Recent hummock establishment in the margin of a subarctic fen, Finnish Lapland

Northern fens, that host unique biota and form a remarkable carbon stock, are sensitive to changes in the moisture balance and, therefore, may be strongly affected by climatic fluctuations. However, long‐term monitoring and palaeoecological studies of fens are relatively rare and, as a result, their responses to past and current climatic fluctuations are poorly known. In this study, we examined the recent vegetation change as well as changes in testate amoeba communities in the mire margin of a subarctic fen in Finnish Lapland with four peat profiles. Testate amoebae were used as indicators of past fluctuations in water table depth. The vegetation showed a drastic shift from sedge‐dominated fen to Sphagnum‐dominated communities during the late 20th and the early 21st centuries. This shift was accompanied by a turnover in the testate amoeba community. Testate amoeba‐based water table reconstructions indicated recent drying. This may be due to the lowering of the water table either from accelerated Sphagnum increment or enhanced evaporation. The observed hummock establishment concurs with the documented hemisphere‐wide expansion trend of hummock communities in fens. This change may strengthen the carbon sink and storage capacity of these peatlands, which could be viewed as a welcome negative feedback process to the ongoing climate warming. However, the change also poses a threat to biodiversity since fens are not only species‐rich habitats but are also endangered ecosystems.

Northern peatlands are globally significant ecosystems in terms of both carbon (C) storage and biodiversity: they store approximately 415 Gt of C (Hugelius et al. 2020), and host a large variety of unique, highly specialized flora and fauna, many of which are considered endangered both in Finland and in Europe (e.g.Rydin & Jeglum 2006;European Commission 2017;Hyv€ arinen et al. 2019).The fate of this C storage as well as specialized peatland species under increasing anthropogenic pressure, such as changing climate conditions, is uncertain (e.g.Herrera-Pantoja et al. 2012;Loisel et al. 2020).
The ongoing climate change is especially marked at high latitudes, where the most sensitive peatland types (i.e.rich fens) are mostly located.Recent studies have shown that the Arctic zone is warming four times faster than the rest of the planet (Rantanen et al. 2022), and increased evaporation has been predicted to cause enhanced soil drying and a decrease in peatland water tables in many regions (Roulet et al. 1992;Gong et al. 2012;Herrera-Pantoja et al. 2012;Laîn e et al. 2014;Helbig et al. 2020).In addition to climate change, many European peatlands previously studied for palaeoecology are also affected directly by land use (e.g.drainage for agriculture or forestry) or indirectly via, for instance, atmospheric deposition of nitrogen, or alteration of regional hydrology (e.g.Swindles et al. 2019).
Northern peatlands can be classified into two major categories that represent different successional stages: minerotrophic fens, which receive additional water and nutrients from the surrounding mineral land, and ombrotrophic bogs, which only receive water and nutrients from rain and the atmosphere.A transition from a fen into a bog is a result of peat accumulation and vertical growth, and it is part of autogenic peatland succession (e.g.Rydin & Jeglum 2006;Tuittila et al. 2013).The process can be accelerated by land use change or by climate change that causes a sufficient decline in the water table or reduction of nutrient-loaded spring floods to isolate the peatland surface from minerotrophic water sources (Hughes 2000;Hughes & Barber 2003;Granath et al. 2010;Tahvanainen 2011;Finsinger et al. 2013;Sallinen et al. 2023).
In comparison with bog species, rich fen taxa typically demand a high water table level and high availability of base cations and, consequently, are less adaptive to changing ecohydrological conditions, such as drying (H ajek et al. 2022), which is manifested as the tendency of fens to readily respond to water level drawdown through a drastic species turnover (Kokkonen et al. 2019).In accordance with the sensitivity of fen species to ecohydrological changes, several studies at northern high latitudes have reported a recent expansion of dry habitat Sphagna in northern fens, which could indicate an ongoing regime shift (e.g.Pedrotti et al. 2014;Kolari et al. 2021Kolari et al. , 2022;;Robitaille et al. 2021;Granlund et al. 2022;Magnan et al. 2022;Piilo et al. 2023).In addition, testate amoeba-based water table reconstructions suggest a trend of deepening water tables in nonpermafrost peatlands throughout Europe during the past c.300 years (Swindles et al. 2019;Zhang et al. 2022).These observations have raised the question of whether the ongoing climate change is extensively driving fenbog transition in the northern peatland areas that are currently fen dominated.
In this study, we address the question by quantifying past and ongoing changes in a subarctic non-permafrost fen, a peatland type that has received little attention.As the importance of replicates has been emphasized in previous studies (e.g.Piilo et al. 2020), we conduct a highresolution analysis on four peat profiles using two complementary palaeoecological proxies: plant macrofossils and testate amoebae.This study aims to investigate the dynamics of the prevailing hummock-lawn mosaic at the mire margin and to assess the stability of prevailing communities over the past centuries.

