,129] and is also the only process that provides the substrate for anaerobic heterotrophic denitrification (and the nitrification-denitrification process) [82,87]. Since nitrification is oxygen dependent and denitrification is inhibited by high oxygen concentration, the balance between the two processes and the ratio of N₂O/NO is controlled by the water-filled pore space (WFPS) in soils [127]. At about 60% WFPS, both nitrification and denitrification processes can occur, and the N₂O/NO ratio is about one with the highest soil N₂O emission. At lower WFPS, nitrification forms N₂O (40–60%) and NO (0–60%). At 60–80% WFPS, N₂O and NO are formed, and at a WFPS above 80% (water saturation), only N₂ is formed by a complete denitrification. In permafrost-affected soils, N₂O emissions were observed at WFPS values of 60–70%, mainly associated with denitrification [82]. Not only oxygen content, but also other environmental parameters control N-cycle processes in controversial ways, e.g., SOM content [59,164,165].

2.2.1. Nitrification

Nitrification, the canonical oxidation of ammonia to nitrite (ammonia oxidation) and further to nitrate (nitrite oxidation) or the complete oxidation of ammonia to nitrate (comammox-process) is mostly carried out by specialized, obligate chemolithotrophic aerobic ammonia-oxidizing bacteria (AOB) and archaea (AOA) and nitrite-oxidizing bacteria (NOB) (reviewed e.g., by [128]) occupying different environmental niches [129,166]. Nitrifiers have been studied and detected in permafrost soils since the early 1990s, mostly using culture-dependent methods [167,168]. However, few nitrifiers were detected, and soils had little or no net nitrifying activity [14,15,169,170,171]. Only isolated sites had significantly higher cell counts and activities comparable to temperate soils [15,112,171]. High net nitrification and denitrification rates correlated with the δ¹⁵N value of the soils [112]. For a long time, extreme environmental conditions such as low pH and temperatures, high water levels associated with anoxic conditions, and low N availability were considered factors inhibiting nitrification in this habitat [64,172,173]. About a decade ago, metagenomic studies contributed significantly to the understanding of the functional capacity of nitrifiers in permafrost habitats [174]. Molecular-based methods enable the detection of ammonia-oxidizing bacteria, archaea and comammox by targeting, e.g., the 16S rRNA genes or the key functional enzyme of ammonia monooxygenase (amoA) [20,110,118,119,120,175,176,177,178,179,180,181,182] and nitrite oxidoreductase gene (nrx) for comammox and nitrite oxidizers [119,155,182]. Using these methods, nitrifying genes were found in the active layer of permafrost-affected soils in high abundance, comparable to other ecosystems, but with low diversity [120,174]. Nitrifiers have also been detected in frozen permafrost cores [101,150,174,183,184] with gene abundance increasing after thawing [185] but may have been lost in ice rich sediments (Yedoma, [87,186]).

As in temperate soils [129,187], AOB amoA could dominate in Arctic soils at neutral pH values [119,178], while in acid soils mainly AOA amoA was found [120,147,181] and some soils showed no correlation with the soil pH values [118]. At neutral alluvial soils with low SOM content, AOB were mainly represented by the genus Nitrosospira, with relative abundance of nitrifying taxa accounting for up to 6% of the total microbial community (indicated by 16S rRNA). The abundance of nitrifiers correlated with potential AOB activities und enriched Nitrosopira-like AOB were characterized as psychrotolerant [119]. The genus Nitrosospira also dominated in samples from Antarctica soils [179,180,188,189] and the Canadian Arctic [163]. However, when AOA dominates, the measured nitrification activity did not correlate with abundance of the amoA gene, as some AOA clades have low nitrification activity and alternative mixotrophic growth [118]. Recently, it was shown that especially the new Candidatus AOA genus Nitrosopolaris with four species are widely distributed in polar and alpine habitats and a new AOA species Ca. Nitrosocosmicus arcticus was isolated from Arctic soils [182,190]. All ammonia-oxidizing organisms (AO) produce N₂O (and NO) as a byproduct during aerobic ammonia oxidation by different pathways (from reactive intermediates and nitrifier-denitrification) and at different amounts, with N₂O yields about 30 times higher in AOB than in AOA and comammox [129,191,192]. The global median N₂O yield from AO is 0.14% per ammonia [193]. Although AOA often outnumber AOB in soils, AOB nitrification is far more susceptible to N₂O formation [192]. Therefore, models of soil N₂O emission could be improved by including not only physiochemical parameters [194], but also microbial traits to distinguish between AO communities [195]. For the final step of nitrification, nitrite oxidation, nitrite-oxidizing bacteria (NOB) of the genera Nitrospira and Nitrotoga might play an important role in permafrost-affected soils [119,155]. To date, two psychrotolerant strains of Nitrotoga have been enriched from active layer (Nitrotoga arctica) and permafrost, both with temperature optima of 17 °C and 22 °C, respectively [155,196].

Ramm et al. [59] synthesized nitrification rates of the active layer of permafrost-affected ecosystems and estimated mean gross nitrification rates of 6.6 µg N g dw⁻¹ d⁻¹ with slightly negative or near zero net nitrification rates. Organic soils have much higher rates than minerals soils. Thus, gross nitrification rates are comparable or even higher than in synthesized data from non-permafrost ecosystems [59]. Nitrification rates were mainly controlled by soil organic carbon (highest rates at medium SOC), C/N ratio, and total N and accounted for about half of ammonification rates, indicating a great importance of autotrophic nitrification.

2.2.2. Denitrification

Nitrate formed by nitrification can be gradually reduced under anoxic conditions by heterotrophic canonical denitrification from nitrate to nitrite, to gaseous nitric oxide (NO) and nitrous oxide (N₂O), and finally to inert dinitrogen gas (N₂). This heterotrophic process uses C compounds as electron donor, therefore depends on the availability of degradable organic carbon in the soil. Denitrification is carried out by various non-specific groups of mainly prokaryotic microorganisms occupying divers ecological niches [130]. The final step of denitrification represents the only known N₂O sink in the biosphere [197]. In permafrost-affected soils, denitrifying organisms were detected based on their various key enzyme genes (nitrate reductases: narGHI, napAB, nitrite reductases: nirS, nirK, NO reductase: norVW, norB and N₂O reductase: nosZ [136]). In general, the main product of denitrification is N₂, but in the absence of N₂O reductase, N₂O is formed. The N₂O/N₂ ratio determines how much N₂O is formed and is regulated by the denitrifying community. High N₂O/N₂ ratios can result from high abundance of denitrifying fungi lacking N₂O reductase [198], or from favorable soil conditions, such as carbon availability or acidity [199,200]. On a global scale, N₂O accounts for about 8% of the terrestrial denitrification flux, and thus has a higher yield than nitrification [193,201]. In permafrost-affected soils, denitrification is the main process of N₂O emission, but nitrification can also be a major contributor (20% [202]) or even dominate ([203], 86% [163]).

Studies of in-situ denitrification rates in permafrost-affected soils are rare (e.g., [112,204]) but have recently been reviewed [58]. Denitrifying communities of permafrost-affected soils remain largely unresolved to date, with one of the first characterizations of denitrifiers occurring in N hotspots of bare permafrost peatlands and sea animal-influenced soils (details see chapters below). In both N hotspots, nirS-type (nitrite reductase) denitrifiers play an important role [189,205], as well as in permafrost-affected soils on the Tibet Plateau [110]. Metatranscriptomic analyses of active layer in Arctic cryosols at representative ice-wedge polgygon in the Canadian high Arctic polar desert detected active denitrifiers using the marker gene nirS, but most of these have not yet been cultured [136]. In this study, the known denitrifiers were members of the genera Thiobacillus, Denitrovibrio, Pseudomonas, Azospirillum, and Azorhizobium, and soil N₂O emission correlated with a particular denitrifier community [136]. At one N hotspot in hillslope thermokarst landscape that produces large amounts of N₂O, denitrification genes accounted for 15–29% (nirS + nirK) of the analyzed N functional genes [87]. A recent metagenomic analysis of tundra soils in northern Finland showed that dentrifiers were dominated by poorly characterize taxa but had truncated denitrification enzymes that lacked one or more denitrification genes, leading to the hypothesis that these organisms contribute significantly to N₂O cycling in tundra soils [181]. However, the study did not find a significant relationship between N₂O flux and the abundance of denitrifying genes. In addition, permafrost samples contained denitrifying genes, as shown in metagenome studies [135]. As in the active layer, the relative abundance of the last denitrification step was low in frozen permafrost, which may lead to the accumulation of N₂O [101,174,183]. The gene abundance of denitrifiers increases after thawing [185], but not during long-term incubations [206].

3. Hotspots of N Availability and Properties

Nitrogen availability is primarily determined by plant N demand. Therefore, hotspots and hot moments of N excess result from lack of competition due to a spatial or temporal separation of N sources and sinks with high N loss potential as N₂O (reviewed in [82]) or nitrate [28,84,207,208,209,210,211,212,213].

Nitrogen hotspots occur, e.g., in the absence of vegetation [30,86,120,211,214,215,216,217,218], with pioneer vegetation following disturbance [87,213] where ecosystem disturbance from wind erosion, cryoturbation, thermokarst, fluviatile erosion, or wildfire have resulted in vegetation loss, or in deep active layers and permafrost without deep-rooting plant species [55,75,76,103,219]. Accordingly, hot moments also occur when plant N demand and immobilization are reduced [28,30,55,59,75,103,217,220,221,222,223], e.g., during spring thaws prior to N uptake by plants [24,40,224,225,226,227] or at the end of the growing season [28,228,229]. A temporal separation of N mineralization and N immobilization was observed in the Arctic tundra [75,230]: In the winter months, N mineralization predominates and DIN is enriched, while in the growing season N immobilization predominates [70,104,224]. Consequently, at the beginning of the growing season, the thawing of frozen DIN, temperature-induced increase in N mineralization, and presumably frequent freeze-thaw cycles associated with microbial cell lysis may result in the substantial release of plant-available DIN [66,75,110,225,231,232,233].

However, N hotspots can also result from additional N input from animal excreta [188,189,234,235,236,237], plants associated with diazotrophic bacteria [111], or N release from thawing permafrost [87,213,238]. Since N hotspots have only been discovered in the last two decades, only a limited number of studies have been conducted at a few sites (Figure 1) that describe their biogeochemical characteristics in relation to N losses. The two best described hotspot types are (1) N hotspots in bare organic lowland soils in permafrost peatlands, where high in situ emissions of N₂O from permafrost environments were initially detected [214], and (2) disturbed mineral soils at retrogressive thaw slumps, where N leaching was first found [28,208,210] (Figure 3). In the following sections, we detail what is known about N availability and the correlation between potential N traits in combination with N losses. In addition, at the end this chapter, we characterize other recently discovered and less well documented probable N hotspots, (3) alluvial soils, (4) animal-influenced, and (5) wildfire-affected soils in permafrost-ecosystems (Figure 4).

