Abrupt ombrotrophication of a boreal aapa mire triggered by hydrological disturbance in the catchment

Summary

1. Hydrological changes due to drainage and climate warming can have great impact on the ecosystem balance of boreal mires. The possibility of ombrotrophication, i.e. the development from fen to bog, in response to altered hydrology has not been previously tested. Here, recent changes in vegetation and surface peat are studied in an aapa mire, a typical boreal mire system dominated by fen vegetation. Drainage in the catchment from 1968 onwards led to the change from richly minerogenous to ombrogenous hydrology, thus providing a long-term ombrotrophication experiment.

2. A sequence of aerial photographs (1941, 1953, 1965, 1974, 1984, 1995, 2005) revealed a dramatic shift from fen vegetation to the nearly complete dominance of peat mosses (Sphagnum) within two decades after the catchment disturbance.

3. A distinct change from Carex peat to Sphagnum peat at the average depth of 23.3 cm (SE 0.8 cm) was found in 18 peat cores. All of the new Sphagnum peat had accumulated within the last four decades. This was verified by the relationship of age and rooting depth of 37 small pines (Pinus sylvestris) and by two pollen density profiles. The ratio Ca/Mg diminished towards the surface of peat profiles indicating change from minerogenous to ombrogenous hydrology. In accordance, extremely low pH (range 3.8–4.2) and conductivity (average 14.5 μs cm⁻¹) were measured in the surface pore water.

4. The average total dry mass of new Sphagnum peat was 7042 g m⁻² (SE 442) and the recent apparent rate of carbon accumulation was 100.6 g m⁻² year⁻¹ (SE 6.3), as calculated for a 35-year period and 50% carbon content.

5. Synthesis. Remarkable potential for vegetation change and increase of peat growth is demonstrated in boreal aapa mires. Ombrotrophication can be initiated within a few decades in response to reduced input of minerogenous water. Future changes in the hydrological cycle, as indicated by climate change models, are similar to the impact of catchment disturbance in aapa mires. Diminished total water budgets during the summer cause a decrease of minerogenous input and a draw-down of water level, both of which may promote the growth of Sphagnum over fen vegetation.

Introduction

Circum-boreal and subarctic mires are peat-accumulating wetlands that contain a large carbon pool (270–370 Pg) that represents almost one third of the global amount of carbon stored in soils (Turunen et al. 2002a). Short-term studies have shown that carbon sinks in mires are sensitive to weather conditions and water level fluctuations. In warm and dry years, mires may turn into carbon sources due to the increased decomposition induced by aeration of surface peat (e.g. Alm et al. 1999; Bubier et al. 2003). Hence, it has been suggested that the role of northern mires as a carbon sink may weaken in response to the hydrological changes caused by climate warming (e.g. Freeman, Ostle & Kang 2001; Kayranli et al. 2010). However, most studies have focused on raised bogs in the temperate to south-boreal zones, while boreal fens such as aapa mires have received less attention. In these two main types of northern mire systems, the potential impacts of changing hydrology are arguably different.

Aapa mires are minerotrophic mire systems dominated by fen vegetation (Ruuhijärvi 1960; Laitinen et al. 2007). In aapa mires, minerogenous water from the catchment enters the central fen areas, while marginal areas are often characterized by bog vegetation. A contrasting hydrological pattern is found in raised bogs, where central bog areas are elevated above the reach of minerogenous water flow and fen vegetation is found in narrow marginal areas only. The geographic zonation of raised bogs and aapa mires in Fennoscandia and northwestern Russia is a well-known case of climate-correlated distribution patterns among mires (Ruuhijärvi 1960; Botch & Masing 1979; Parviainen & Luoto 2007). In Finland, the climatic limit between raised bogs to the south and aapa mires to the north roughly equates with the 1100 °C isocline of growing degree days above +5 °C (Ruuhijärvi 1960). Climate change scenarios imply dramatic changes to this zonation. For example, according to several climate models (see http://www.finessi.info/finessi/), the A2-climate scenario (Nakicenovic & Swart 2000) would result in the northward movement of the 1100 °C isocline by 600–700 km within 100 years. The straightforward implication from such models is that aapa mires might start to develop into raised bogs. The results of such a major, ecosystem-scale response are highly speculative, however, since most of the knowledge on the ombrotrophication process is based on palaeoecological studies and there are no experimental studies exploring its mechanisms and potential speed within contemporary mires.

The development of raised bogs includes the isolation of the mire surface from minerogenous water that eventually follows from the accumulation of peat. This autogenic succession has probably been the main mechanism of ombtrotrophication during the Holocene and it is generally considered to require moist climate conditions in order to proceed (Rydin & Jeglum 2006). However, the ombrotrophication process can also be initiated by allogenic hydrological changes (Hughes & Barber 2004). The existence of several possible mechanisms of ombrotrophication may explain part of the inconsistency between climate variation and ombrotrophication events during the Holocene (Kuhry et al. 1992; Korhola & Tolonen 1996; Hughes & Barber 2004). Indeed, ombrotrophication has been connected to warm and dry climatic periods in some cases (Almquist-Jacobson & Foster 1995; Pajula 2000), as well as to cool and moist periods in other cases (Robichaud & Bégin 2009). Considering the climate feedback of mires in the future, it is important to recognize the possibility that hydrological changes may trigger the allogenic ombrotrophication process in northern fens.