Study site
The study site is a subarctic fen called Lompoloj€ ankk€ a.It is located in Finnish Lapland, 150 km north of the Arctic Circle, in Pallas-Yll€ astunturi National Park, near to Pallas-Ounastunturit fell chain and in the catchment area of Lake Pallasj€ arvi (67°59.835 0N, 24°12.546 0E; Fig. 1).The area is part of the northern aapa mire region.The Finnish Meteorological Institute has conducted continuous greenhouse gas measurements at

BOREAS
Recent hummock establishment in the margin of a subarctic fen, Finnish Lapland Lompoloj€ ankk€ a since 2006, providing useful background information of the study site.The annual mean temperature in the area is À0.55 °C and the annual mean precipitation sum is 548 mm (Finnish Meteorological Institute 2022a, b).Both variables show an increasing trend during the measurement period (Finnish Meteorological Institute 2022a, b; Fig. 2).
Lompoloj€ ankk€ a is an aapa mire, or a minerotrophic valley fen (Zhang et al. 2020), characterized by mesotrophic Caricion Fuscae-type vegetation (sensu Peterka et al. 2017).Peatland initiation began nearly 10 000 years ago, and ~2.5 m of peat has accumulated to date (Mathijssen et al. 2014).The hydrology is strongly affected by a small stream that flows through the peatland.The current vegetation composition varies according to the moisture and nutrient gradients: Salix lapponum is highly abundant around the stream, whereas most of the fen is dominated by Carex spp.(e.g. C. lasiocarpa, C. rostrata, C. canescens) and forbs (e.g.Comarum palustre, Menyanthes trifoliata).The moss cover is discontinuous, characterized by fine-scale variation, and is dominated by minerotrophic peat mosses, such as Sphagnum riparium, Sphagnum teres, Sphagnum warnstorfii, Sphagnum subsecundum and Sphagnum fallax.Some eutrophic brown mosses, such as Scorpidium spp., are also present at the site.
Peat coring points were located in the margins of Lompoloj€ ankk€ a (Fig. 1).Typical of an aapa mire (Rydin et al. 1999;Laitinen et al. 2007), the margins at Lompoloj€ ankk€ a are naturally drier and more nutrient poor than the central fen.Also, the species composition at the coring locations differed from that of the central fen area and was characterized by shrubs and peat mosses, such as Sphagnum fuscum and S. warnstorfii.
While Lompoloj€ ankk€ a is situated in a scarcely populated area, where anthropogenic impact is relatively low when compared with many European peatlands (e.g.Swindles et al. 2019), it should be noted that there are shallow ditches in the vicinity of the study site, and the small road that allows access to the study site was constructed in the early 2000s.

Fieldwork
To investigate the recent development of the microforms in the mire margin, we collected four peat profiles in September 2020 from four low hummocks.The coring points were located within few metres from each other.We used a box corer to collect approximately 50-cm-deep peat cores (hereafter profiles), which were wrapped in plastic and packed inside plastic tubes for transportation.
Testate amoeba samples were collected in 2021 and 2022 to create a local training set for water table reconstruction.The peat samples (5 9 5 9 5 cm) were excavated with a knife and packed in plastic bags (Charman et al. 2000).A subset of the 33 samples (n = 11) was collected from the permanent greenhouse gas measurement points in Lompoloj€ ankk€ a, from which the water table was measured throughout summer 2021.The remaining sample subset (n = 22) was collected from a neighbouring fen, Lompolovuoma, when the water table was measured only once during sample collection.