3.1. Bare Organic Lowland Soils in Permafrost Peatland

3.1.1. Habitat, Soil Characteristics and Inorganic N

The high content of ground ice in permafrost peatlands leads to the formation of various shapes of patterned ground and ice wedge polygonal networks, as well as elevated mounds, such as palsas and peat plateaus. Northern permafrost peatlands cover an area of about 1.7 ± 0.5 million km² and store about 185 ± 70 Pg organic carbon and 7 ± 4 Pg nitrogen in their soils [33]. Altogether, northern peatlands are at present-day a sink for CO₂ (0.10 ± 0.02 Pg y⁻¹) as well as a source of CH₄ (0.026 ± 0.002 Pg y⁻¹) and of N₂O (0.022 ± 0.005 Tg y⁻¹, but based on data from only a limited number of sites) with C and N losses to aquatic systems of 0.022 ± 0.02 Pg C y⁻¹ and 0.7 ± 0.5 Tg N y⁻¹ [33]. Permafrost peatlands with ombrotrophic palsa and peat plateaus surrounded by minerotrophic unfrozen fen surfaces represent a typical permafrost-affected ecosystem at sporadic and discontinuous permafrost zone in subarctic regions, in contrast to ice-wedge polygons that occur in the continuous permafrost regions [11,239,240]. While in the sporadic permafrost zone palsa mires occur with a large proportion of the mire surface without permafrost, in more northerly with higher permafrost cover extensive frozen peat plateaus occur surrounded by only narrow fen stripes [11] (Figure 3 and Figure 4). Uplifted peat surfaces have often bare, non-vegetated patches of varying sizes-called peat circles up to 25 m in diameter-created by the expansion of the ice body and the resulting surface uplift, erosion (wind abrasion) or cryoturbation [214,241,242,243] (Figure 3 and Figure 4). Due to higher elevation compared to the non-frozen mire surface, the palsas and peat plateaus are well-drained, ombrotrophic and the vegetated parts have typical bog vegetation [86]. The uppermost peat is usually old, minerotrophic and only partially covered by a thin layer of moss, but without any vascular plants [86,244]. Assuming a spatial coverage of peat plateaus (and palsas) of about 20% in the Artic [214] and a 1–4% proportion of bare sites in these uplifted permafrost peatlands, the area of bare peat plateaus in the Arctic could be 0.015–0.06 million km² [214]. However, the precise identification of these small-scale features on a pan-Arctic scale is difficult. Bare peat surfaces on palsa and peat plateaus are typical landscapes in permafrost peatlands in Canada [245], Europe [240], northwestern Russia, and Siberia [241,242]. However, so far, only a few bare peat sites in Finland and eastern Russia have been studied for their N properties, N turnover, and microorganisms involved (Figure 1), which are described below.

The bare peat soils in uplifted palsas and peat plateaus have been termed as Cryic Folic Histosol [246], and the surfaces are characterized by old peat material (age up to 6000 years) with a high degree of peat decomposition [243,244]. The bare peat surfaces have higher total labile carbon and nitrogen, lower C/N ratio, lower pH of about 3–4, higher bulk density, and lower phosphate content, unlike the surrounding vegetated peat soils (Fibric Histosols) [86,202,216,241,242,244,246,247]. In addition, the water table in the palsas and peat plateaus is lower than in the surrounding peatlands without permafrost, with the water content as water-filled pore space (WFPS) in the bare peat surfaces often at 60–70% [86,202,217], i.e., in the intermediate range favorable for N₂O emission. Oxygen concentration in bare peat circles of a Russian peat plateau varied between 17–20% in the first 35 cm and was <1% at 45 cm depth, allowing simultaneous aerobic microbial processes in the topsoil and anaerobic processes in deeper parts [217]. In general, the bare peat circles had higher concentrations of inorganic nitrogen ammonium and nitrate at lower DOC/DIN ratios than the vegetated parts of the peat plateau [86,120,202,214,216,247,248]. For the Seida site in Russia, the mean values of all studies for ammonium were 60 ± 14 µg NH₄⁺-N g dw⁻¹ (Table 1) with the highest mean of 116 ± 39 µg NH₄⁺-N g dw⁻¹, [202] (Table S1). Meanwhile, for nitrate, the mean values were 117 ± 3 µg NO₃⁻-N g dw⁻¹ (Table 1 and Table S2, Figure 5C) with the highest mean of 419 ± 54 µg NO₃⁻-N g dw⁻¹, [247] (Table S1). Nitrate concentration in bare peat was even two orders of magnitude higher than ammonium with concentrations up to 1 mg NO₃⁻-N g dw⁻¹, and with increasing soil depth nitrate content decreased and ammonium increased [217].

3.1.2. C and N Mineralization

While the vegetated subsites of the peat plateau are carbon sinks and have the highest SOC stock per unit area in the region, the bare peat circles act as carbon sources [249,250,251]. Peat circles exhibited comparable in situ soil microbial respiration to vegetated parts despite the presence of old carbon [243], indicating that peat is highly available for decomposition [223,244,247] (Table S1). However, in laboratory incubations, peat circles had lower C-mineralization rates than vegetated sites [223,242] (27 µg compared to 92 µg CO₂-C g dw⁻¹ d⁻¹ in incubations at 15 °C [244] (Table S1). Bare peat sites are usually small or negligible CH₄ sources [251] or even sinks [252] in contrast to waterlogged, surrounded fens with high CH₄ emissions. The ability of bare palsa surface for atmospheric CH₄ uptake was also demonstrated under dry conditions in laboratory incubations with intact peat cores [223]. Peat circles were found to have lower microbial biomass with low diversity of specialized microorganism for decomposition of SOM [242]. In anoxic incubation experiments, fermentation potentials are lower for bare peat circles than for the vegetated sites [205,253]. The soils of peat circles were inhabited by the bacterial class-level taxa Alphaproteobacteria, Acidimicrobiia, Acidobacteria, Verrucomicrobiae, and Bacteroidia, as well as by archaeal Nitrososphaeria and non-methanogenic Euryarchaeota [253]. Cryoturbation destroys plant roots and leads to suppression of filamentous fungi, so more unicellular fungal forms were found [242]. In contrast to C mineralization, bare peat surfaces show higher gross N mineralization rates (16.8 ± 9.7 µg N g dw⁻¹ d⁻¹, mean rates of [86,202]) in oxic laboratory incubation experiments at 15 °C and in situ ¹⁵N labeling experiments than sparse and vegetated sites on the peat plateau (9.0 ± 10.8 µg N g dw⁻¹ d⁻¹ mean rates of [86,202]) (Table 1). Gross nitrification [120,202] (Table 1 and Table S1) and in situ net N mineralization [202,216] (Table S1) were also higher in the bare peat plateau than in the vegetated one. Net N mineralization rates in peat circles were an order of magnitude lower than gross rates for peat circles [216], indicating a surplus of available N in bare peat circles [202]. The gross nitrification rates of bare peat circles (8.4 ± 7.3 µg N g dw⁻¹ d⁻¹ mean rates of [120,202]) were in the upper range of those reported for permafrost-affected soils (mean 6.6 µg N g dw⁻¹ d⁻¹ in [59]) and in the middle range of other ecosystems [164], while vegetated sites on the peat plateau (0.2 ± 0.1 µg N g dw⁻¹ d⁻¹, [202]) were in the lower range [59] despite the low pH values found (Table 1).

3.1.3. Gaseous N Loss

In conjunction with high N turnover rates, peat circles show exceptionally high N₂O rates of 0.89 ± 0.10 g N₂O-N m⁻² per year (and a daily emission of 6.55 ± 2.83 mg N₂O-N m⁻²) during the 100-day growing season, being the highest N₂O emission of any natural terrestrial ecosystems [86,214] (Table S1, Figure 4A). The first observation of high N₂O emissions from permafrost peatlands represented the highest N₂O emissions from northern pristine terrestrial ecosystems and was estimated at the time to be up to 0.6% of global annual N₂O emissions [214]. In one study, the N₂O emissions from the peat circle were approximately equal to the net N mineralization, such that almost all of the N surplus was lost to N₂O emission [216]. N₂O concentrations in theses soils were 0.35–5 ppm and increased with depth [217]. Compared to the mean value of N₂O fluxes of 1.98 ± 3.19 mg N m⁻² d⁻¹ from various studies on peat circles at Seida site in Russia, N₂O emissions from vegetated peat plateaus (0.01 ± 0.03 mg N m⁻² d⁻¹) and surrounding water-logged fens, and from upland soils in tundra heath were negligible or soils even consumed N₂O [86,120,202,214,254] (Table 1 and Table S2, Figure 5A,B). In sparsely vegetated peat mounds, the N₂O emissions were averaged only 12% of those from peat circles [86]. The bare soils of the peat plateau also emitted N₂O outside the growing season, with cold season emissions accounting for 20–69% of annual peat circles emissions [86]. In two seasons studied, bare sides produced N₂O at the beginning of the growing season, and after N₂O decreased at the end of the growing season, there was a significant increase of CO₂ and CH₄ in the soil [247]. Nitrous oxide emission of bare sites during growing season correlated positively with air and soil temperature, total N content, inorganic N and nitrate content, WFPS, pH and C/N ratio with emission measured only below a threshold of C/N ratio <25 [86,120,202,247]. Gil et al. [217] additionally found a positive correlation between N₂O concentration and ammonium and CO₂ concentration and a negative one with nitrate concentration in the soil profile. Recently, bare peat sites on permafrost peatlands with high availability of inorganic N have also been identified as important sources of nitric oxide (NO, 1.42 ± 0.13 mg N m⁻² d⁻¹) and nitrous acid (HONO, 0.06 ± 0.02 mg N m⁻² d⁻¹) [255] (Table S1, Figure 4A). As with N₂O, low emissions were detected from adjacent vegetated surfaces. This was the first report of these gases from high latitudes soils. At the maximum, reactive N gas fluxes were similar to N₂O emission rates from these permafrost peatlands and about twice as high as average global emissions. Both NO and HONO, like N₂O, are formed by microbial nitrification and denitrification processes. Reactive N gas fluxes correlated negatively with C/N ratio and positively with total inorganic N, nitrate, and ammonium content with variable moisture effects.

Table 1. Soil properties, gross and net N turnover rates, abundance of functional N genes, and in situ N₂O fluxes at two types of N availability hotspots: exemplified by a best described site of bare surfaces on permafrost peatland and thermokarst disturbed and revegetated retrogressive thaw slump (RTS), and undisturbed sites associated with the hotspots: vegetated surfaces on permafrost peatland and intact, fully vegetated soil adjacent to thaw slump, respectively. For RTS sites, a distinction was made between two subsites: slump floor (SF) and thaw mound (TM).