Bogs are typically dominated by Sphagnum mosses and dwarf shrubs, while Amblystegiaceae mosses and sedges are characteristic to fens (Gorham & Janssens 1992). In the strict sense, the complete change from fen to bog vegetation (e.g. Sjörs & Gunnarsson 2002) is a very slow process because many deep-rooted fen species that are absent from true raised bogs (e.g. Menyanthes trifoliata, Carex rostrata, Eriophorum angustifolium) may survive for long periods after changes in habitat conditions (Gorham 1950, 1957). However, it can be argued that the most significant phase of the transition is the establishment of habitat conditions and the dominant vegetation becoming that typical of bogs, rather than the final disappearance of the last fen plants. Several palaeoecological studies have shown that the dominance of ombrotrophic Sphagnum species over fen vegetation can develop within decades or centuries (see Gorham & Janssens 1992; Kuhry et al. 1993), i.e. very rapidly considering the multi-millennial history of mire development. I refer to the development of dominant bog vegetation as ‘ombrotrophication’ and consider the time-lag of the complete exclusion of deep-rooted fen species as ‘biological inertia’sensuGorham (1957). Within these frames it can be hypothesized that ombrotrophication may take place as a response to the allogenic change from minerogenous to ombrogenous hydrology.

The average annual hydrograph differs between the raised bog and the aapa mire zones, the latter being characterized by later and richer peak discharge after snow melt and by richer discharge also during the growing season (Sallantaus 2006). Respectively, the total water budget is greater in aapa mires than in raised bogs. Furthermore, the rich minerogenous hydrology of aapa mires has an impact on water quality. Mineral alkalinity and rich flushing are important factors in maintaining the typical water pH (5–7) of fens (Tahvanainen et al. 2002; Siegel, Glaser & Janecky 2006). The changes in the hydrological cycle expected to follow from climate change include increased winter precipitation, increased summer evaporation, diminished average flood discharge and earlier timing of spring floods (Silander et al. 2006; Lotsari et al. 2010). Altogether, these changes would force the hydrology in the aapa mire zone towards that of the contemporary raised bog zone.

At the local scale, catchment hydrology determines the water source of mire basins. Especially large catchment area may result in sufficiently rich minerogenous water input to maintain aapa mires in the raised bog zone. Correspondingly, extra-zonal raised bogs are met in the aapa mire zone in places where minerogenous water input is reduced by special catchment conditions, such as within river bends (Ruuhijärvi 1960). In general, both the local and the climate-correlated distribution of raised bogs and aapa mires are connected to hydrology (see Sallantaus 2006). Disturbance in the catchment that reduces input of minerogenous water in the mire may therefore have similar impact as climate change on the hydrology and future development of aapa mires. Such disturbance can be produced by drainage activities in the catchment that redirect minerogenous water flow away from the mire basin.

In this study, I explore the response of vegetation and surface peat to hydrological disturbance caused by drainage in the mire catchment. The treatment reduces minerogenous water input to the mire, thus changing the hydrology towards ombrogenous conditions. The main hypothesis is that such a change should initiate the ombrotrophication process in mire vegetation. Since there are no long-term monitoring data of mire vegetation available, a multi-proxy approach using aerial photographs and surface peat stratigraphy was used to answer the question: how rapid can the ombrotrophication process be, and does it immediately lead to increased peat accumulation?

Materials and methods

Study area

The object of this study is a patterned aapa mire, which forms a clearly separable hydrological unit of the Valkeasuo mire system (Tolonen 1967). The study site is located 40 km southeast of the city of Joensuu in North Karelia, eastern Finland (110 m asl, N 62°24′, E 30°16′). The area belongs to the middle-boreal climatic-phytogeographical zone (Tuhkanen 1984) and it is located close to the climatically determined limit between the raised bog and the aapa-mire zones (Ruuhijärvi 1960). The mean annual precipitation is 600 mm, the average annual temperature is +2 °C and the average temperatures of the coldest and warmest months are −12 °C and +14 °C, respectively.

Before its extensive utilization, the Valkeasuo mire system was among the biggest continuous mire areas in southern Finland. Tolonen (1967) described the Valkeasuo mire system and its development in detail. The total continuous mire area was approximately 50 km². Typically for larger mire systems in such locations, aapa mire areas with fen vegetation form a mixed mire system with raised bogs (Tolonen 1967). In such mixed systems, aapa mires receive minerogenous water from the catchment and bogs develop in places with weaker minerogenous influence. While most of the Valkeasuo mire area has been devastated by peat mining (c. 22 km²) or directly disturbed by drainage (c. 25 km²), the aapa mire area (c. 80 ha) under investigation has remained untouched (Fig. 1).

Figure 1.  Changes in the study area between 1965 (above) and 2005. The 1965 photograph shows 1-m contours of the slope of the original mire surface, along with black arrows pointing the directions of water flow and the transect in Fig. 2 (white arrow). Drainage ditches and the blocks of peat mining are superimposed on the 2005 photograph. The study site is seen in the lower middle part of the photographs.