Chronology
Chronology for the peat profiles was determined by applying Pb-210 and C-14 dating methods.Pb-210 dating was performed at the Department of Chemistry, University of Helsinki.A dried subsample of bulk peat (>0.2 g) was analysed for each 1 or 2 cm interval by performing polonium (Po) precipitations as spontaneous electrodeposition of Po on silver discs, and by measuring the discs with alpha-spectrometry.As Pb-210 dating is suitable for samples younger than 150 years (Appleby & Oldfield 1978;Turetsky et al. 2004), three samples from each core were also C-14 dated in the Finnish Museum of Natural History (LUOMUS, Helsinki, Finland).The samples consisted of either bulk peat, from which rootlets were removed, or selected plant macrofossils (Table 1; Holmquist et al. 2016).Age-depth models were produced using the Rplum package (Blaauw et al. 2021), which combines Pb-210 and C-14 data, in R version 4.2.2 (R Core Team 2022).

Peat properties
Peat bulk density was derived from the dry mass of 5 cm 3 fresh volumetric samples.Peat carbon content and C/N ratio were measured with Leco TruSpec Micro CHNS at the Department of Environmental Sciences, University of Helsinki.

Plant macrofossils
Changes in vegetation were investigated with plant macrofossils.The peat columns were first analysed at 4cm resolution, and then at 1-cm resolution, where the strong changes in plant assemblages were detected.The analysis followed the Quadrat and leaf count protocol first introduced by Barber et al. (1994) and since developed by V€ aliranta et al. (2007).Subsamples of 5 cm 3 were taken from each 1-cm-thick slice and washed over a 140 lm sieve.The retained remnants were placed in a transparent Petri dish with 2-3 mm of water, and the proportion of each peat component was determined by estimating their cover by examining the samples systematically under a stereomicroscope.Sphagnum species were identified further to the species level using a light microscope: a random selection of approximately 100 Sphagnum leaves was identified and the results were expressed as percentages of the total Sphagnum proportion.The R package RiojaPlot (Juggins 2022a) was used to produce the stratigraphic diagram.Non-metric multidimensional scaling (NMDS) ordination was performed using the R package vegan (Oksanen et al. 2022).

Testate amoebae and water table reconstructions
Testate amoebae are a commonly used proxy to reconstruct past changes in peatland water table depth (e.g.Tolonen 1986;Warner & Charman 1994;Zhang et al. 2018a).Testate amoeba assemblages were analysed at 2 cm intervals.The processing of testate amoeba samples followed a modified version of the standard method (Booth et al. 2010).Volumetric peat samples (4 cm 3 ) were boiled in distilled water for 90 min, after which they were then sieved with a 300-lm mesh and back-sieved with a 15-lm mesh.Material retained on the

BOREAS
Recent hummock establishment in the margin of a subarctic fen, Finnish Lapland 15-lm sieve was allowed to settle overnight.A total of 150 individual shells were counted for each sample (Payne & Mitchell 2009) whenever possible.Additional potassium hydroxide (KOH) treatment was applied to the most decomposed samples to remove some of the organic material.The shells were identified to species or 'type' level under a light microscope at a magnification of 200-400.Taxonomy followed Charman et al. (2000), supplemented with online sources (Siemensta 2022).For the water table reconstruction, taxa were grouped according to Amesbury et al. (2016).Stratigraphic graphs were produced using the R package RiojaPlot (Juggins 2022a).
Past water table fluctuations were reconstructed based on a local modern training dataset of 33 samples using R-package Rioja (Juggins 2022b) in R version 4.2.2.Based on absolute water table depths, we calculated the z scores as follows: z > 0 indicates drier than average and z < 0 wetter than average conditions (Amesbury et al. 2016).

Chronology
The basal ages of the profiles varied widely: the basal age of Profile 4 at 59-60 cm depth was 3655 a BP (before present), while the basal age of Profile 1 at 37-38 cm depth was 948 a BP.The basal ages of Profiles 2 and 3 were 2287 a BP at 44-45 cm, and 2616 a BP at 52-53 cm, respectively (Fig. 3A-D).The age-depth model was based on overlapping Pb-210 and C-14 dating methods, which produced different age estimates for specific depths and is indicative of the uncertainty associated with the results.The model preferred Pb-210 ages over C-14 ages, which left the C-14 from the middle of Profiles 3 and 4 and from the surface of Profile 4 as outliers.