	Permafrost Peatland		Mineral Upland Soils in Thermokarst Landscapes
Soil and Microbial Properties	Bare Surfaces *	Vegetated Surfaces *	Disturbed, Revegetated RTS **		Undisturbed Fully Vegetated Site Next to RTS **
			Slump Floor (SF)	Thaw Mound (TM)
pH	3.7 ± 0.5 [86,247]	3.6 ± 0.3 [86,247]	7.2 ± 1.5 [87,256]	7.9 ± 0.0 [256]	5.7 ± 0.3 [87]
C/N	22 ± 4 [86,247]	51 ± 16 [86,247]	15 ± 1 [87,256]	13 ± 0 [256]	38 ± 4 [87]
WFPS (%)	53 ± 25 [86,247]	22 ± 7 [86,247]	60 ± 11 [87,256]	60 ± 3 [256]	13 ± 6 [87]
SOM (%)	95 ± 1 [86,247]	98 ± 0 [86,247]	11 ± 4 [87,256]	8 ± 0 [256]	27 ± 3 [87]
TN (%)	2.4 ± 0.3	1.0 ± 0.3	0.30 ± 0.10	0.34 ± 0.01	0.29 ± 0.03
	[86,247]	[86,247]	[87,256]	[256]	[87]
δ15N in bulk soil (‰)	n.d.	n.d.	1.4 ± 0.5 [87,256]	2.0 ± 0.1 [256]	1.2 ± 0.3 [87]
Ammonium (µg N g dw−1)	60.1 ± 14.4	19.6 ± 5.7	1.4 ± 1.8	0.0 ± 0.0	9.0 ± 11.7
	[86,202,205,216,247,255]	[86,202,205,216,247,255]	[87,256]	[256]	[87]
Nitrat (µg N g dw−1)	116.8 ± 2.8	4.4 ± 7.7	0.7 ± 0.1	81.6 ± 24.3	0.0 ± 0.0
	[86,202,205,216,247,255]	[86,202,205,216,247,255]	[87,256]	[256]	[87]
DIN/TN (%)	1.1 ± 1.3 [86,247]	0.3 ± 0.1 [86,247]	0.1 ± 0.1 [87,256]	2.4 [256]	0.3 [87]
Gross N mineralization	16.8 ± 9.7 [86,202]	9.0 ± 10.8 [86,202]	15.1 ± 10.1 [87]	n.d.	b.d.
(µg N g dw−1 d−1)
Net N mineralization	n.d.	n.d.	4.0 ± 4.9 [87,256]	1.4 ± 0.5 [256]	−20.9 ± 16.6 [87]
(µg N g dw−1 d−1), p.a.s. 0.8 [59]
Gross nitrification	8.4 ± 7.3 [120,202]	0.2 ± 0.1 [202]	b.d.	n.d.	b.d.
(µg N g dw−1 d−1), p.a.s. 6.6 [59]
Net nitrification	n.d.	n.d.	1.6 ± 1.5 [87,256]	1.4 ± 0.5 [256]	−6.0 ± 2.9 [87]
(µg N g dw−1 d−1), p.a.s. −0.4 [59]
Net denitrification	0.56	<0.004	2.8 ± 1.8 [87]	n.d.	0.01 ± 0.00
with (without acetylene) (µg N g dw−1 d−1)	(0.37) [205]	(b.d.) [205]	(0.3 ± 0.4) [87,256]	(0.20 ± 0.02)	(0.01 ± 0.00)
				[256]	[87]
Functional nitrification gene
AOA amoA (copies gdw−1)	6.4 × 108 [120]	8.0 × 106 [120]	2.0 × 107 [87]	n.d.	5.4 × 107 [87]
AOB amoA (copies gdw−1)	b.d.	b.d.	3.3 × 107 [87]	n.d.	4.2 × 106 [87]
Functional nitrification gene amoA (% of 16S rRNA)
	n.d.	n.d.	7 [87]	n.d.	1 [87]
amoA (% of N genes)	n.d.	n.d.	3.4 ± 1.0 [87]	n.d.	0.5 ± 1.0 [87]
Functional denitrification gene (% of 16S rRNA [205], N genes [87])
narG	7.6 ± 2.8 [205]	0.04 ± 0.01 [205]	42 ± 0.9 [87]	n.d.	45 ± 9.3 [87]
nirS + nirK	0.34 ± 0.08 [205]	0.88 ± 0.13 [205]	29 ± 0.1 [87]	n.d.	15 ± 0.6 [87]
(nirS + nirK)/nosZ (%/%)	0.20 × 103 [205]	8.88 × 103 [205]	2.4 ± 0.2 [87]	n.d.	0.5 ± 0.1 [87]
N2O fluxes	1.98 ± 3.19	0.01 ± 0.03	1.64 ± 2.61 [87]	n.d.	−0.001 ± 0.018 [87]
(mg N m−2 d−1)	[86,120,202,247,255]	[86,120,202,247,255]

* Peat plateau in Seida, Komi Republic, Russia (N 67°03′, E 62°57′, Figure 1 and Figure 2), bare peat surfaces covered by a thin moss layer but lacking vascular plants, vegetated site with typical bog vegetation [86]. ** Retrogressive thaw slump, Kurungnakh Island, Lena Delta, Russia (N 72°20, E 126°17′, Figure 1 and Figure 2), undisturbed site: Holocene cover deposits with moss, sedge and dwarf shrub vegetation communities on top of the RTS, disturbed thermokarst site: Yedoma revegetated with grasses (Arctagrostis arundinacea) subdivided into slump floor (SF, SF3V in [256]; SF3 in [257], Yedoma mixed with younger Holocene) and thaw mounds (TM, TM1V in [256] dominated by late Pleistocene Yedoma) as described in Marushchak et al. [87]. Data are mean values ± standard division calculated from the mean values of the cited publications, therefore they are mainly variations between sampling years. All gas fluxes were measured with static chamber technique, more detailed parameters and information can be found in Table S1. amoA = gene of subunit A of ammonia monooxygenase, AOA = ammonia-oxidizing archaea, AOB = ammonia-oxidizing bacteria, b.d. = below detection limit, dw = dry weight, DIN = dissolved inorganic nitrogen, narG = gene of nitrate reductase, n.d. = not determined, nirS, nirK = genes of nitrite reductases, nosZ = gene of N₂O reductase, p.a.s. = mean rates of permafrost-affected soils from Ramm et al. [59], SOM = soil organic matter, TN = total nitrogen, WFPS = water-filled pore space.

Table 1. Soil properties, gross and net N turnover rates, abundance of functional N genes, and in situ N₂O fluxes at two types of N availability hotspots: exemplified by a best described site of bare surfaces on permafrost peatland and thermokarst disturbed and revegetated retrogressive thaw slump (RTS), and undisturbed sites associated with the hotspots: vegetated surfaces on permafrost peatland and intact, fully vegetated soil adjacent to thaw slump, respectively. For RTS sites, a distinction was made between two subsites: slump floor (SF) and thaw mound (TM).

	Permafrost Peatland		Mineral Upland Soils in Thermokarst Landscapes
Soil and Microbial Properties	Bare Surfaces *	Vegetated Surfaces *	Disturbed, Revegetated RTS **		Undisturbed Fully Vegetated Site Next to RTS **
			Slump Floor (SF)	Thaw Mound (TM)
pH	3.7 ± 0.5 [86,247]	3.6 ± 0.3 [86,247]	7.2 ± 1.5 [87,256]	7.9 ± 0.0 [256]	5.7 ± 0.3 [87]
C/N	22 ± 4 [86,247]	51 ± 16 [86,247]	15 ± 1 [87,256]	13 ± 0 [256]	38 ± 4 [87]
WFPS (%)	53 ± 25 [86,247]	22 ± 7 [86,247]	60 ± 11 [87,256]	60 ± 3 [256]	13 ± 6 [87]
SOM (%)	95 ± 1 [86,247]	98 ± 0 [86,247]	11 ± 4 [87,256]	8 ± 0 [256]	27 ± 3 [87]
TN (%)	2.4 ± 0.3	1.0 ± 0.3	0.30 ± 0.10	0.34 ± 0.01	0.29 ± 0.03
	[86,247]	[86,247]	[87,256]	[256]	[87]
δ15N in bulk soil (‰)	n.d.	n.d.	1.4 ± 0.5 [87,256]	2.0 ± 0.1 [256]	1.2 ± 0.3 [87]
Ammonium (µg N g dw−1)	60.1 ± 14.4	19.6 ± 5.7	1.4 ± 1.8	0.0 ± 0.0	9.0 ± 11.7
	[86,202,205,216,247,255]	[86,202,205,216,247,255]	[87,256]	[256]	[87]
Nitrat (µg N g dw−1)	116.8 ± 2.8	4.4 ± 7.7	0.7 ± 0.1	81.6 ± 24.3	0.0 ± 0.0
	[86,202,205,216,247,255]	[86,202,205,216,247,255]	[87,256]	[256]	[87]
DIN/TN (%)	1.1 ± 1.3 [86,247]	0.3 ± 0.1 [86,247]	0.1 ± 0.1 [87,256]	2.4 [256]	0.3 [87]
Gross N mineralization	16.8 ± 9.7 [86,202]	9.0 ± 10.8 [86,202]	15.1 ± 10.1 [87]	n.d.	b.d.
(µg N g dw−1 d−1)
Net N mineralization	n.d.	n.d.	4.0 ± 4.9 [87,256]	1.4 ± 0.5 [256]	−20.9 ± 16.6 [87]
(µg N g dw−1 d−1), p.a.s. 0.8 [59]
Gross nitrification	8.4 ± 7.3 [120,202]	0.2 ± 0.1 [202]	b.d.	n.d.	b.d.
(µg N g dw−1 d−1), p.a.s. 6.6 [59]
Net nitrification	n.d.	n.d.	1.6 ± 1.5 [87,256]	1.4 ± 0.5 [256]	−6.0 ± 2.9 [87]
(µg N g dw−1 d−1), p.a.s. −0.4 [59]
Net denitrification	0.56	<0.004	2.8 ± 1.8 [87]	n.d.	0.01 ± 0.00
with (without acetylene) (µg N g dw−1 d−1)	(0.37) [205]	(b.d.) [205]	(0.3 ± 0.4) [87,256]	(0.20 ± 0.02)	(0.01 ± 0.00)
				[256]	[87]
Functional nitrification gene
AOA amoA (copies gdw−1)	6.4 × 108 [120]	8.0 × 106 [120]	2.0 × 107 [87]	n.d.	5.4 × 107 [87]
AOB amoA (copies gdw−1)	b.d.	b.d.	3.3 × 107 [87]	n.d.	4.2 × 106 [87]
Functional nitrification gene amoA (% of 16S rRNA)
	n.d.	n.d.	7 [87]	n.d.	1 [87]
amoA (% of N genes)	n.d.	n.d.	3.4 ± 1.0 [87]	n.d.	0.5 ± 1.0 [87]
Functional denitrification gene (% of 16S rRNA [205], N genes [87])
narG	7.6 ± 2.8 [205]	0.04 ± 0.01 [205]	42 ± 0.9 [87]	n.d.	45 ± 9.3 [87]
nirS + nirK	0.34 ± 0.08 [205]	0.88 ± 0.13 [205]	29 ± 0.1 [87]	n.d.	15 ± 0.6 [87]
(nirS + nirK)/nosZ (%/%)	0.20 × 103 [205]	8.88 × 103 [205]	2.4 ± 0.2 [87]	n.d.	0.5 ± 0.1 [87]
N2O fluxes	1.98 ± 3.19	0.01 ± 0.03	1.64 ± 2.61 [87]	n.d.	−0.001 ± 0.018 [87]
(mg N m−2 d−1)	[86,120,202,247,255]	[86,120,202,247,255]

* Peat plateau in Seida, Komi Republic, Russia (N 67°03′, E 62°57′, Figure 1 and Figure 2), bare peat surfaces covered by a thin moss layer but lacking vascular plants, vegetated site with typical bog vegetation [86]. ** Retrogressive thaw slump, Kurungnakh Island, Lena Delta, Russia (N 72°20, E 126°17′, Figure 1 and Figure 2), undisturbed site: Holocene cover deposits with moss, sedge and dwarf shrub vegetation communities on top of the RTS, disturbed thermokarst site: Yedoma revegetated with grasses (Arctagrostis arundinacea) subdivided into slump floor (SF, SF3V in [256]; SF3 in [257], Yedoma mixed with younger Holocene) and thaw mounds (TM, TM1V in [256] dominated by late Pleistocene Yedoma) as described in Marushchak et al. [87]. Data are mean values ± standard division calculated from the mean values of the cited publications, therefore they are mainly variations between sampling years. All gas fluxes were measured with static chamber technique, more detailed parameters and information can be found in Table S1. amoA = gene of subunit A of ammonia monooxygenase, AOA = ammonia-oxidizing archaea, AOB = ammonia-oxidizing bacteria, b.d. = below detection limit, dw = dry weight, DIN = dissolved inorganic nitrogen, narG = gene of nitrate reductase, n.d. = not determined, nirS, nirK = genes of nitrite reductases, nosZ = gene of N₂O reductase, p.a.s. = mean rates of permafrost-affected soils from Ramm et al. [59], SOM = soil organic matter, TN = total nitrogen, WFPS = water-filled pore space.