In the 1960s, the dominant vegetation type of the study area was minerotrophic flark fen with open water and mud-bottom surfaces (Tolonen 1967). The mire was patterned by low hummock strings covered by sedges (Carex sp., Eriophorum sp.) and scattered cranked Scots Pine (Pinus sylvestris). The landowner remembers the dominance of wet, muddy-bottom surfaces without moss cover in the 1960s, as well as the greenness due to the abundance of sedges along the strings (M. Heiskanen, personal communication). The contemporary vegetation of the study area is characterized by dominance of Sphagnum mosses. The mire margins and the thin-peat pine fen areas (c. 2 km²) of the higher catchment area in the immediate vicinity of the study area were drained in order to promote timber production in 1968 (Fig. 1). According to Markku Heiskanen, the area was ‘easily’ drained thanks to the sandy sub-soil. The drainage activities connected to the peat mining in the Valkeasuo mire system started in 1969 and peat mining was started in 1973. However, the peat mining areas in the catchment of the study area (Fig. 1) were drained in early 1980s. The hydrological disturbance of the catchment has been continuously intensified by improvements of drainage network connected to peat mining, forestry and road construction.

Importantly, the whole of the patterned, central aapa mire area has remained untouched by the drainage activities. Furthermore, the threshold of runoff from the mire has not been severely altered, as the string patterns have remained intact and the downstream drainage has not been so effective. Instead, the input of minerogenous water from the catchment to the aapa mire has definitely been interrupted by the upstream drainage activities. The hydrological patterns before and after the catchment disturbance (Fig. 2) were reconstructed according to peat stratigraphies, elevation and peat inventory data (Tolonen 1967 and unpublished field notes), as well as new field observations. The Ahveninen Lake receives runoff from the mineral soil catchment, in the direction of the water flow path from the catchment towards the aapa mire (Figs 1 and 2). The small but relatively deep (5 m) lake lacks any outlet stream and it is bordered by an excentric bog dome, which does not have any minerotrophic vegetation. The only possible path for outflow from the lake is seepage through an aquifer formed by sand and Carex peat strata underneath the ombrotrophic Sphagnum peat, acting as an aquitard. All drainage ditches between the study area and the mineral soil area of the catchment are at least 1-m deep in relation to the original surface levels. The ditches penetrate well below the level of the peat surface of the study area. Intermediate values of pH (5.6–5.9) and conductivity (53–91 μs cm⁻¹) were measured in the minerogenous water from the ditches draining from Carex peat in the peat mining area.

Figure 2.  Suggested changes in the hydrology of the study area. In the natural conditions, the sorted subsoil and Carex peat strata of the upper mire areas acted as an aquifer and up-welling of ground water characterized the hydrology of the patterned aapa mire (above). Disturbances in the upper catchment have changed the hydrology by redirecting the minerogenous water flow to the drainage network (below). Water flow according to hydraulic head is indicated by arrows.

Analysis of aerial photographs

Aerial photographs over a 68 year period were used to explore the pace and extent of the overgrowth of the study area by Sphagna. In general, Sphagnum mosses have high reflectance and bog areas dominated by Sphagnum are recognized by bright tones in aerial photographs of mire areas. In contrast, surfaces with open water or wet fen vegetation are seen as darker tones. Such contrasts make analysis of aerial images a powerful tool for recognition of patterns in the main vegetation types in mires (e.g. Glaser et al. 1981; Laitinen et al. 2007; Dissanska, Bernier & Payette 2009).

Eight aerial photographs were purchased from the mapping unit (Topografikunta) of the Finnish Defense Forces. The photographs were taken in 1941, 1953, 1965, 1970, 1974, 1984, 1995 and 2005. The 1953 photograph was taken on 19th September, during the autumn senescent period, and the 1941 photograph on 29th May, i.e. in spring. All other photographs were taken during the summer season (June–July). All photos were scanned from original prints to 8-bit greyscale TIFF-files (255 bands) with approximately 1-m field resolution. The greyscales were rescaled, in order to adjust for differences between the photographs (arising from the exposure, light conditions, atmospheric dimming) and enhance their comparability. Prior to further treatment, all photos were smoothed using a Gaussian filter with 5 × 5 pixel overlap, which reduced the noise in the greyscale histograms and diminished the impact of extreme values. Minimum and maximum tones of grey were sought in areas surrounding the study area and set as the black (0) and white (255) bounds of the rescaling. The darkest tones were represented by shadows in conifer forests and the brightest tones were found from Sphagnum surfaces, both of which were found in each photograph. Subsequently, greyscale histograms were determined from a 24.2 ha area of the central, patterned mire area. Exactly the same delineation (shown in Appendix S1, Supporting information) of the mire area was used for each histogram.

Non-metric multidimensional scaling (NMDS) was used in an analysis of similarities between the greyscale histograms derived from the photographs. For each photograph, relative cover of 64 greyscale stacks, each representing eight consecutive greyscale bands, were calculated and used as the scaling variables. The options used in the NMDS were: Euclidean distance measure, maximum of three dimensions, 50 runs with real data using random starting configurations, a stability criterion of 0.00001 and varimax rotation. The NMDS model was then applied to similar histogram samples obtained from an excentric bog area, located 3 km northwest from the study area that was included in the 1965 to 2005 photographs. These bog histograms were included in the analysis as passive samples in order to facilitate the interpretation of the temporal movement of the aapa mire samples.

The photographs were pretreated in Adobe Photoshop CS4. The smoothing and the histogram analyses of the photographs were conducted using ImageTool 3.00. The program PC-ORD 5.0 was used for the NMDS analysis.