Changes in the vegetation composition
A clear species-turnover in the vegetation was detected in all four profiles (Fig. 4A-D).Before 1900 CE, the vegetation resembled the communities that currently prevail in the mire centre: the extensive coverof Cyperaceae, as well as the presence of brown mosses but only few Sphagna, was indicative of relatively wet and nutrient-rich conditions.During the 20th and 21st centuries, the abundance of Acutifolia-type Sphagnum mosses increased drastically, whereas the previously dominant Cyperaceaeassemblage nearly disappeared.The timing and the rate of the change differed among the profiles.
In the NMDS ordination (Fig. 5), the x-axis (NMDS1) reflects a fen-bog gradient driven by pH, nutrient status and water table level, where typical fen taxa are concentrated at the left end of the axis and bog taxa at the right end of the axis.The NMDS2-axis reflects variation within the fen-bog related gradient.
In Profile 1 (current hummock), Cyperaceae cover had already decreased in the 1800s, and was replaced initially by Polytrichum sp. and shrubs.The high proportion of unidentified organic matter (UOM) implied that a substantial part of the peat was highly humified and could not be identified, which hindered the interpretation.Sphagnum fuscum appeared in the 1980s and was dominant by 1997 CE.During the 1990s, the proportion of S. fuscum increased from 15% to >80% in less than a decade.Pleurozium schreberi appeared c.2014 CE.
In Profile 2 (current hummock), Cyperaceae disappeared rather drastically after the 1960s, and were directly replaced by S. fuscum, while P. schreberi was dominant c. 1996-2014 CE.
In Profile 3 (current hummock), the proportion of Cyperaceae decreased gradually after 1920 CE, while the proportion of Sphagnum sect.Acutifolia and shrubs became more abundant.From c. 1970 CE onwards, S. fuscum was clearly the dominant species, while P. schreberi appeared in 1988 CE.
In Profile 4 (current hummock), the change occurred later but was more rapid than in the other profiles: Cyperaceae disappeared after 2000 CE, and S. warnstorfii was the dominant species by 2006 CE.

Testate amoebae and water table reconstructions
Testate amoeba assemblages were analysed for Profiles 1 and 2. In total, 57 testate amoeba taxa were identified.Selected taxa are presented in Figs 6, 7.
In Profile 1, the data covered the timespan from 1818 CE to the present.Until the 1990s, the community was characterized by typical fen taxa, i.e.Centropyxis platystoma, Cyclopyxis arcelloides type, along with Difflugia pulex.The latter was replaced by E. rotunda in the 1990s, which was shown in the water table reconstruction as a wet shift.From 2014 CE onwards, generalists, such as Corythion-Trinema type, Assulina muscorum, Hyalosphenia papilio and Hyalosphenia elegans, took over (Zhang et al. 2018b), which indicated a shift towards drier conditions typical of hummocks.
Testate amoeba data from Profile 2 encompassed a longer time span, from the Medieval Climate Anomaly (MCA, 1178 CE) to the present.The data suggested fluctuating water table levels during both the MCA and the subsequent cold climate phase, the Little Ice Age (LIA), i.e. until the late 1800s.Fen-related taxa, including C. arcelloides type and C. platystoma persisted until 1990 CE.The dry peaks in the 1500s and the late 1800s were driven by Bullinularia indica, Trigonopyxis arcula, and Heleopera sp.At the turn of the 21st century, H. papilio, H. elegans, Archerella flavum, Physochila griseola, and A. muscorum were very abundant, which was reflected as a shift towards drier conditions in the water table reconstruction.
The first NMDS axis (Fig. 7) ordered the samples from the oldest to the most recent reflecting the overall change

BOREAS
Recent hummock establishment in the margin of a subarctic fen, Finnish Lapland in time in the testate amoeba community.The variation in the second NMDS axis was not related to any environmental variation and did not show a clear trend but appears to be more random.
In both profiles, it was notable that the change in the testate amoeba community occurred only after the change in the plant community composition: in Profile 1, the first signs of the vegetation change were visible already from the 19th century onwards, and the rapid Sphagnum expansion took place in the 1990s, whereas the major shift in the testate amoeba community was first noticeable around 2002, and even more clearly after 2014.Similarly, in Profile 2, the major changes in the vegetation occurred around the beginning of the 1960s, followed by a change in testate amoebae in the 1990s and the early 2000s.

Peat properties
Peat bulk densities varied between 0.024 and 0.148 g cm À3 .In all four profiles, peat bulk density values declined towards the surface.In Profiles 1, 2 and 4, % % % % % % % % % g mL -1 a steep decline accompanied the change in peat type (from sedge peat to Sphagnum peat).In Profile 3, the decline was more gradual, as was the change in peat type.