3.1.4. Microbial Based N Processes

In anoxic incubations, samples from cryoturbated bare peat circles show high N₂O production rates (0.37 µg N₂O-N g dw⁻¹ d⁻¹ and 0.56 µg N₂O-N g dw⁻¹ d⁻¹ at 20 °C in the absence and presence of acetylene, respectively), but not from adjacent vegetated sites (Table 1 and Table S2, Figure 5E,F), suggesting anaerobic denitrification as the main N₂O source and the potential to be responsible for the in situ N₂O emissions [205]. Instead, soils samples from vegetated sites showed high N₂O consumption potential in incubations [254]. However, samples from vegetated sites produced N₂O after the addition of nitrite and nitrate. Therefore, the content of inorganic N had limited denitrification at vegetated sites [202,205,254]. Since N₂O production from peat circles was not stimulated by substrate, these samples were not N-limited [205]. N₂O emissions from peat circles were associated with a specific, diverse, and acid-tolerant denitrifier community that had different compositions at sites with low and high N₂O emissions [205,253,258]. Concentrations of key denitrification enzymes were significantly higher in peat circles than in vegetated sites, with, e.g., nitrate reductase (narG) accounting for about 8% of bacterial 16S rRNA (Table 1) and the nirS-type (nitrite reductase) denitrifier playing a major role. The high emission of N₂O could be caused by a specialized denitrifying community with a high percentage of denitrifiers that lack the final step to produce N₂ from N₂O, as shown by very low amount of the N₂O reductase gene (nosZ) [205]. Denitrifiers found in bare and vegetated soils belong to Actino-, Alphaproteo-, and Betaproteobacteria, but most key enzyme sequences are only distantly related to cultured denitrifiers, so they are not yet cultured and their ecophysiological properties are therefore unknown [205,254]. At bare sites, acetate consuming key players of the Burkholderiaceae family coexist with Rhodanobacter sp. during complete denitrification, which under natural conditions might be limited of electron donors [253]. In situ ¹⁵N labeling studies at the same bare peat circle have shown that aerobic ammonia-oxidizing nitrification is responsible for about 20% and denitrification processes for about 80% of the emitted N₂O, which was regulated by moisture (WFPS) and thus oxygen status [202,217]. The process of nitrifier denitrification could be excluded as no ammonia-oxidizing bacteria (AOB) were detected [120,253]. Therefore, the 20% N₂O produced by ammonia oxidation were abiotically formed from intermediates or byproducts of ammonia oxidation pathway of the detected ammonia-oxidizing archaea (AOA). These nitrifiers were described to be adapted to acid soil conditions despite low concentrations of their actual substrate ammonia [129]. Although nitrifiers contribute only slightly to N₂O production in bare peat circles, nitrification is the most important limiting step in N₂O emission, as it regulates the supply of nitrite and nitrate for denitrification.

In addition, in four Arctic and subarctic permafrost peatlands in Russia and Finland, AOA were most abundant in bare peat soils compared to vegetated parts, and neither AOB nor comammox were detected [120] with only 0.01% abundance of bacterial nitrifiers of 16S rRNA [253]. In addition, high gross nitrification rates at bare sites with up to 13.5 ± 1.1 µg NO₃⁻-N g dw⁻¹ d⁻¹ (in incubations at 15 °C, Table S1) were assigned only to AOA by inhibition experiments [120], which could be due to the low pH values that generally favor AOA over AOB [129,190,259,260,261] and, in peat soils [262]. On average, the abundance of amoA genes was about 200 times higher in bare peat than in vegetated peat surfaces [120]. AOA showed low diversity. Only two phylotypes of clades NS-ζ-1.2 and NS-γ−2.3.2 clades were detected associated with the taxonomic order Nitrososphaerales, with the closest relatives Candidatus Nitrosocosmicus arcticus Kfb [190]. Nitrate concentrations correlated positively with AOA (amoA) gene abundance and N₂O fluxes. Due to the high availability of inorganic N in the peat circles, biological N_2-fixation (BNF) and N fixing cyanobacteria were absent there, in contrast to the vegetated parts of the peat plateau [216]. Therefore, the high N mineral cycle is fed only by internal N sources of N mineralization and no additional external N input from BNF.

In summary, all soil parameters of the bare peat sites indicate high N availability, i.e., N hotspot character, caused by high mineralization relative to N immobilization, which together with intermediate moisture supports high gaseous N losses through active nitrification and denitrification processes (Figure 2).

3.2. Disturbed Mineral Upland Soils in Hillslope Thermokarst Landscapes

Similar to bare peat surfaces caused by wind erosion, high N losses by gaseous fluxes and/or leaching have been reported from thermokarst-influenced upland soils, where N availability is increased by the N release from permafrost layers after thawing, but also by loss of plant cover [28,42,75,103,213] (Figure 3 and Figure 4). This disruption of the N cycle is long-lasting and is not reversed by initial revegetation with plants [87,125,213,218,263,264,265,266,267].

3.2.1. Habitat and Soil Characteristics

Thermokarst is defined as a process that causes subsidence of the ground surface due to thawing of ice-rich permafrost [11,97,268,269,270]. About 20% of Northern Hemisphere permafrost areas are currently affected by thermokarst, with a substantial increase projected by the end of the century (to 20–50% for upland areas [271,272,273]). Thermokarst landscapes can be divided into wetland, lake and hillslope thermokarst [97,270]. Hillslope thermokarst currently covers only a small portion of about 5% of the northern permafrost region [270] but its spatial extent is expected to increase in the future, and it has the potential to emit about one-third of total carbon loss associated with abrupt thaw [99]. Upland thermokarst landforms are divided into three types: retrogressive thaw slumps (RTS), the most widespread and active form, active layer detachment (ALD) and thermal erosional gullies, collectively referred to as thermo-erosional features [11,97,268,269,270]. Retrogressive thaw slums are particularly common along riverbanks, coasts, and thermokarst lakes in the American Northwest and in the Siberian Arctic, but are also found on slopes on the Tibetan Plateau, where the ground is underlain by massive permafrost in areas of continuous permafrost [11,268,270,274,275,276,277,278] (Figure 1). Unlike the gradual thaw, which affects permafrost only centimeters at a time over decades, the abrupt thaw affects many meters at daily and annual intervals [275,279,280,281] and causes long-lasting feedbacks on hydrology, soil biochemical properties, redox conditions, soil and vegetation cover, SOM decomposition and erosion [125,213,218,263,264,265,266,267]. In particular, the ice-rich (up to 80 vol-%) Yedoma sediments deposited during the Weichselian Ice Age (≤MIS 5) in the Late Pleistocene are susceptible to abrupt permafrost thaw, leading to the formation of RTS in eastern Siberia. These loess-dominated silty and fine sandy sediments, formed under cold, arid conditions, store large amounts of N, as indicated by a low C/N ratio (median of 9.3) and easily degradable SOM and organic matter content of 4–10% [3,42,239,272,282,283,284,285].

In the Yedoma regions, RTS can be described by the main morphological components: (1) a steep, nearly vertical icy headwall that can be up to 25 m high, (2) a slump floor with a low slope (typically 2–15°) and (3) the exposed thaw mounds (in Russian ‘baydjarakhs’) within the slump floor (Figure 3). While the slump floor contains eroded and mixed Pleistocene permafrost, Holocene active layer sediments, and remnants of surface vegetation [213,257,269,286,287], thaw mounds contain only Pleistocene permafrost sediments as remnants of former ice-wedge polygon centers [257,280,287,288]. Soils of RTS were classified as Protic Cryosols due to the presence of permafrost and lack of soil horizon development, meaning that organic horizon is reduced or absent in homogenous soil material [269,286,287,289]. Compared to undisturbed permafrost, elevated soil temperatures and greater active layer thickness can be observed on non-vegetated exposed soils [88,218,269,286,290] with increased UV light exposure and photodegradation of organic matter [276]. After several decades of erosion, disturbed soils from RTS and hermos-erosion gullies still have lower SOM content with lower DOC and total nitrogen, lower soil moisture (leading to adjacent oxic and anoxic conditions), higher pH (around 7), and higher bulk density than undisturbed soils [29,88,213,218,266,269,286,289,290,291,292].

Different thaw stages of thermokarst features form a chronosequence from bare, freshly eroded to already stabilized areas revegetated by pioneer plants. Stabilization of RTS may occur when the headwall is no longer affected by coastal waves, when vegetation has recovered, or when other eroded sediments protect the ground-ice from thawing [269]. The vegetation on disturbed sites differs from that on undisturbed sites, which persists for centuries [11,125,218,264,266,269,289,290,291,293,294]. In the initial phase, erosion usually mobilizes large amounts of SOM into aquatic ecosystems, especially readily biodegradable carbon and nitrogen in the form of DOC and DON [210,281,293,295]. Based on model simulations, thermal erosional events are assumed to result in a net loss of 10% carbon and 5–7% of nitrogen and phosphorus in the ecosystem within 100 years [67]. Therefore, despite high N availability in the early decades, SOM and vegetation renewal may be nutrient limited at later vegetation stages [67].

3.2.2. C and N Mineralization and Inorganic N

Upland thermokarst exposes SOM to mineralization from long-term frozen Holocene and Pleistocene deposits, such as the ice-rich sediments of Yedoma. The reported decomposability of Yedoma is inconsistent, suggesting greater variability between geographic locations or microbial limitation of C mineralization in recently thawed sediments [186,296]. Lower, similar, or higher rates of C mineralization and thus gaseous C released as CO₂ and CH₄ have been described at hillslope thermokarst landforms compared to undisturbed neighboring sites [88,257,276,277,286,287,297]. In comparison to undisturbed soils, soils in hillslope thermokarst landscapes are generally characterized by low C/N ratios and higher but highly variable inorganic N availability, dominated by ammonium, [29,87,125,213,256]. The increased inorganic N content is most likely due to the following: (1) thermo-erosion exposes mineral soils and thaws permafrost subsoil layers, which typically contain higher concentrations of inorganic nitrogen (mainly ammonium) than topsoils [28,42,77,103,117,213], (2) physical mixing increases aeration and drainage, which promotes aerobic N mineralization [186,208,213], (3) reduced or absent uptake of plant nutrients due to disturbed vegetation. When permafrost thaws, the previously conserved ammonium and, to a lesser extent nitrate, are diluted with melted ice and mixed with freshly mineralized ammonium [213]. In general, the concentration of total dissolved nitrogen increases significantly in the early stage of thermokarst and decreases in the late phase [125,213,292] in association with decreasing SOC and total N [125]. A higher bioavailability of nitrogen at early stages is also indicated by early colonization of RTS by nitrophilous grasses (e.g., Alopecurus alpinus) [213,294]. The imbalance between the C/N ratio of resources (bulk soil) and microorganisms indicates greater microbial N limitation at the late stage [125].