Field sampling and peat analyses

Eight evenly distributed flarks were selected from the central mire area (Appendix S2, Supporting information) for sampling of surface peat. At each of the selected flarks, two sampling sites were chosen: one at a Sphagnum papillosum-dominated lawn and one at a Sphagnum majus and/or Sphagnum jensenii-dominated carpet. At each site vegetation was recorded using visual species cover estimates from 1 m² plots. Pipe wells with 0.3 mm slits (diameter 32 mm) supplied with filter gauge (Eijkelkamp, Germany) were inserted into the peat to 40-cm depth for water sampling. The wells were emptied twice before sampling of the water with a vacuum syringe. Measurements of pH and electric conductivity (reference temperature 25 °C) were performed immediately in the field using a Consort 535 multimeter and standard electrodes. Conductivity values were corrected for the specific conductance of protons (350 μs cm⁻¹ meq⁻¹ H⁺) according to the pH measurements. Water-table depth was measured from the pipe wells. Surface peat samples were collected using a sharpened steel cylinder (diameter 115 mm) down to the depth of 32 cm. The samples were cut in the field, longitudinally to quarter size and then into 4-cm vertical segments, and brought to the laboratory for analyses of bulk density (BD) (see below). The sampling was conducted on 17–18 August 2009.

In addition to the surface peat samples, two full peat profiles were collected in October 8th 2009. Profile 1 was located in the lower part (in direction of slope) of the central mire area, approximately at the thickest peat deposit. Profile 2 was located at an upper, central area with average peat thickness. Both profile sites were Sphagnum papillosum-lawns. Peat samples were collected using the steel cylinder to a depth of 35 cm and by a ‘Russian type’ peat corer (half-cylinder with 50 mm diameter) for deeper strata. Two replicate, whole peat cores were transported from each site to the laboratory, where peat samples were carefully cut into vertical segments. The lengths of the segments were adjusted from 1 to 5 cm according to the apparent variation of peat quality down to 30-cm depth, below which the segments were regularly 5-cm long. Volumetric peat samples were dried at 70 °C to constant weight and the BD was obtained as g cm⁻³of dry weight. Ash content (% of dry weight) was measured as the residual mass after ignition in 510 °C. Concentrations of calcium and magnesium were analysed by atomic adsorption spectroscopy from selected samples down to 45-cm depth. The mass ratio Ca/Mg was then used as an indicator of minerogenous versus ombrogenous hydrology (Steinman & Shotyk 1997).

From the second, parallel samples, small volumetric (1–6 cm³) samples were carefully cut and prepared for the analysis of pollen density of P. sylvestris. It can be expected that, under relatively constant pollen deposition, the density of pollen grains in peat depends on the growth rate and decomposition of peat (Middeldorp 1982). Age of a peat stratum can be calculated by dividing the cumulative pollen count by the average annual pollen deposition. P. sylvestris was selected because it has formed the dominant tree stands over the studied period in the region and because its pollen deposition pattern is relatively well known (Hicks et al. 2001). The volumetric samples were mixed with 10% KOH, homogenized and heated for 5–10 min. The samples were then sieved and washed repeatedly through an 800 μm mesh. The samples were centrifuged, adjusted to a known volume (14 ml) and stained with safranin. It was checked under microscope that no abundant pollen was left in the sieved coarse peat material and regular shaking of samples was done before careful pipetting of each subsample. All Pinus pollen grains and Sphagnum spores were counted from repeated (3–8) volumetric subsamples (10 μl) pipetted under 20 × 20 mm cover slips, and average density values were calculated as number of pollen grains or spores per cm³. Standard error was calculated for each average density value and cumulative standard error was used to assess the reliability of the cumulative pollen counts. The average yearly deposition of Pinus pollen was estimated by dividing the cumulative pollen count at a certain depth by the age of small pine individuals rooted at that depth. In addition, two Sphagnum fuscum hummocks were sampled, in order to estimate the local deposition of Pinus pollen. These samples were dated by counting of annual growth increments of Polytrichum strictum.

The degree of humification of peat was determined using the von Post scale (H₁–H₁₀). Plant species composition of peat samples was judged visually and critical samples were later verified microscopically (data not shown). Selected peat samples were also qualitatively explored for diatom assemblages, in order to explain patterns of ash content. For this purpose, small amounts of amorphous peat material were dried on microscope slides, embedded in resin and investigated by light microscope.

Small (< 50 cm) pines (P. sylvestris) were collected from S. papillosum surfaces from the sampled flarks for age determination of surface peat layers. A mark was carved with a knife on each pine at the location of peat surface (bryophyte capitula) and the depth from the root collar to the mark was measured. In the laboratory, several thin cross-sections were made from the basal portion of stem and annual growth rings were counted under microscope for each tree.

Results

Vegetation and habitat conditions

Covering the whole of the central mire area, two distinct vegetation types (S. papillosum lawns and S. majus carpets) were found in the flarks, both of which were dominated by Sphagnum species that are ombrotrophic in the region (see Appendix S2). In the lawns, Sphagnum magellanicum and Sphagnum angustifolium were met frequently in addition to the dominant S. papillosum. Trichophorum cespitosum, Eriophorum vaginatum and Andromeda polifolia were the commonest vascular plant species in the lawns. In the carpets, Sphagnum majus, S. jensenii and S. compactum formed the dominant bryophyte flora. Rhynchospora alba was markedly abundant in all carpets and Carex limosa was found frequently. Deep-rooted vascular plant species considered as minerotrophic indicators included Carex rostrata, Carex lasiocarpa, Eriophorum angustifolium and Menyanthes trifoliata that were all found frequently over the central mire area. In the hummock strings, low-hummock vegetation dominated by S. angustifolium, S. magellanicum and Eriophorum vaginatum and high-hummock vegetation characterized by Sphagnum fuscum and Betula nana were found.