S . w a r n s t o r f i i S . s t r a m i n e u m
The low degree of decomposition in the surface peat, indicated by low or absent UOM values, is associated with the decline.In all four profiles, the C/N ratio increased rapidly towards the surface and followed the change in peat type.The sedge peat exhibited a rather constant C/N ratio of around 20, although the C/N ratio increased up to 40-70 when the peat type changed to Sphagnum peat (Fig. 4A-D).

Discussion
While hummocks are widely observed at the margin of aapa mires (Laitinen et al. 2007), our palaeodata reveal that, in our study site, they have only appeared since the onset of the ongoing warming phase in the early 1900s, with fen-typical taxa persistent throughout the previous variable climate phases.Next, we discuss the potential causes and implications of the observed change.

Medieval Climate Anomaly and Little Ice Age
During the MCA (c.800-1200 CE), the vegetation in the Lompoloj€ ankk€ a margin was characterized by the dominance of sedges accompanied by a small number of Sphagna and Polytrichum species, which are well adapted to dry conditions.In northern Fennoscandia, the MCA was a relatively warm and dry period (Helama et al. 2009b;Hanhij€ arvi et al. 2013;Linderholm et al. 2018).While shifts from fen vegetation into a S. fuscum community have occurred in other northern Fennoscandian peatlands during the MCA (Zhang et al. 2018a), such change was not detected in our data.Even though palaeosamples might underestimate the original abundance owing to degradation, Sphagnum cover during the MCA in our samples was probably not as extensive as currently.In Profile 4, the presence of Sphagnum was not only restricted to the MCA, as Sphagnum, to a limited extent, was also present throughout most of the profile.Only two samples in our testate amoeba data covered the MCA period (Profile 2) and the presence of B. indica and T. arcula in those samples suggests relatively dry conditions, which was also reflected in the water table reconstruction.
During the LIA (c.1250-1900 CE), which was a cold climate period in northern Europe (Hanhij€ arvi et al. 2013), no major shifts in the plant record were detected, although the vegetation remained largely sedge dominated.In contrast, the testate amoeba taxa varied more and as a result, the water table reconstruction fluctuated between drier and wetter phases within the LIA.This observed fluctuation is in agreement with the inconsistent hydrological patterns for the LIA in Fennoscandia reported by Linderholm et al. (2018).