The following summarized results are organized by major geographic region: In studies from America Northwest, it was found that not only DIN content, but also its composition and microbial N turnover rates depended on RTS stabilization. Young RTS soils (about five years after disturbance) showed equal amounts of ammonium and nitrate, but in older stages stabilized by revegetation (up to 150 years), only ammonium was found, in contrast to undisturbed sites where no DIN was detectable [292]. In conjunction with the highest DIN content, young RTS soils showed the highest gross N ammonification rates (280 ± 150 µg N dw⁻¹ d⁻¹ in soil incubations, −10–2 °C, six days [292]). Since the addition of vegetation leachate only increased the phosphorous content of microbial biomass but not C and N mineralization, the organisms seem to be limited to phosphorous and not to C and N [292]. In contrast, in the Canadian Arctic, Tanski et al. [213] found a decrease in ammonium and an increase in nitrate content with the progress of stabilization and revegetation. High nitrate concentrations were found primarily in fast-flowing mudflows and cracked surfaces, which are likely to have more oxygen. Larger nitrate pools in older, revegetated retrogressive thaw slumps than in younger thermokarst features could be additionally caused by colonization by N-fixing plants such as green alder and soapberry (Astragalus, Oxytropis in ref. [294], Sheperdia canadensis (L.) Nutt. in ref. [291]). However, other studies reported despite higher DIN contents in disturbed soils from thermo-erosion gullies, there were no differences in net N mineralization and N nitrification rates with median values around zero [29]. However, higher potential nitrification rates and suppressed denitrification were observed in the same thermo-erosion gullies than in undisturbed soils. In contrast to in situ measurements, model simulations of recovery from thermal erosion disturbances in the Alaskan tundra have assumed that nitrification is negligible since inorganic N availability is limited and ammonium released in the non-vegetated stage is immediately immobilized with only low inorganic losses instead of high DON losses [67]. At a thermo-erosion gully in Greenland, low DIN, negligible net nitrification, and negative net N mineralization were observed with slightly higher nitrification at disturbed sites [69].

At thermo-erosion gullies of the Tibet Plateau, DIN availability also depends on the age of the disturbed landform. The highest nitrate concentrations were found at the early vegetated stage (three years after collapse) with 2.7 ± 0.6 µg NO₃⁻-N g dw⁻¹ and at the exposed bare site with 4.5 ± 0.4 µg NO₃^--N g dw⁻¹ in contrast to the low nitrate content at the older disturbed site and the undisturbed control site with intact vegetation cover [218].

In addition, RTS formed in the ice-rich Yedoma permafrost riverbanks in northeastern Siberia were found to have the highest ammonium content in recently thawed bare sediments (35.3 ± 8.2 µg NH₄⁺-N g dw⁻¹), but highest nitrate content after stabilization and revegetation (0.7 ± 0.3 µg NO₃⁻-N g dw⁻¹, RTS Lena River [87]; Table S1). When differentiating between subsides, soils of thaw mounds in particular had increased nitrate content with a higher DIN/DON ratio (means of all sites 27.7 ± 36.3 µg NO₃⁻-N g dw⁻¹, 0.4 ± 0.6 µg NH₄⁺-N g dw⁻¹, 4.9 ± 3.0 data of one site see at Table S1)). In contrast, mixed Holocene and Pleistocene sediments at the slump floors were mainly enriched in ammonium and had a lower DIN/DON ratio (means of all sites 8.3 ± 15.3 µg NO₃⁻-N g dw⁻¹, 9.4 ± 7.7 µg NH₄⁺-N g dw⁻¹, 0.8 ± 0.8, data of one site see at Table S1 [256]).

3.2.3. Gaseous N₂O Loss and Based Microbial N Processes

Hillslope thermokarst landscapes have only recently been identified as a significant source of gaseous N losses in form of the GHG N₂O. Abbott et al. [264] provided initial evidence and measured elevated N₂O concentration in thermokarst features in upland soils in northern Alaska. In addition, N₂O emissions were detected in thermokarst-affected soils of the Tibet Plateau. Exposed non-vegetated sites of thermo-erosion gullies with low soil moisture (~40%) emitted significantly higher amounts of N₂O (72.60 ± 2.42 mg N₂O-N m⁻² d⁻¹ in August) and CH₄ (41.47 ± 12.10 mg CH₄ m⁻² d⁻¹ in August) than in the vegetated parts of the thermo-erosion gully, but less than in the undisturbed, vegetated adjacent soils [277]. At a comparable site, but with higher soil moisture (~67%), differentiation in a thaw sequence showed the highest N₂O emissions three years after initial disturbance ((1.97 ± 0.32 mg N₂O-N m⁻² d⁻¹, 100 fold increase and 0.12 ± 0.01 mg N₂O m⁻² d⁻¹, seven-fold increase from non-vegetated and vegetated sites in contrast to undisturbed sites, respectively [218]). However, in the middle (12 years) and late (20 years) stages, N₂O emissions decreased and were comparable to undisturbed sites (0.017 ± 0.01 mg N₂O-N m⁻² d⁻¹, [218]). N₂O fluxes correlated positively with nitrate concentration and negatively with water content [266]. Significant correlations were also found with microbial parameters: N₂O flux correlated positively with the copy number of denitrification genes (nirS but not nir K, nitrite reductase) and negatively with the abundance of the gene encoding N₂O reduction (nosZ, N₂O reductase). Overall, about 80% of the variability in N₂O fluxes was explained by substrate availability and denitrifier gene abundance [218]. High N₂O fluxes at the young stage (instead of N₂) and low nosZ abundance (better expressed by the ratio of nirS + nirK/nosZ) were explained by lower pH (around five) at these sites, which suppresses N₂O reductase similar to bare peat soils. CH₄ was also emitted at the same site. However, in contrast to N₂O emission, the highest CH₄ fluxes (1.3 ± 0.5 mg CH₄ m⁻² d⁻¹) were found at the late stage (20 years after collapse), associated with the highest amount of methanogen gene (mcrA, [266]). The global warming potential of N₂O emissions after thermokarst formation was about 60% of CH₄ emissions on a decal scale.

In addition, two RTSs on two major rivers of northeastern Siberia, namely Lena and Kolyma, in the ice-rich Yedoma permafrost region showed significant but highly variable N₂O emissions. However, in contrast to the findings on the Tibet Plateau, the highest N₂O emission rates (median 0.55 (0.13–6.29), mean 1.64 ± 2.61 mg N₂O-N m⁻² d⁻¹) were found after drying (WFPS = 42–84%, mean of 60 ± 8), stabilization, and revegetation of slope sediments and emissions were significantly higher than at undisturbed soils and freshly thawed bare slumps [87] (Table 1 and Table S2, Figure 5A,B). Based on measured N₂O emissions, N₂O loss under optimal conditions may be 54.8 mg N₂O-N m⁻² per year, which is 0.14% of the initial inorganic N content of exposed Yedoma [87]. Marushchak et al. [87] reported from bare, disturbed slump floor sites with high water content (WFPS ≥ 67%), but also undisturbed, dry sites (WFPS = 6–15%, mean of 13 ± 6) an N₂O uptake [87] (Table 1 and Table S2, Figure 5B) being in contrast to a study at the same site where even non-vegetated bare slump with intermediate water content (WFPS of 63%) had significant in situ N₂O emissions (2.03 ± 1.39 mg N₂O-N m⁻² d⁻¹) [256]. High N₂O emissions on slump floor sites were detected only with measurable nitrate content (RTS Lena River [87]; Table S1). However, not only nitrate, but also water content might play an important role. On dry thaw mounds at another site, no N₂O emission was detected at high nitrate content (39.4 ± 44.1 µg NO₃⁻-N g dw⁻¹), probable due to the lack of anaerobic microsites for denitrification (bare and moss-covered, WFPS~50%, RTS Kolyma River).

The net mineralization and net nitrification rates of slump floor soils measured in laboratory incubations increased with the thaw stage of RTS in Siberia. The lowest rates were observed on undisturbed, vegetated control sites (C/N 38), followed by young, non-vegetated disturbed sites (C/N ~15, both mainly with negative rates), and the highest rates were observed on revegetated, stabilized sites (gross N mineralization 15.1 ± 10.1 µg N g dw⁻¹ d⁻¹, net N mineralization 4.0 ± 4.9 µg N g dw⁻¹ d⁻¹, net nitrification 1.6 ± 1.5 µg N g dw⁻¹ d⁻¹, in aerobic incubations 4–5 °C, Table 1 and Table S1 [87,256]). Comparing of slump floor and thaw mounds soils, both sites showed similar net N mineralization in aerobic slurry incubations (Table 1 and Table S1, [256]). Strong positive correlations were observed between in situ N₂O fluxes and net mineralization and net nitrification rates, whereas no correlation was found between in situ N₂O fluxes and gross N mineralization rates [87].

Laboratory incubations confirmed the spatial in situ N₂O emission pattern with higher N₂O production under anoxic conditions, suggesting denitrification as the main source [87]. The highest anoxic N₂O production rates were found on revegetated slump floor with rates up to 2.8 ± 1.8 µg N₂O-N g dw⁻¹ d⁻¹ (WFPS 52 ± 9%) in contrast to undisturbed sites (0.01 ± 0.00 µg N₂O-N g dw⁻¹ d⁻¹, WFPS 13 ± 6%, both with acetylene at 5 °C for six days [87], Table 1, Tables S1 and S2, Figure 5E,F). In contrast, in another study at the same RTS, N₂O production was detected only on one bare slump floor site (0.02 ± 0.05 µg N₂O-N g dw⁻¹ d⁻¹, WFPS 52–67%) and was high at all thaw mound sites (mean 0.07 ± 1.0 µg N₂O-N g dw⁻¹ d⁻¹, WFPS 45–74%, in anaerobic laboratory incubations at 4 °C, 40 days [256]). In the same incubations, higher anaerobic CH₄-production was observed in the slump floor than in the thaw mounds with only negligible CH₄ production. After the addition of nitrate (110 µg NO₃⁻-N g dw⁻¹), which removed the probable N limitation, the two sites responded differently. Thaw mounds with old Yedoma material showed a decrease in anaerobic N₂O production (probable due to N₂ formation) [256], but a sharp increase in N₂O production was observed in mixed slump floor samples ([87], with the highest rates of about 3.3 ± 0.4 µg N₂O-N g dw⁻¹ d⁻¹ in ref. [256]). In all incubations, CH₄ production was only detectable when N₂O production was absent or very low, and after nitrate was added, CH₄ production disappeared in slump floor samples, both suggesting that increased DIN availability could affect methanogenesis at sites affected by thermokarst.

N turnover and increases of N₂O fluxes at different thaw stages of the RTS were accompanied by increases in the relative abundance of functional genes for nitrification, denitrification and dissimilatory nitrate reduction [87]. Ammonia oxidation functional genes, measured as total amoA, increased from 0.5 ± 1.0% and 1% in undisturbed soils (and freshly disturbed soils) to approximately 3.4 ± 1.0% and 7% in revegetated slump floor, in terms of captured functional N genes and total 16S rRNA, respectively (Table 1, Tables S1 and S2, Figure 5G,H). In contrast to an increase of bacterial amoA, there was no increase in archaeal amoA (Table 1 and Table S1). In addition to the nitrification genes, the functional genes of denitrification, nirS + nirK of nitrite reductases, increased from 15% to 29% (in % of functional N genes) with the increase of the ratio (nirS + nirK)/nosZ (Table 1, Tables S1 and S2, Figure 5G,H). However, there was no increase in norB (NO reductase). In addition, the functional gene of dissimilatory nitrate reduction DNRA, nrfA increased from ~1.8 to 3.2% (in % of captured functional N genes).

3.2.4. Lateral N Loss

High DIN availability in hillslope thermokarst landscapes may lead to lateral N losses to aquatic systems. Recent studies show that Arctic thermokarst features are hotspots for dissolved N losses and increase nutrient loading to wetlands, rivers, and lakes, which could stimulate primary production [29,88,208,211,212,264,281]. Thermokarst abruptly releases nutrients, temporarily increasing aquatic nutrient availability while decreasing terrestrial stocks [238]. Increased and sustained inorganic N leaching of ammonium, and in some cases, of nitrate was observed in thaw slumps, thermo-erosion gullies, and active layer detachments in thaw streams, rivers, and oceans compared to undisturbed areas. In addition, DON and other nutrients such as sulfate, phosphate, potassium, calcium, and magnesium were also leached [28,208,210,211,212,213]. It has been shown that nutrient leaching can continue for several decades after disturbance, but this varies on a regional scale and depends on the type of feature, age of the landscape, and season [29,211,212,264]. Nitrate output decreased in older, stabilized thermokarst features [264] and reached its maximum at the end of the season [212]. Isotopic evidence indicates that nitrate is effectively retained or removed by organic soils in undisturbed watersheds, in contrast to disturbed watersheds with large areas of mineral soils. Unlike organic soils, mineral soils facilitated the export of nitrogen from nitrification processes, while less N was immobilized by vegetation or removed by denitrification [211].