The depth of the water-table was 14–21 cm in the lawns and 1–9 cm in the carpets. The average of water pH was 4.08 in the lawns and 4.04 in the carpets (total range pH 3.81–4.23). The averages of the corrected conductivity values were 14.6 and 8.7 μs cm⁻¹, respectively (total range 3–32 μs cm⁻¹).

Aerial photographs

In the sequence of aerial photographs from 1941 to 2005, a major vegetation shift can be seen to have taken place in the aapa mire (Fig. 3). The dark, blackish flark surfaces (open water, mud-bottom, Carex) between the hummock strings in the 1941–1974 photographs are completely replaced by the whitish cover of Sphagnum in the 1984–2005 photographs. The 2005 photograph appears as an almost negative image of the 1941 and 1965 photographs, as the originally pale hummock strings become darker, probably due to the increase of cover of dwarf-shrubs (mainly B. nana, Chamaedaphne calyculata, Empetrum nigrum). The median greyscale moved from between 62 and 125 (years 1941–1974) to around 148–165 (years 1984–2005). The proportion of cover of pixels with values higher than 128 depicts the increase of Sphagnum, increasing from 30–60% to around 95%.

Figure 3.  Aerial photographs of the study area. Histograms representing percent cover (y-axis = 0–2.5%) of each greyscale band (x-axis = 0–255) of the central mire area. Median greyscale is depicted in each histogram as a vertical line.

The 1941 and 1965 photographs showed a distinct mode of dark tones (around 50) in the greyscale histograms, as produced by the wet flark-fen surfaces (Fig. 3). The 1953 photograph was taken in September when the reflectance of the mire surface was probably increased by autumn colours and senescent leaves, reducing the dark-tone mode of the fen flarks. In 1970, the median of the greyscale histogram was at a comparably high level (125) and the flarks were not as dark as in all previous photographs. In 1974, the dark tones were found again in the flarks.

The NMDS of the proportions of greyscale tones resulted in a two-dimensional ordination of the time series. The Euclidean distances in the ordination explained 96% of the Euclidean distances in the original 64-dimensional space (cumulative R² = 0.958). In the ordination, the 1941–1974 data showed a zigzag change over time, from which the 1984–2005 data were clearly separated (Fig. 4). The 1995 and 2005 photographs gave very similar values to the bog histograms.

Figure 4.  NMDS ordination of the greyscale histograms of the central mire area (filled circles). The bog samples (open circles) represent greyscale histograms from an ombrotrophic bog area between 1965 and 2005 and are included as passive samples in the analysis.

Surface peat accumulation and peat profiles

A distinct transition from Carex peat to Sphagnum peat (C/S transition) was found in all peat cores close to the peat surface. At the S. papillosum lawns, the average depth of the C/S-transition was 24.5 ± 1.1 cm (± SE). At the S. majus carpets the C/S-transition was found at 21.8 ± 1.0 cm depth, on average. There was no significant difference in thickness of the new Sphagnum peat between the S. papillosum and the S. majus sites (t-test, p = 0.128). In most cases, the transition of peat quality was very clearly visible in the field. Microscopic analyses further confirmed that the C/S-transition represented change from Carex to Sphagnum peat (not shown).

The BD analyses of surface peat showed typical profiles from c. 0.02–0.03 g cm⁻³ at the top layer to 0.03–0.05 g cm⁻³ at 15–20 cm depth for decomposed Sphagnum peat. In the underlying Carex peat strata, a zone of high BD (>0.1 g cm⁻³) was met at ca. 35–55 cm depth, under which the peat quality was rather homogenous (Fig. 5). The high BD layers showed the highest concentrations of pollen of P. sylvestris. The profiles of ash % showed rather homogenous over-all patterns of low ash contents (1.5–2.5%) except at the C/S transition, where marked peaks of ash contents were met (Fig. 5). The mass ratio Ca/Mg showed clear decrease to the level of ombrotrophic S. fuscum across the C/S transition in both profiles (Fig. 5).

Figure 5.  Properties of peat profile 1 (a) and profile 2 (b). On the left, BD, ash content and degree of humification are presented for the whole peat profiles. On the right, concentrations of pine pollen and Sphagnum spores, as well as the mass ratio of calcium and magnesium are shown for the top 50-cm layers. The horizontal reference lines depict the depth of transition from Carex peat to Sphagnum peat. The Ca/Mg reference depicts the ratio of ombrotrophic Sphagnum fuscum in the region.

In the qualitative investigation of macrofossils, abundant remains of cladocerans (Chydorus sp., Cladocera) were met in the Carex-peat strata below the C/S transition (Profile 1: 25–50 cm; Profile 2: 30–50 cm). The samples with peak ash content at the C/S transition were very rich in diatoms, namely Eunotia paludosa, Eunotia parallela, Pinnularia sp. and Frustulia sp.. Very few remains of bryophytes were found in the Carex peat below the C/S transition. After screening tens of replicate slides, Sphagnum subsecundum, S. majus and S. compactum were identified from several samples below the C/S transition, however (not shown).

