Urban Gunnarsson, Department of Plant Ecology, Evolutionary Biology Centre, Uppsala University, Villavägen 14, SE-752 36 Uppsala, Sweden (tel. + 46 18 4712850; fax: + 46 18 553419; e-mail Urban.Gunnarsson@ebc.uu.se).
1We present data on calcium concentrations and pH in mire waters collected from different mire types in central and northern Sweden, compiled from published literature or calculated from field determinations of electrical conductivity and pH.
2Measurements of electrical conductivity (after subtracting that of H+ ions) were used to calculate the most probable Ca concentrations, but only when pH was 4.7 or higher. At lower pH the possible errors become too large.
3The data support a continuous gradient in water chemistry, but with considerable overlap, from mineral-poor ombrotrophic bogs and minerotrophic extremely poor fens, to moderately poor fens, intermediate fens, moderately rich fens and finally to extremely rich fens, rather than a discrete division. However, for hydrological reasons, we wish to retain the separation of ombrotrophy (bog) from minerotrophy (fen).
4The frequency histogram of pH values from central and northern Sweden shows a bimodal tendency, although a considerable number of mires still occur close to the minimum (pH 5.3). Over-sampling in species-rich areas could have contributed to the depth of this apparent minimum, but not to its position along the pH scale.
5Changes in calcium concentrations in mire waters during the last 50 years are discussed and related to changes in pH and conductivity.
The first published pH determinations in Swedish mire waters were made by Du Rietz (1932) and the first cation analyses by Witting (1947, 1948, 1949). Although some of the numerous data on pH and electrical conductivity were summarized in Sjörs (1952), most remain unpublished or appear only in reports to Swedish authorities.
Recently Wheeler & Proctor (2000), Vitt (2000) and Økland et al. (2001) discussed the relationship between water-chemistry and peatland vegetation. Water chemistry has been more frequently studied than peat chemistry, although it may be less relevant for the composition of the vegetation (compare Sjörs 1961; Malmer 1962; Persson 1962). There are abundant data on pH and electrical conductivity from central and northern Sweden, but fewer analyses of ionic conditions. We hope that the large data set presented here may shed light on the nature of the correlation between water chemistry and vegetation.
Gorham et al. (1987), Wheeler & Proctor (2000) and Vitt (2000) discuss a possible bimodality in water acidity and a large overlap in both pH and Ca in relation to vegetation units. One aim of this paper is to find out to what extent these gradients are continuous.
As the bulk of our material is quite old, we also include modern data to show the considerable changes that have occurred during the last 50 years in relation to anthropogenic deposition. Few Swedish studies have focused on the chemical changes in mire waters, but most of these show lowered pH in rich and intermediate fens (Hedenäs & Kooijman 1996; Thygesen 1997; Gunnarsson 2000; Gunnarsson et al. 2000; Gunnarsson et al. 2002). We further discuss how the changes in pH may be related to increased Ca concentrations in mire waters.
Materials and methods
We have compiled water-chemistry data (mostly from unpublished sources) gathered in several parts of north and central Sweden (Fig. 1) between 1942 and 1971.
Water was collected at the surface of shallow mire pools (hollows, flarks, etc.) and notes were made of the type of adjacent vegetation (see below), classified as in Sjörs (1952) with ‘intermediate fens’ between ‘moderately (transitional) poor fen’ and ‘moderately (transitional) rich fen’ in the Du Rietz (1942, 1949, 1954) terminology.
Up to 1952 pH determinations were made colourimetrically, with a correction of +0.2 pH units for ‘salt error’, but subsequently a glass electrode was used and no correction was needed. Conductivity was measured in small cells, in some cases with the Pt electrodes platinized (no difference), and standardized to 20 °C. Care was taken not to allow mixing with interstitial or deeper water.
When both variables were log transformed Ca was linearly related to conductivity in samples where both were determined (Fig. 2). This regression was used to convert conductivity to Ca concentrations when only conductivity values were available. The uncertainty resulting from this conversion, although considerable, is very much smaller than the overall variation in Ca concentration (Fig. 2). Vitt (2000) presents a similar diagram from central Canada.
Whilst valid for central and northern Sweden, the regression cannot be used for the more maritime south, where considerably more Na, Mg and Cl contribute to the conductivity (Malmer 1962). Most of our area is only weakly exposed to maritime influences, but for some waters in westernmost Jämtland (VJ, Fig. 1) a greater sea salt effect may have resulted in slightly increased estimated values for Ca, although levels in most samples remain very low.