Total N release from thawing permafrost at an ice-rich Yedoma permafrost riverbank RTS in the central Lena River Delta in northeastern Siberia (with a length of 1.7 km) was estimated 1.7 kg N m⁻² y⁻¹ and associated with an inorganic N release as ammonium of 39 g N m⁻² y⁻¹ (between 2012 and 2019, [87]). This is a remarkably high amount of added N compared to the major external atmospheric N inputs by BNF (20–2000 mg m⁻² y⁻¹ [53,59,60,61] and N deposition (<200–300 mg m⁻² y⁻¹ [62,63]). This highlights the importance of thawing permafrost as a source of nutrients. At another 1.7 km long Yedoma cliff in the Lena Delta, 400 t N were eroded annually between 2015 and 2018 [281]. For an RTS along a riverbank in Yukon, Canada, with an area increase from 6000–10,000 m², a total N loss of about 19.3 t per year was found over a three-year period [298].

In summary, these results show that disturbance on hillslope thermokarst landscapes can be substantial sources of N₂O and DIN. Thawing of collapsed soils increases inorganic N content, N turnover rates, abundance of functional N genes, and N₂O production only when optimal moisture conditions allow simultaneous aerobic processes for nitrate production and anaerobic processes for N₂O production. Sub-sites of RTS, such as slump floor and thaw mounds, appear to have different levels of N availability. In particular, ammonium content at slump floors indicates the second level of N availability (Figure 2II), while high nitrate availability associated with high N₂O production indicates the third level, the N hotspot (Figure 2III). In situ nitrate limitation of slump floor, unlike thaw mounds with typically high nitrate content, can be eliminated by leaching, which can increase total N₂O loss [256].

3.3. Bare Soils in the Transition between Terrestrial and Aquatic Ecosystems

In the transition zone between terrestrial and aquatic ecosystems, young, initial soils without vegetation in floodplains have been formed by continuous fluvial deposition and erosion by river or the sea and are therefore often disturbed [11]. Preliminary evidence of N forms and microbial N-based processes suggest that these soils are likely N hotspots with high N availability. Alluvial soils on lakeshores and riverbanks of Arctic river deltas and coasts have deep active layers and relatively high soils temperatures, with large variations in soil properties like hydromorphology [299,300,301]. Soils in, e.g., the Lena River delta in northeastern Siberia, are stratified with layers of silt loam and sand containing low but more degradable allochtonic SOM [38,39,302,303,304].

Figure 5. Biochemical traits of (A,C,E,G) N hotspots with high N availability and (B,D,F,H) adjacent control sites with N limitation using one described site as an example. (A,B) water-filled pore space (WFPS), C/N ratio (C,D) dissolved inorganic nitrogen (DIN, ammonium + nitrate), ratio of DIN to total N (DIN/TN) and nitrate. (E,F) microbial net N turnover: N mineralization, nitrification, denitrification, and (G,H) abundance of genes of key functional enzymes of nitrification (amo, ammonia monooxygenase in % of 16S rRNA and **log10 gdw⁻¹), denitrification (nirS + nir K, nitrite reductases in % of 16S rRNA, *% of N genes and **log10 gdw⁻¹) and (nirS + nirK)/nosz ratio (nitrite reductase/N₂O reductase *** × 10³) in subarctic bare (BP) and vegetated peatland (VP, references see Table 1), Arctic retrogressive thaw slump (RTS) and undisturbed site (URTS, references see Table S1), thermoerosion-gully (TEG) and undisturbed site (UTEG) on the Tibet Plateau [218], Arctic bare alluvial soils (AS) and vegetated floodplain (VP) [119,302,304] and animal-influenced permafrost affected soils of the Arctic (AIA) [234] and Antarctic (AIAA) [188,189,236] and non-influenced soils of Arctic (NIA) and Antarctic (NIAA). For more detail, see Table S2.

Figure 5. Biochemical traits of (A,C,E,G) N hotspots with high N availability and (B,D,F,H) adjacent control sites with N limitation using one described site as an example. (A,B) water-filled pore space (WFPS), C/N ratio (C,D) dissolved inorganic nitrogen (DIN, ammonium + nitrate), ratio of DIN to total N (DIN/TN) and nitrate. (E,F) microbial net N turnover: N mineralization, nitrification, denitrification, and (G,H) abundance of genes of key functional enzymes of nitrification (amo, ammonia monooxygenase in % of 16S rRNA and **log10 gdw⁻¹), denitrification (nirS + nir K, nitrite reductases in % of 16S rRNA, *% of N genes and **log10 gdw⁻¹) and (nirS + nirK)/nosz ratio (nitrite reductase/N₂O reductase *** × 10³) in subarctic bare (BP) and vegetated peatland (VP, references see Table 1), Arctic retrogressive thaw slump (RTS) and undisturbed site (URTS, references see Table S1), thermoerosion-gully (TEG) and undisturbed site (UTEG) on the Tibet Plateau [218], Arctic bare alluvial soils (AS) and vegetated floodplain (VP) [119,302,304] and animal-influenced permafrost affected soils of the Arctic (AIA) [234] and Antarctic (AIAA) [188,189,236] and non-influenced soils of Arctic (NIA) and Antarctic (NIAA). For more detail, see Table S2.

The nitrogen content of these soils varies between 0.3 and up to 1.4 kg m⁻² [303]. Differences in frequency and duration of flood events can affect nitrogen cycling by creating temporal mosaics of oxic and anoxic conditions that are susceptible to nitrification and denitrification processes [305]. At a beach site on a Holocene terrace in the Lena Delta on Samoylov Island (Figure 3D), the alluvial soils are predominantly pH neutral with narrow C/N ratios (around 15). High inorganic nitrogen concentrations (>2 µg N g dw⁻¹) were found in high amounts only in non-vegetated soils in sandy loam layers with more than 2% SOM. DIN quality changed during the summer season from mid-July to late August, with ammonium decreasing and nitrate content increasing to high levels of 89 ± 57 µg N g dw⁻¹ (about 4.6% of TN) [302]. Therefore, nitrate content is comparable to known N hotspots [86,87,120,218] (Figure 5C, Table S2). In bare soils, DIN content correlated with a high potential net N mineralization (1.7 µg N g dw⁻¹ d⁻¹, 20–25 °C, [304]) and substrate-induced net nitrification rates (13.0 ± 0.3 µg N g dw⁻¹ d⁻¹, in situ temperature with mean of 5.2 °C, [119]). Both N turnover rates are in the range of the previous described N hotspots as RTS (for nitrification even five-fold higher [87], Figure 5E, Table S2). In contrast, only low microbial N turnover rates were observed at sparsely vegetated or non-vegetated beach sites (Figure 5F). Nitrification rates were performed by ammonia-oxidizing bacteria (AOB) rather than by archaea (AOA) and correlated with the abundance of the functional gene amoA of AOB but not AOA. The gene amoA accounted for 4.8% of 16S rRNA at bare sites, in contrast to 0.7% of vegetated sites [119], which is almost equal to the proportion detected at N hotspots of RTS [87] (Figure 5G, Table S2). All these biogeochemical traits indicate that these soils may have potential for N losses, but there was only preliminary evidence that these soils leach inorganic nitrogen into the Lena River or release N₂O (aerobic N₂O production in incubations [304], Figure 4D). Temperature elevation experiments on enriched Nitrosospira-like AOB showed that heating increased ammonia oxidation rates with Q₁₀ values of about 4 (12–22 °C, with highest nitrifying activity at 22–30 °C). In contrast to these alluvial soils with SOM content lower than 2%, the bare riverbank sands of a neighboring island, which receive meltwater from the thawing Yedoma exposure, had only negative net microbial N turnover and negligible N₂O fluxes, but significantly higher than the adjacent vegetated Holocene soils [87].

In addition to alluvial soils on beaches, there was further evidence of high N availability in bare soils sampled on vertical cliff faces. Soils containing high DIN content were measured, along with high potential net mineralization, nitrification rates, as well as aerobic N₂O production [119,302,304]. On bare lakeshore soils in an Antarctic dry valley, high ammonium levels (300 µg N g dw⁻¹) but only low nitrate levels were found in combination with high variable in situ N₂O emission rates (highest rates: 1.49 mg N₂O-N m⁻² d⁻¹, [306]) associated with high nitrification rates [307], indicating another likely N hotspot. High N₂O emissions reduced CH₄ emissions at this site [306].

3.4. Animal-Influenced Soils

Studies over the last decade have shown that animals, particularly seabirds in the Arctic and penguins and marine animals in the Antarctic, increase the input of external nitrogen in permafrost-affected soils through excretions. This external N input enhances the described traits of high N availability and is associated with N₂O emissions (Figure 4E). A comparison of two cliffs inhabited by seabirds in the European High Arctic revealed that the cliff with higher bird density had higher in situ N₂O fluxes (mean 0.40 ± 0.24 mg N₂O-N m⁻² d⁻¹), which correlated with high potential denitrification rates (96 µg N g dw⁻¹ d⁻¹, 10 °C with acetylene) and high abundance of functional denitrification genes (nirS, nirK and nosZ) (Figure 5A,E,G, Table S2, at BL T-M site of [234]). In addition, these animal-influenced soil sites had higher nitrate (2270 µg N g dw⁻¹) and ammonium contents (6 µg N g dw⁻¹), lower C/N (~10), and higher δ¹⁵N in bulk soil (max. 19.4‰) compared to the other bird cliffs without N₂O emissions and compared to the non-influenced tundra soils (Figure 5A–D). However, as soils far from bird colonies had low DIN and low δ¹⁵N contents in bulk soil and also did not emit N₂O, but had high gene abundance of denitrifiers and denitrifying activity, soil N forms might be a better indicator of high N₂O emissions than microbial traits for animal-influenced soils.

In addition, in maritime Antarctica, highly animal-influenced soils had comparably high N₂O emissions (mean of penguin, seal and skua colony sites 0.22–0.88 mg N₂O-N m⁻² d⁻¹ with mean of 0.38 ± 1.15 mg N₂O-N m⁻² d⁻¹ of active penguin colony [236]), in contrast to the animal-lacking tundra (Figure 5A,B, Table S2, [235,236,237]). In addition, these animal-influenced soils had higher ammonium and nitrate content (e.g., with mean values of 55.4 ± 24.2 µg N g dw⁻¹ and 90.2 ± 49.1 µg N g dw⁻¹, respectively in penguin colony [236]), higher abundance of functional genes, and higher activities of nitrifiers and denitrifiers than uninfluenced soils [188,189] (Figure 5C,D,G,H, Table S2). Nitrogen input by animals increased the abundance of bacterial ammonia oxidizers (AOB amoA) and potential ammonia-oxidizing activity (e.g., mean value of 1.3 ± 0.3 µg N g dw⁻¹ d⁻¹ penguin colony, 24 h, 15 °C, Figure 5E,F), which was correlated with AOB amoA abundance, but decreased the abundance of archaeal functional genes (AOA amoA) [188]. The higher abundance of functional denitrification genes (nirS + nirK) and activities (e.g., mean value of 3.7 ± 2.8 µg N g dw⁻¹ d⁻¹ in penguin colony, 8 h, 10 °C) in ornithogenic soils indicate that nirS-encoding denitrifiers dominate in contrast to adjacent tundra soils [189] (Figure 5E–H, Table S2).