The total mass (dry weight) of new peat in the Sphagnum surface-peat layer was 7900 ± 612 g m⁻² at the S. papillosum lawns and 5970 ± 410 g m⁻² at the S. majus carpets, including biomass. The difference between the S. papillosum and S. majus sites was nearly significant (t-test, p = 0.080). During the 35-year period after the onset of the vegetation change, the recent apparent rate of carbon accumulation (RARCA) in the new Sphagnum peat was 112.9 g m⁻² year⁻¹ (SE 8.7) at the S. papillosum lawns and 85.3 g m⁻² year⁻¹ (SE 5.9) at the S. majus carpets, on average (assuming 50% carbon content of dry weight). The average RARCA of all sites was 100.6 g m⁻² year⁻¹ (SE 6.3).

Pollen density and pine dating

All examined small pines in S. papillosum carpets were rooted in Sphagnum peat above the C/S transition (Fig. 6). The average age of the pines was 20 years and the oldest pine had established in 1974. The thickness of surface peat above the root collars of the small pines approximately followed the power function:

• 

Figure 6.  The age-depth relationship of surface peat as reconstructed by rooting depth of 37 small pines (open circles). Exponential regression curve and 95% confidence curves are fitted to the pine data. Embedded in the figure are age-depth curves derived from cumulative pollen counts of profiles 1 and 2.

The model was highly significant (F = 42.61, p < 0.001) and explained 55% of variation in thickness of surface peat (R² = 0.549). The model gives estimated depths of 24.8 cm for 40-year-old peat and 23.5 cm for 35-year-old peat at the S. papillosum lawns.

The two pollen-density profiles of P. sylvestris pollen grains showed low density in the Sphagnum layer and remarkably higher densities in the underlying Carex peat (Fig. 5). The average annual deposition of pine pollen was 6022 grains cm⁻² in profile 1 (three pine dates) and 6708 grains cm⁻² in profile 2 (two pine dates). The two S. fuscum samples resulted in 6488 and 7128 grains cm⁻² average deposition over 6 years (2003–2009). The average of the four estimates (6587 grains cm⁻²) was used as the constant value for pollen deposition. The cumulative pollen counts (± cumulative standard error) at the C/S transition were 238 958 grains cm⁻² (± 26231) in Profile 1 and 261 446 grains cm⁻² (± 37063) in Profile 2. The estimated ages of the C/S transition were 36 years (± 4) and 40 years (± 6 years), respectively. The age-depth curves of the cumulative pollen counts were within the 95%-confidence interval of the age-depth regression curve based on the pine dating (Fig. 6).

Discussion

Interpretation of observed changes

The aerial photographs showed that a major ecosystem-scale change took place at the studied mire between 1974 and 1995. The timing of the change was verified by dating of surface peat using root-collar depth of small pines and density profiles of pine pollen. Vegetation shifted rapidly from Carex-fen vegetation to Sphagnum-dominated vegetation. The change took place after the beginning of drainage and peat excavation activities in the catchment and it is very likely that the vegetation change was a response to this hydrological disturbance. This conclusion is supported by the fact that the whole mire area changed simultaneously. An extensive, allogenic hydrological impact, such as a major reduction of minerogenous water input to the mire basin, is required to explain the phenomenon.

The 1970 aerial photograph, taken 2 years after the first drainage in the catchment, possibly represents an exceptionally dry situation. This is indicated by the lack of dark tones typical of open water surfaces. Apparently, no major vegetation change had taken place by this time, as the 1970 histogram still reflected the same shape as the 1941 and 1965 greyscale histograms, being only truncated to a narrower scale. The onset of recruitment of pines in the flarks started in mid-1970s, which may indicate a lowered water table (see Gunnarsson & Rydin 1998), and the growth of old pines also temporarily increased in the early 1970s in the central mire area (T. Tahvanainen, unpublished data). Furthermore, the C/S transition was characterized by increases of Eriophorum vaginatum and Trichophorum cespitosum, possibly indicating lowered or increasingly fluctuating water levels. Elsewhere, E. vaginatum has been found to characterize the transition from fen to bog (Hughes & Barber 2003, 2004). Rapid increase of E. vaginatum is a characteristic feature also in mire restoration areas (Jauhiainen, Laiho & Vasander 2002). The blocking of ditches in restoration areas and the moderate drying of flarks in aapa mires are quite different treatments, but they appear to have similar results, i.e. in conditions favourable to E. vaginatum and Sphagna. The tussocks formed by E. vaginatum provide sheltered microenvironments for the establishment of Sphagnum (Tuittila et al. 2000).

Despite the reduction of the total water budget due to the catchment disturbance, the mire remains very wet. The water level was located well above the C/S transition at all places examined. Furthermore, the water-table depth relative to bryophyte capitula was just a few centimetres in the carpets dominated by S. majus and S. jensenii, which are indicators of extremely wet conditions (Tahvanainen & Tolonen 2004). Prior to the catchment disturbance, the flarks of the studied aapa mire were very wet, open-water pools. This was indicated by the dark tones of flarks in aerial photographs and by the abundance of the aquatic cladocerans in the Carex peat. In such watery environments, survival and growth of Sphagna may be hampered by flooding (Tuittila, Vasander & Laine 2003), although some Sphagnum species do grow well in aquatic environments. In every case, a moderate lowering of water level may have promoted Sphagnum growth when sufficiently moist conditions still prevailed (see also Hughes 2000). It is apparent that the water level did not fall below the surface of the old Carex peat and, hence, the availability of water to plants was not limited by the hydrological change. Without considering effects on water quality, it is difficult to explain why Sphagnum species had not thrived before the change, however.