Recent community shift in Lompoloj€ ankk€ a suggests ongoing change in northern peatlands
The major transition in the vegetation during the past century was characterized by the disappearance of Cyperaceae, and the expansion of Acutifolia-type Sphagna, along with the appearance of forest mosses.This shift reflects an adaptation to nutrient-poor and dry conditions.The onset of these changes coincidedwith the beginning of the post-LIA warming, which started in Finnish Lapland in the mid-19th to early 20th century (Helama et al. 2009a;Linderholm et al. 2018; Fig. 4A-D).The onset of Sphagnum expansion in our study site varied from 1920 to 2000 CE, and its rate, which varied from decades to years, has accelerated towards the present.After 2000 CE, the change in the vegetation was followed by a compositional change in the testate amoeba communities from typical fen taxa to assemblages more typical of hummocks and ombrotrophic peatlands (Lamentowicz et al. 2007;Zhang et al. 2018b).This order of the change, i.e. vegetation before testate amoebae, was unexpected as testate amoebae are considered to be a more sensitive hydrological proxy than plants (V€ aliranta et al. 2012).
The water table reconstructions based on the testate amoeba composition from Profiles 1 and 2 indicated rather constant and intermediate conditions for most of the 20th century.Profile 1 showed a wetter phase during the 1990s, supported by the disappearance of Polytrichum spp., with the site turning drier from c. 2009 CE onwards (Fig. 6A).Profile 2 indicated a turn towards drier conditions after 1990 CE (Fig. 6B).The appearance of P. schreberi at the turn of the 21st century and again in c. 2015 CE supports our reconstruction.Since a drying trend was indicated by the testate amoeba composition in both cores and supported by plant data, our results agree with previous studies that indicate a drying trend in many northern peatlands (Swindles et al. 2019;Zhang et al. 2022).However, the water table reconstructions include uncertainties, such as differences in the vertical micro-distribution between testate amoeba taxa.Living testate amoebae can be found even at 10 cm depth, especially taxa that use mineral particles in their construction (Jassey et al. 2011;Roe et al. 2017).
As we did not separate living and dead specimens, we cannot conclude with certainty that the apparent changes observed in the uppermost 10-cm layer, e.g. the T. arcula peak in the 1990s, are not merely due to living specimens inhabiting these peat layers.However, the taxa that appeared only in the upper-most layer certainly represent a recent change.
Since the 1960s, the mean annual, summer and September temperatures in our study area have risen by approximately 2, 0.5 and >1 °C, respectively (Fig. 2).Sphagnum growth is known to benefit from warmth, provided that moisture availability is sufficient (Dorrepaal et al. 2004;Breeuwer et al. 2008;Keuper et al. 2011;Loisel et al. 2012;Bengtsson et al. 2021;K€ oster et al. 2023).This may have given Sphagna a competitive advantage over vascular plants.K€ uttim et al. (2020) have reported greater moss length growth in autumn than in summer, which could indicate that Sphagna have a better ability than vascular plants to take advantage of autumn 0 2 0 5 10 0 5 10 0 5 10 0 5 10 0 25 50 0 5 10 0 5 10 0 5 10 0 5 10 % % % % % % % % % % % % % % % % % % % % scores warming.Since the prevailing S. fuscum community is better adapted to drier conditions than the previous Cyperaceae-dominated vegetation, the observed shift could also reflect peatland drying owing to increased evaporation.While this cannot be completely ruled out based on the current data, the water table reconstruction and the fact that major changes occur in vegetation before testate amoebae rather indicate that the deepening of the water table relative to the surface is a result of vertical Sphagnum growth (see also Kolari et al. 2021).
The observed changes in the testate amoebae community may as well reflect changes in water chemistry associated with Sphagnum establishment, such as Sphagnum metabolites (Jassey et al. 2011;Sytiuk et al. 2021) and lowered pH (Heal 1964;Lamentowicz et al. 2007).
In addition to the climatic drivers, the impact of anthropogenic disturbances in the study area cannot be ruled out entirely.A fen-bog-transition in a northern fen has earlier been chronologically associated with road construction (Finsinger et al. 2013).However, the road allowing access to our study site was built at the beginning of the 21st century, therefore the roadwork cannot explain the detected changes that began already some decades before.In addition to the access road, there are also some small ditches in the proximity of our study site.Unfortunately, we were not able to exactly date their digging, but found out that the ditches are missing from the 1968 map (Kutilainen 2023).The current overgrowth suggests that they were dug several decades ago but after the map from 1968 was drawn.Thus, while it seems likely that the ditches do not explain the first signals of change from the late 19th and early 20th centuries, they may have accelerated the process that was naturally triggered earlier.
The recent community changes seem extraordinary, since the past climate phases covered in the data have not triggered similar hummock growth in our study site (Fig. 4A-D; Mathijssen et al. 2014).Over the millennialong timespan covered in our dataset, the forest moss, P. schreberi, appeared in our profiles for the first time in the 1980s.Similarly, certain testate amoeba taxa only appeared in recent decades.
Our observations are in line with the recently documented process in which hummock species have been observed to expand over northern fens (Pedrotti et al. 2014;Kolari et al. 2021Kolari et al. , 2022;;Magnan et al. 2022;Granlund et al. 2022;Piilo et al. 2023).Often, the change has been described as an ecosystem regime shift, i.e. a fen-bog transition (Magnan et al. 2022;Kolari et al. 2022).In our case, new hummock communities were only established recently in the mire margin, which is considered more sensitive to hydrological changes than the central parts of an aapa mire.While these hummocks could represent a starting point for a wider regime shift, ecosystem-scale ombrotrophication in the near future does not seem likely in this particular peatland, as the central parts of the fen receive minerogenic water from a stream and mesotrophic Sphagnum mosses, sedges and forbs remain abundant.Palaeovegetation from the central part of Lompoloj€ ankk€ a has been studied by Mathijssen et al. (2014), but they did not detect notable changes comparable with ours.
Our results confirm previous work on the substantial internal variation of peatlands (e.g.Piilo et al. 2020).This was also demonstrated by the observed variation in timing and rate of change, even though our four profiles were located within a few metres of each other.