3.5. Wildfire-Affected Soils

Wildfires have rarely affected tundra permafrost soils of late, but they have increased in recent years and will increase rapidly with climate change [11,308,309,310]. Fires in boreal ecosystems and their effects on N availability, particularly short-term increases in DIN, have been studied more frequently, but data on N turnover, the abundance of N-cycling microorganisms, and N₂O emissions in post-fire soils are scare [140,311,312]. In tundra ecosystems, wildfire effects on N availability are less studied, mainly in a single case, the largest fire in the Alaskan Arctic (2007) [238,313,314]. Fire increases the depth of active layer, even years after the fire [238,308], can cause thermokarst [315], remove vegetation, and alter nutrient status through combustion and ash residuals [313,316]. Recovery of the tundra ecosystem after wildfire is strongly influenced by nutrient losses from post-fire leaching and volatilization. In situ measurements [316], simulation of long-term recovery [317] and incubations experiments revealed high DIN content in soils with up to 14-fold increase in nitrate and four-fold increase in ammonium content [318] in the first few years after the fire. In addition, in the adjacent watersheds DOC, TDN and TDP were three times higher and ammonium was four-times higher, while nitrate content was lower even four years after the fire event [319]. Ten years after fire disturbance, watersheds did not have significantly elevated inorganic N anymore, but did have elevated dissolved organic nutrients (increase of 25% and 59% for organic C and N, respectively), and two to ten times higher lateral N flux than BNF and N deposition [238]. In situ measurements and long-term simulations showed that C and N contents (DIN, DON, SOM) decreased in soils disturbed by wildfire due to continuous loss [317]. For example, total ecosystem C stocks decreased by 5.7–31% and N stocks decreased by 34–18% within a decade in moderately and severely burned areas compared to unburned sites [238]. In contrast, low-intensity experimental fire did not affect soil C, N or P pools, but ¹⁵N labeling experiments showed an increase in N retention in the residual soil N pool due to increased microbial N assimilation in the dry heath tundra of western Greenland [320]. In this experiment, a three-fold increase of post-fire soil N₂O emissions (5.54 ± 1.09 ng N₂O-N m⁻² d⁻¹) under wet conditions with WFPS of 70% and a 51-fold increase of potential denitrification activity (0.84 µg N g dw⁻¹ d⁻¹, 10 °C with acetylene) were observed [318]. Fire makes nitrogen available by bypassing the N mineralization step rather than through the loss of vegetation [318] or through supply from external sources (BNF, N deposition, [317,321]). In long-term simulations, N mineralization was about 350% higher 200 years after fire than on control sites (46.6 mg N m⁻² d⁻¹). Higher DIN levels in soils and watersheds and simulated higher N mineralization are preliminary indications that burned permafrost soils are likely short-lived hotspots of N availability where microbial nitrification and denitrification rates and N₂O emissions are likely elevated.

4. Hotspots of N Availability and Climate Change

4.1. Contribution of N Hotspots to Climate Change

As shown in the last chapters, hotspots with higher N availability are characterized by a more open N cycle and N imbalance caused by lower N immobilization compared to N mineralization, resulting in higher N susceptibility to vertical and lateral N transport. Therefore, hotspots of N availability have an important influence on climate change (1) directly through the formation of the potent GHG nitrous oxide (N₂O) and (2) indirectly through dissolved N losses and N transport, which affect the C balance, C source functions (SOM decomposition and GHG emission of CO₂ and CH₄), and C sink functions (C uptake and storage) on-site or off-site of N hotspots.

As recently reviewed by Voigt et al. [82], vegetated permafrost-affected soils are generally small but distinct sources of N₂O during growing season with a mean 150 ± 420 µg N₂O-N m⁻² d⁻¹. Bare soils emitted significantly higher amounts with a mean of 1025 ± 1727 µg N₂O-N m⁻² d⁻¹ (and all sites together with a mean of 288 ± 858 µg N₂O-N m⁻² d⁻¹). Therefore, N₂O fluxes from N hotspots summarized in this review, 1980 ± 3190 µg N₂O-N m⁻² d⁻¹ from bare surfaces on permafrost peatlands, 1640 ± 2610 µg N₂O-N m⁻² d⁻¹ from disturbed but revegetated RTS (Table 1), 1970 ± 320 µg N₂O-N m⁻² d⁻¹ from thermo-erosion gully on the Tibet Plateau [218] are remarkable N sources. Voigt et al. [82] calculated that all permafrost regions could be a source of 0.14–1.27 Tg N₂O-N y⁻¹ (each calculated with mean), corresponding to 0.8–7.1% of the current global N₂O-N budget (but not including currently found N₂O emission of RTS). Recent estimates of N₂O emissions at the pan-Artic scale using a terrestrial ecosystem model are twice as low [322], but in perspective of this review this seem to be an underestimate, as the representation of N cycling processes in high latitude soils in process models is still not fully developed. All in all, great uncertainty remains regarding the magnitude of N₂O emissions from natural ecosystems [323]. Since the described hotspots of N availability are significant sources of N₂O, they can contribute significantly to climate change [83]. Therefore, due to increased N₂O emissions, N hotspots can have high non-C feedback on climate warming. The first back-of-the-envelope extrapolation of data from bare peat circles at a Russian study site to the entire Arctic suggests that emissions from these hotspots could be ~0.1 Tg N₂O-N y⁻¹, representing 4% of the global warming potential of Arctic CH₄ emissions [214] and about 1% of natural N₂O sources [324]. In mesocosm laboratory experiments involving these bare peat circles, N₂O emission was 14% of CO₂ emission under dry conditions [223]. More recently, bare peat soil circles have been described as likely sources of the atmospherically reactive gases nitric oxide (NO) and nitrous acid (HONO), which may therefore also play a critical role in N hotspots affecting atmospheric chemistry and climate [255]. For the hillslope thermokarst landscape types in the northern permafrost region, the upper end of pan-Arctic N₂O emissions can be estimated using in situ N₂O emissions (with a median of 54.8 mg N₂O-N m⁻² per year [87]) and area estimates of all three thermokarst landscape types at 0.91 × 10⁶ km² [270]. This would be a maximum of N₂O emission of ~0.05 Tg N₂O-N y⁻¹. This is about half the calculated N₂O emissions from the N hotspot in bare peatland in the discontinuous permafrost region, but it does not include N₂O emissions from hillslope thermokarst at the Tibetan plateau. The additional N hotspots proposed here, i.e., alluvial, animal influenced, and wildfire-affected soils, could have further implications for the global N₂O budget due to their greater spatial extent.

In addition to gaseous loss, hotspots of N availability contain high amounts of inorganic N. In particular, alluvial soils likely leach dissolved N into adjacent N-poor terrestrial or aquatic ecosystems, as was observed on the hillslope thermokarst features. The relative location, timing, and magnitude of N release will likely determine whether these N fluxes are used by plants or organisms in the soil or are susceptible to transport [55]. Since N is a highly demanded nutrient by primary producers and SOM decomposers, it regulates ecosystem C cycling, determines whether permafrost soil functions as a net CO₂ sink or source, and regulates microbial mediated CH₄ production and emissions. However, how and to what extent nutrient availability affects the C balance in permafrost-affected ecosystems remains unclear today [47,50,52,55,89,90,91,92,93,94,95,96]. In general, the short-term change of SOM decomposition by addition of (in)organic forms is referred as priming effect [325]. However, reactions on N addition to permafrost-affected soils or stream water were contradictive and no, positive (increase of SOM decomposition) or negative priming (decrease of SOM decomposition) effect depending on extend of N limitation and SOM degradability was reported [52,71,125,326,327,328,329]. A positive priming effect was observed in C limited mineral and cryoturbated subsoils as microbes were investigated for their additional N in the production of extracellular enzymes to degrade SOM (N mining) [52,326,329]. In another study, the microbes invested additional N in their own growth instead of enhancing the degradation of SOM (negative priming) because their N requirements were met by N addition obviating the need to produce enzymes that degrade carbon-nitrogen compounds [330,331]. A negative priming effect was also found in disturbed permafrost soils with high N availability on the Tibetan Plateau [125]. Whether SOM degradation is enhanced by N is not clarified yet and depends on other nutrients such as phosphorous, as well as many environmental factors, such as soil minerals, acidity, redox state, water availability, vegetation cover, and changes in soil microbiome composition and functions [327,332,333,334]. Recently, model simulations including the mineral N pool locked within the permafrost have shown that uptake of additional nitrogen from thawing permafrost reduces N limitation of vegetation, resulting in a negative feedback on climate by slowing the rate of increase of atmospheric CO₂ [335]. In contrast, the effect of N availability on net CH₄ flux in permafrost-affected ecosystems is poorly studied [204,336,337,338]. Increased N availability could remove the N limitation of both methanogenic and methanotrophic microorganisms thus promoting both processes [339,340,341]. In contrast, CH₄ production is inhibited by higher DIN content due to competition with denitrifiers and denitrification intermediates are toxic for methanogens and methanotrophs [11,339,342,343,344,345,346,347]. Nitrogen fertilization experiments stimulated N₂O emission from Arctic soils [163,203,204] or had no effect [227,336] and increased NO [348].

It is not clear whether higher C release or uptake is induced by higher N availability at the hotspots described. Despite higher N availability, bare peat soils of permafrost peatlands usually had low net CO₂ emissions than adjacent vegetated sites on peat plateau-which act as CO₂ sinks [250], although soil microbial heterotrophic respiration, estimated using, e.g., an in situ ¹⁴C partitioning approach, was similar or lower in bare peat plateau than in vegetated surfaces [223,242,244]. However, the well-drained bare peat plateaus act as CH₄ sinks or small sources in contrast to the surrounding waterlogged fens [223,252], and even in deeper water-saturated soil layers CH₄ production is suppressed by nitrate content, which promotes N₂O production [247]. Overall, bare peat surfaces apparently function as net C sources due to higher CO₂ losses and little or no CH₄ emissions. No clear effects on ecosystem respiration and CH₄ fluxes were found for hillslope thermokarst N hotspots, with CO₂ and CH₄ release lower, similar, or higher than undisturbed sites depending on thermokarst stage, vegetation, and moisture conditions [88,257,277,286,287,297], but CH₄ production was detected only in the absence of N₂O [88,256]. Atmospheric CO₂ uptake ceases when the vegetation cover is disturbed, but on the other hand, once stabilized, highly productive pioneer plants that feed on the nutrients released when permafrost thaws rapidly colonize these disturbed sites. Higher N availability in a Tibetan thermo-erosion gully had a negative priming effect caused by increased microbial metabolic efficiency, thus reducing carbon loss as CO₂ [125].

4.2. Impact of Climate Change on N Hotspots

Permafrost regions are affected by drastic changes due to anthropogenic-induced climate change [11,97,98] with the average air temperature in the Arctic increasing by twice the global average over the past 100 years—Arctic amplification—which has serious ecosystem implications [349,350]. In addition, the fastest changes in temperature and precipitation are predicted to occur along the coasts of Arctic Ocean [270,351], affecting soils in the transition between terrestrial and aquatic ecosystems, particularly N hotspots on thaw slumps and alluvial soils. Furthermore, temperature increase favors thermokarst processes [99,100], leads to higher soil temperature, gradually deeper summer thawing of the active layer, loss of permafrost [5,352,353], and affects vegetation by lengthening the growing season [354] and shifting plant cover [355]. In addition, warming enhances soil microbial processes and causes changes in microbial communities [139,334,356,357,358,359]. Therefore, changes could directly increase the activity and abundance of microbial driven N processes and organisms, primarily by increase of temperature [119,155,196]. These studies showed that experimental warming increased nitrifying activities, and that most enriched nitrifiers were cold-adapted rather than psychrophilic. Thawing of permafrost releases long-stored SOM, and climate changes can accelerate decomposition processes that serve as source of C-containing GHG CO₂ and CH₄ [93,265,352,354,360,361,362,363,364], as well as N₂O [82], further contributing to global warming through positive feedback processes [365].