Following the reduced minerogenous water input, water chemistry apparently changed towards poorer mineral content and lower pH at the studied mire. The observed low pH and conductivity values of surface water were similar to bogs and poor fens elsewhere in eastern Finland (Tahvanainen et al. 2002; Tahvanainen 2004). Evidence of change from minerogenous to ombrogenous hydrology was found from Ca/Mg profiles of peat. The ratio decreased from minerogenous levels (5–10) below the C/S transition to less than 3.5 at the surface, thus corresponding to ombrotrophic S. fuscum in the region. This pattern cannot be expected under unchanged hydrology because, in ombrogenous peat profiles, Ca is typically depleted with depth in relation to Mg, which is more efficiently accumulated in peat (Damman 1978; Damman, Tolonen & Sallantaus 1992). The present ombrogenous water chemistry was evidently preceded by more minerogenous conditions, which could explain the absence of Sphagnum cover prior to the hydrological disturbance.

The profiles of ash content showed a rather invariable overall pattern of relatively low ash content through the peat profiles, except for the marked increases at the C/S transition. Kokfelt, Struyf & Randsalu (2009) observed similar elevated ash content connected with peaks of biogenic silica and diatom abundance at transitional phases of a peat profile from southern Sweden. They suggested that abundance of diatoms may indicate major transitions in mire development. In line with their suggestion, the peak ash contents at the C/S transition were paralleled by abundant diatoms.

Impact on carbon balance

The rapid development of Sphagnum-dominated vegetation has the potential to increase carbon sequestration. Within the decadal time scale observed here, the new surface peat had accumulated approximately 7 kg m⁻² dry weight. The RARCA was 101 g m⁻² year⁻¹ as calculated for the 35-year progressive period after the vegetation change and 50% carbon content. This RARCA estimate is comparable to the average of 126 g m⁻² year⁻¹ (37 years, 50% carbon content) of pristine bogs in Finland (Tolonen & Turunen 1996). In pristine fens, the respective RARCA was estimated to average 77 g m⁻² year⁻¹ (Tolonen & Turunen 1996). Hence, about 60% increase of carbon sequestration in the surface peat could be expected to follow from ombrotrophication, as according to the average RARCA values. In the current case, the increase was probably much higher. The aapa mire was characterized by open water and ‘mud-bottom’ surfaces and carbon accumulation was probably very low prior to ombrotrophication, which seems to be the typical case in patterned aapa mires (Mäkilä, Saarnisto & Kankainen 2001; Turunen et al. 2002b).

It must be remembered, however, that although the RARCA represents an important part of the carbon cycle, it is not a direct measure of the total carbon balance. Changes in the loss of carbon from deeper peat strata could not be estimated here. It is possible that decomposition of the old fen peat becomes constrained by acidification and reduced nutrient availability after ombrotrophication. Also the suppression of fen vegetation by Sphagnum, hence of the input of new carbon substrates and oxygen via roots, may inhibit decomposition of old fen peat. Indeed, it is possible that ombrotrophication conserves the underlying fen peat to some extent. Simultaneously, methane production can be assumed to fall after the development of a thick, aerobic Sphagnum layer (Bubier 1995). The increase of carbon sequestration after ombrotrophication occurs subject to the condition that water-table level does not fall too low within the fen peat. In such cases, carbon loss due to increase of aerobic layer and decomposition would be expected (Moore 2002) instead of the response observed here. Clearly, more detailed studies are needed to assess the effects of ombrotrophication on the carbon balance.

Sensitivity of aapa mires to hydrological changes

It is known from several palaeoecological studies that ombrotrophication can be a relatively rapid phenomenon (Gorham & Janssens 1992; Kuhry et al. 1993). According to Kuhry et al. (1993) the time lapses of the transition from rich fen to poor fen and bog were only 50–350 years, as estimated by linear interpolation between available dates of five peat profiles in boreal, continental Canada. Similarly rapid transitions were found by Hughes & Barber (2003) from bogs in temperate, oceanic climate in Wales. In this study, it is demonstrated in a boreal aapa mire that the major vegetation change of the ombrotrophication process can, indeed, take place within a few decades.

Abrupt ombrotrophication is connected to the rapid lowering of pH (Gorham & Janssens 1992; Kuhry et al. 1993). In mire waters, extremely low pH (around 4) is principally caused by organic acidity, while high pH (above 6) is buffered by mineral alkalinity. Intermediate pH range (5–6) corresponds to a narrow range of alkalinity (c. 0–100 meq L⁻¹), making it very susceptible to changes of the ionic balance (Gorham, Bailey & Schindler 1984). Nevertheless, aapa mires in Fennoscandia are typically ‘poor fens’ characterized by low alkalinity and intermediate pH, while ‘rich fens’ with high pH are relatively rare (Eurola et al. 1991; Sjörs & Gunnarsson 2002). The intermediate pH level can be maintained despite its chemical instability if the minerogenous hydrology is stable enough to prevent the dominance of organic acids over the pH reaction (Tahvanainen et al. 2002). However, the chemical instability of intermediate pH values does make aapa mires particularly sensitive to hydrological changes. The fact that pH consistently correlates with the main gradient of mire vegetation (Sjörs 1952; Tahvanainen et al. 2002; Tahvanainen 2004) implies that acidification is a potential key factor to mediate vegetation change during ombrotrophication.