Implications for carbon dynamics and biodiversity
Currently, the fate of the carbon stored in peatlands remains uncertain.The shift from sedge communities to Sphagnum domination often implies decreased methane (CH 4 ) emissions and enhancement of the C sink function (e.g.Turunen et al. 2002;Riutta et al. 2007;Turetsky et al. 2014;Mathijssen et al. 2016;Loisel & Bunsen 2020).Moreover, peatland models have suggested an increasing capacity of C sequestering in northern peatlands (Gallego-Sala et al. 2018).Despite its substantial C sink, Lompoloj€ ankk€ a has had a net warming impact on the climate owing to its high CH 4 emissions (Aurela et al. 2009;Mathijssen et al. 2022).Despite the change towards drought-adapted communities at Lompoloj€ ankk€ a margins, the water table level remained high in the central parts of the fen even during the extremely warm and dry summer of 2018 CE.In the stream-and groundwater-fed parts of Lompoloj€ ankk€ a, increased peat temperatures and unchanged water table levels led to exceptionally high CH 4 emissions, while in four other adjacent peatlands, the extreme weather event led to decreased emissions (Rinne et al. 2020).Even though the shift to dry Sphagnum-dominated communities seems to be a widely occurring trend throughout the northern hemisphere, vast areas currently remain sedge-dominated, open fens.Those peatlands typically emit high amounts of CH 4 , whose production is largely controlled by hydrology and temperature (Dunfield et al. 1993).This highlights the high internal variation in peatland C dynamics, which has been also observed in palaeostudies with multiple study points per peatland (e.g.Mathijssen et al. 2014Mathijssen et al. , 2016Mathijssen et al. , 2022;;Piilo et al. 2019Piilo et al. , 2020)).While the C accumulation rate may increase following Sphagnum expansion, CH 4 fluxes from sedge-dominated communities may increase with warmer soils if the water level stays sufficiently high in the peat profile.On the other hand, prolonged drought or drying owing to land-use change may lead to net C losses owing to increased decomposition (e.g.Laine et al. 2019).The net C balance (including lateral C fluxes) of a peatland depends on the tradeoff between these flux components.
As stated by several authors (e.g.European Commission 2017; Kolari et al. 2021;Granlund et al. 2022;H ajek et al. 2022), hummock expansion over wet fen areas poses an additional threat to fen habitats and to their specialized species that are already under pressure from the anthropogenic impact.Warming is known to benefit mire generalists and forest taxa at the expense of fen-specialized mosses and herbs (M€ akiranta et al. 2018;Kokkonen et al. 2019), at least partly because the pH niche of fen specialists becomes narrower under warming climate conditions, making them weaker competitors (H ajek et al. 2022).Another possible successional pathway for nutrient-rich peatlands following water table drawdown is a regime shift into a forest ecosystem.In nutrient-rich fens, succession may lead to a closed-canopy forest in as little as two decades (Kokkonen et al. 2019).In both cases, the fen habitat converts into a new habitat, and its species are lost from the area.

Conclusions
Our data indicate a clear change in the vegetation type and associated testate amoeba community in all four study points in the Lompoloj€ ankk€ a mire margin.This finding suggests that dry hummock habitats are expanding from the mire margin towards the central area of this fen.This change commenced in the 20th century and has been accelerating towards the present day.Our results show that the establishment of Sphagnum hummocks over fen vegetation in aapa mire margins may occur in less than a decade.Our study timespan covered a period from one to nearly four millennia but no previous changes comparable with the recent shift were detected.Neither was such development observed previously in a Holocene-scale peat study from the same peatland.While the impact of local anthropogenic activities cannot be completely ruled out, the changes observed in this study concur with similar findings that have been reported throughout the northern high latitudes that have been interpreted as climate changeinduced.Thus, we suggest that the recent changes in Lompoloj€ ankk€ a reflect a hemisphere-wide trend, which is likely to affect peatland carbon dynamics and biodiversity, thereby posing a threat, especially to species adapted to fen habitats.

C
Fig. 1.A, B. The location of Lompoloj€ ankk€ a peatland.C. The location of the study points in Lompoloj€ ankk€ a peatland.D. The study site: mire margin in Lompoloj€ ankk€ a. E. Coring points 1-4.

Fig. 4 .
Fig. 4. A-D.Selected taxa in palaeovegetation and peat properties in Profiles 1-4.Taxa that represent >5% of the community in at least one sample are presented.Note the varying scaling of x-axes.UOM = Unidentified organic matter.Years CE = Common Era; BCE = Before Common Era.

Table 1 .
Description of the 14 C dating samples.