There is evidence that climate change and associated permafrost degradation in tundra ecosystems increase N availability [15,45,108], partly through increased N mineralization due to temperature rise and lengthening of the growing season [75,78,102,104,366,367] and partly through the release of inorganic N from thawing permafrost [28,55,75,103,213]. The described hotspots of N availability may therefore illustrate how climate change will affect microbial N cycling in permafrost-affected ecosystems. Increased N availability combined with extremely biodegradable carbon inputs from permafrost degradation could increase N₂O emission by providing substrates for microbial processes in combination with reduced plant demand [219,322]. In frozen permafrost samples, N cycle genes were generally present and increased after thawing, but there may be higher potential for N₂O emissions due to the lack of a functional gene for converting N₂O to N₂ (nosZ gene) [139]. Recently, a meta-analysis has shown that experimental warming alters biogeochemistry in cold ecosystems, increasing N mineralization rates and N₂O emissions, but does not affect N fixation and thus further N input [102]. The changes described are more closely linked to enzyme activities than to changes in the abundance of functional nitrification and denitrification genes or the total abundance of archaea or bacteria. Therefore, the activities of N-cycling organisms, which are used as traits for N availability, might be more affected by climate change than their gene abundances. Recently, it has been shown that climate induced N loss, combined with enhanced CO₂ enrichment and an increased N demand by plants, is leading to a global decline of N availability of non-agricultural soils [105,368]. Initial evidence for a decline of N availability in permafrost-affected ecosystems was provided by a decline in δ¹⁵N in Arctic lake sediments [369], in wood rings of shrubs due to warming [370], and in leaves on a transect in the Tibetan Plateau due to increased N loss [95].

Climate change could have differential effects on the areal extent of N hotspots by reducing the extent of palsa and peat plateau areas due to palsa collapse and increasing thermokarst at hillslope landscapes. Reconstruction of habitat changes during the Holocene of subarctic peatlands in Fennoscandia indicates that bare permafrost peatlands had the greatest spatial extent about 2000 years ago and then declined [243]. Satellite data have shown that bare areas in Fennoscandia have decreased by about 20% within a decade (2007–2015, area of about 1 kha [243]) and in northern Norway by 33–71% over the past 60 years (1950–2010, area of 11 kha [371]). In addition, bioclimatic models have shown that Fennoscandia is no longer predicted to have a suitable climate for palsa and peat plateau by 2040 [372]. As a consequence, N hotspots in permafrost peatlands may be lost in the near future and degraded wet peatlands may instead act as N₂O sinks [82,86,214]. However, large spatial variations are expected in permafrost peatland development, which may also lead to landscape desiccation [373] and thus displacement of other likely N hotspots. In contrast, hillslope thermokarst landscapes, particularly affected by RTS, have increased by 331% over the last two decades in northern Siberia [374] and by 617% over the past decade on the Tibetan Plateau [278] and will increase in area from 0.1–3% in permafrost region between 2000 and 2300 [99]. The expansion of the area will likely result in higher releases of N₂O and nutrients, both of which are likely to provide positive climate feedbacks. In addition to the known N hotspots, the area of the likely alluvial N hotspot could also change due to sea level rise, and wildfire-affected soils will increase due to the projected increases in wildfire in the Arctic tundra [238,309,375,376], meaning that the spatial extent of both potential N₂O hotspots will change.

Preliminary direct results from climate manipulation studies indicated that climate change might lead to increased N losses from bare peat soil circles. In laboratory experiments on thawing peat cores from bare and vegetated palsa surfaces, higher N₂O release was observed in a dry scenario, representing a gradual deepening of the active layer, than in a wet scenario, representing a complete collapse of the palsa [248]. When gradual thawing reached the upper permafrost layer, the N₂O emissions from bare peat increased fivefold compared to the fluxes during thawing of the active layer. The presence of vegetation did not prevent thaw-induced N₂O release under dry conditions, in contrast to waterlogged conditions [248]. Since ammonium began to accumulate in the mesocosms immediately after thawing at the interface between active-layer and permafrost only in the dry scenario, N mineralization from freshly thawed peat appeared to be a longer-term major source of N₂O [248]. The same authors showed the first in situ evidence for increasing N₂O emissions upon warming of bare peat sites as well as vegetated peat surfaces [247]. Therefore, climate change will not only alter the area of certain N hotspots, but may also lead to increased N₂O emissions from existing hotspots. However, these changes are highly dependent on the magnitude and direction of changes in hydrology and soil moisture content. In these experiments, in addition to N₂O, CO₂ emission increased after thawing regardless of moisture, and thawing also increased CH₄ uptake under dry conditions (but remain neutral at higher moisture) [223]. Altogether, climate change that increases drought, heat, and vegetation stress on bare peat soils in particular could lead to higher GHG emissions, as suggested by bioclimatic model stimulations for palsa and peat environments [377,378]. In general, permafrost peatlands will be affected by climate change, as recent climate models shown. Using a modeled global warming stabilization scenario, Hugelius et al. [33] estimate that the current carbon sink of northern peatlands will be transformed into a carbon source when 0.8–1.9 million km² of permafrost-affected peatland thaw, mainly through enhanced CH₄ emissions, slightly enhanced CO₂ emissions, and minor N₂O losses, but they also predict losses of peat to fluvial systems of 0.4–09 Pg N. In addition, the simulation of increased N availability due to probable climate change by adding nitrate to disturbed soils of bare peatland soils and RTS resulted in increased N₂O emission in incubation experiments of the described N hotspots [87,253,256]. Thus, if N availability from N hotspots increases, this could further increase N₂O emissions. Overall, it is not yet clear how climate change will directly affect the biogeochemical properties of thermokarst sites through temperature or changes in moisture, but it could have a significant impact on the N cycling, as exposed sites on Yedoma in particular have high levels of inorganic N and labile organic matter, which increases N availability [42].

5. Conclusions and Perspectives

In the last two decades, certain sites in permafrost-affected soils have been identified as important sources of the potent greenhouse gas N₂O (reviewed in [82]) and as sites of inorganic N leaching [28,210,213], both of which are associated with direct and indirect feedbacks on climate. Based on the synthesis information about these N hotspots and linking their geochemical properties to N losses, this review proposes that these hotspots of N losses are caused by high N availability. This confirms our hypothesis that tundra permafrost-affected soils are not always N-limited but show a great spatial diversity in terms of N availability. An open N cycle instead of closed N cycle is typical for hotspots of high N availability. We have shown that certain soil and microbial traits can be used to predict N availability status, especially at the highest level of N availability, the N hotspot level, and thus identify sites with high N loss. The characteristics of the N hotspots differed significantly from the adjacent non-N-releasing environments, which are associated with typical traits of the N-limitation level. The key traits of N hotspots with high N availability are as follows and should be used in combination:

• Soil properties: Low C/N below a threshold of 25 [29,86,87,120,202,234,247,255,256], high δ¹⁵N of bulk soil [234,304], WFPS around 60–70% [82,86,87,120,202,217,247,256] and high DIN [28,29,86,87,120,188,189,202,208,210,213,214,234,247,255,256,292,304], sometimes expressed as higher DIN/DON ratio [256]. Nitrogen loss often correlates with dominance of nitrate over ammonium [29,87,213,216,217,247,255,256], but at some sites a higher ammonium content was found in association with higher N₂O emission [86,188,189,248,306]. Therefore, the total inorganic N content would be a better indicator for N loss.
• Microbial process rates measured in situ or in laboratory incubations correlated with N loss and were therefore appropriate traits of high N availability. They were gross [202,216] and net mineralization, the sum of ammonification and nitrification [87,304], gross [111,120,202] and net nitrification [29,69,87,119,188,304,307] and denitrification [189,205,234,254,256]. Since gross N turnover rates are also indicative of the total amount of circulating N and theses rates were higher than currently expected in this ecosystem [59], we propose to use net rates as an indicator of N limitation status instead. Net rates could provide a better indication of how much N might be lost from the tight N cycling system of permafrost-affected soils. However, net rates estimated using laboratory incubation techniques are highly dependent on conditions such as temperature and have inherent biases related to functional groups, so they do not characterize the contribution of, e.g., nitrifying organisms in the natural habitat [379]. Recently, however, in situ N₂O flux rates from an N hotspot were shown to be highly correlated with anaerobic N₂O production in laboratory incubation tests, as well as with net mineralization and net nitrification [87], so laboratory tests could be a good tool for predicting field fluxes.
• The abundance of functional genes or the detection of known 16S rRNA gene sequences of nitrifiers [119,120,188], denitrifiers [189,218], or both [86] have been correlated with high N₂O losses in N hotspots. For nitrifiers, especially ammonia-oxidizing bacteria (AOB) and archaea (AOA), the gene amoA was mostly applied. However, gene AOA amoA abundance also dominated in permafrost-affected soils [118,120,147,182], but generally did not correlate with nitrification activity and N₂O production rates. The reasons are that AOA do not exclusively use ammonia as substrate and that AOA produce N₂O with about 30 times lower yields [129]. Therefore, it will likely be useful to distinguish between AOA and AOB amoA when using as traits for N₂O production in the future. To predict N₂O production by denitrifiers, high levels of their functional genes, mainly nirS + nirK genes of nitrite reductase, low levels of the gene nosZ (N₂O reductase), or a high ratio of (nirS + nirK)/nosZ, were found in N hotspots in correlation with high N₂O emission [205,218] and could therefore be a suitable microbial trait for N losses. In addition, metatranscriptomic analysis, which detects actively produced transcripts, could be another good tool to detect active microbial N cycling and N losses. This detection method has recently shown a correlation between denitrification activity and N₂O production in permafrost-affected soils [136].

Because microbial traits were more often correlated with high gaseous N loss than with N leaching, it is not clear whether high lateral N loss can also be predicted by microbial traits. However, in any case, when analyzing large volumes of soils samples or using existing data to predict gaseous and lateral N losses, it would be appropriate to use only geochemical traits that are easier to measure and thus available for more sites.

Importantly, N hotspots in permafrost-affected soils were described for only a few study sites, and it can only be assumed that the results are representative of geochemically similar sites throughout the tundra permafrost-affected ecosystem. Therefore, the status of N availability, as well as process rates related to N cycling and sink/source character throughout the tundra biome, remain uncertain. More data are needed through in situ measurements combined with geomicrobiochemical properties to improve spatial and temporal resolution, especially for the non-growth period [82,380]. In addition, different revegetation stages of N hotspots and thus levels of N availability could be used for a more detailed analysis of plant nutrient requirements relative to their growth rates and the role of plant functional types. Although the current areal extent of N hotspots and their evolution in context of climate changes are not yet clear, increases in areal extent will result in N losses that will significantly affect regional biochemistry and the global climate. In addition, there is evidence of increased mineral N cycling and N losses with warming and climate-change-induced disturbances, but more observations, e.g., obtained through manipulation experiments, are urgently needed with a view to how climate change directly affects N processes in Arctic soils. Elevated N concentrations in formerly N-limited permafrost soils could affect the carbon cycle, but studies of C-N interactions, particularly the effects of N cycling processes on methanogenesis, are lacking. Knowledge of the existence, spatial extent, and biogeochemical properties of these N availability hotspots is important for understanding current and future GHG release of all three gases. The proposed N availability traits could be used to develop ecosystem models for simulating emissions of all three GHGs.