Another element adding sensitivity towards ombrotrophication is the common, although subordinate, occurrence of Sphagnum species with wide ecological amplitude across the bog-fen gradient (e.g. S. angustifolium, S. compactum, S. jensenii, S. papillosum, S. magellanicum, S. majus) in vegetation of aapa mires (Ruuhijärvi 1960). Following changed conditions, these species may readily proliferate without delays due to dispersal limitation. In the current case, such generalist Sphagnum species formed the dominant vegetation, while Sphagnum balticum, which is typically the dominant species of lawns of pristine bogs in the region (Eurola 1962; Tolonen 1967), was markedly scarce. It is likely that S. balticum has a relatively narrow tolerance towards minerotrophy and it probably was not present in sufficient frequency to immediately benefit from the changed conditions in the studied mire. Among the bryophytes, only scattered indications of weak minerotrophy (Sphagnum fallax, Sphagnum pulchrum) were found at the mire margins and the sides of some hummock strings. High abundance of S. papillosum is usually only found in minerotrophic fens, but the species is not a strict indicator, as it frequently occurs also in ombrotrophic bogs in eastern Finland (personal observations; see also Tolonen 1967). Sphagnum subsecundum, a clearly minerotrophic species indicative of intermediate pH (Gorham & Janssens 1992), was found in some peat samples below the C/S transition, while it appears to be completely absent from the flora today. Among vascular plants, several deep-rooted fen species (C. lasiocarpa, C. rostrata, E. angustifolium, M. trifoliata, Molinia caerulea) were met in relatively high frequency. Such species may persist for extended periods after ombrotrophication as relicts from the preceding minerotrophic phase (Gorham 1950, 1957). Thus, the dominant ‘bog vegetation’ today is not representative of raised bogs in the region, but rather a collection of (dominant) wide-amplitude bog species and (subordinate) remnants of deep-rooted fen plants.

A model case of climate impact on aapa mires?

Previously, drainage in situ has been used in many studies as a treatment to model climate change impact on bogs (e.g. Laine et al. 1996; Whittington & Price 2006). However, while discharge from raised bogs is typically close to zero in the summer (Kellner & Halldin 2002), it is generally richer in aapa mires (Sallantaus 2006). Therefore, water level is not so readily affected in aapa mires. Before draw-down of water level (i.e. decrease of water storage) can be anticipated, the rate of water flow (volumes of minerogenous input and discharge) must fall in aapa mires. This effect can be imitated by the artificial removal of the minerogenous influence.

The possibility of the development of fens into bogs in response to climate change has not been subjected to experimental investigation. According to Aurela, Laurila & Tuovinen (2004), the meteorological conditions in spring (temperatures and the timing of snow melt) best explained variation in the annual carbon sink of a subarctic aapa mire. They suggested that lengthening of the growing season would increase the carbon sink. In the long term, increased peat accumulation could lead to autogenic development of aapa mires into bogs. Autogenic development has been considered to be the most typical trait of the development of Sphagnum-dominated bogs (e.g. Rydin & Jeglum 2006), but it is deemed very slow as it is limited by the vertical growth rate of peat. However, bog development can also be triggered by allogenic factors such as reduced minerogenous input due to climate warming or catchment disturbance (Van Diggelen et al. 1991; Almquist-Jacobson & Foster 1995; Hughes 2000; Hughes & Barber 2003). Until this study, this possibility has not been recognized for aapa mires.

Pristine aapa mires do not appear to have experienced notable changes during recent decades. Mäkilä, Saarnisto & Kankainen (2001) found very low long-term carbon accumulation rates (8.0 g C m⁻²) in an aapa mire in northern Finland and they suggested that peat accumulation was ceasing. In the Patvinsuo mire in eastern Finland, Turunen et al. (2002b) found even lower long-term accumulation rates (5.6 g C m⁻²) and apparently arrested development of aapa mires. Dissanska, Bernier & Payette (2009) observed slightly increased areas of open water in a patterned aapa mire, instead of any signs of Sphagnum invasion in Quebec, Canada. Aapa mires have been in an apparently steady equilibrium with the cool and moist climate conditions of the suboceanic boreal regions. However, the current results show that abrupt vegetation change may readily follow in aapa mires in response to strong enough changes of hydrology.

This study shows that hydrological disturbance in the catchment can affect the ecosystem balance of aapa mires even when the mire itself was not drained. This type of disturbance may provide a model for climate-change impact on aapa mires, but more detailed hydrological modeling is required to assess the potential hydrological changes predicted by climate models. In every case, it should be recognized that ombrotrophication is not degradation. Ombrotrophication is a progressive response that leads to increased peat accumulation, and thus is a desirable climatic feedback. In the same time, however, a severe threat is generated to the diversity of fen biota. The potential impacts of catchment disturbances and of climate change to the hydrology of aapa mires calls for attention in conservation planning. Ombrotrophication creates strong carbon sinks, while the remaining aapa mires will be increasingly important as habitats for threatened fen species.

Acknowledgements

Mr Markku Heiskanen and Professor Kimmo Tolonen provided information about the natural state of the study area in 1960s. Mr Iivari Laakkonen and Mr Tuomo Takala are thanked for their assistance in field work. Mrs Anne Ryynänen and Miss Marika Lax are acknowledged for laboratory assistance. Dr Heikki Simola is thanked for the identification of diatoms. The two anonymous referees are thanked for valuable comments. The study was started with a preliminary project funded by the Finnish Cultural Foundation.