When pH values are low, residual conductivity is customarily determined by subtracting the conductivity due to H+ ions. The subtracted term is excessively large in very acid waters and residual values are therefore strongly dependent on the accuracy of the pH determinations. For instance, at pH 4.7 an error of 0.1 pH-units would on average correspond to a difference of about 0.17 mg L−1 in Ca concentration. Therefore, samples with pH below 4.7 were used when comparing Ca with pH only if it was analytically determined (Fig. 3a), rather than estimated from conductivity values (Fig. 3b).
The similarity of Fig. 3(a) and 3(b) shows that estimation of Ca from conductivity was quite reliable. The most obvious discrepancy is the higher frequency of low Ca values in the pH 5–7 region in Fig. 3(b), due to the more diverse geographical origin of these samples. Low Ca is often associated with a pH well above 5 in the siliceous areas that prevail in most of the higher forested parts of Fennoscandia. Nevertheless, calcareous areas are over-represented in our material, notably in central Jämtland and certain parts of the mountains (Du Rietz 1949; Witting 1949; Persson 1961, 1962).
There are diurnal and seasonal variations in water pH, and also a vertical gradient, all caused mainly by differences in the concentration of dissolved CO2 (Teemu Tahvanainen, personal communication). These variations are largest at intermediate and higher pH. Loss of CO2 caused by warming and photosynthesis in mosses and algae causes pH to rise by day and fall at night. We used data from surface water, which tends to have a higher pH than that in moss capitula or in the peat below the water table. It is, however, difficult completely to avoid contamination by such water, and this may occasionally have affected pH measurements and cause increased scatter within a vegetational group (compare Sjörs 1952).
Results and discussion
DISTRIBUTION OF WATER PH AND CA WITHIN VEGETATIONAL GROUPS
Ombrotrophic bogs in central and northern Sweden occur most frequently between pH 3.7 and 4.2, although some have higher pH values (Fig. 4). Some of these samples originate from small shallow bogs of northern type (see Rydin et al. 1999). In western Jämtland, as mentioned, the maritime influence could give rise to higher Mg concentrations and thus increased pH even in ombrotrophic sites. In a few bogs there, wind-transported fine sand from eroded lakeshores was observed, and this may have increased the pH. The majority of Swedish ombrotrophic bogs are distinctly more acid than most British ones and have very low Ca concentrations, mostly well below 1 mg L−1 (Witting 1948; our Fig. 3a).
The so-called mineral soil water (indicator) limit (Thunmark 1942; Du Rietz 1954; Witting 1947, 1948; see also Gorham 1952) is not distinct in our data, although sites close to this boundary are well represented. During high-water conditions, minerotrophic water may have a temporary influence over areas which normally have only ombrotrophic water (e.g. Malmer 1962). There is considerable overlap in water pH between extremely poor fens and ombrotrophic bogs (Fig. 4). Although the large, fairly dry, mostly wooded marginal mire parts may not be as ombrotrophic as they look, they rarely yield free water for sampling and are thus absent from our material.
Moderately (transitional) poor fens (botanically a rather heterogeneous group) show neither distinct pH optimum (Fig. 4) nor sharp boundaries with other vegetation units.
Intermediate fens are almost exclusively found in the boreal region (from north of Lake Vänern up to the treeline, and even higher). Poor fen species tend to be dominant or prominent (e.g. Sphagnum papil1osum Lindb.), but are mixed with scattered examples of rich fen species, most frequently bryophytes. The water is nearly always non-calcareous, but not very acid pH (mostly 5.5–6.4, Fig. 4). Most intermediate fens are sloping and water seeps downhill, implying a relatively rapid change of the water. This may cause fairly high metal saturation of the peat, despite the low concentrations in the water.
Moderately (transitional) rich fens show much overlap with intermediate fens, with optimum pH only slightly higher (5.8–7.1, Fig. 4). The Ca concentration is very variable, often as low as about 2 mg L−1 but sometimes much higher (Fig. 3). Vascular rich fen species are mostly not very numerous, but the typical rich fen bryophytes play a great role. Some sites are particularly rich in iron (e.g. Sjörs & Een 2000).
Extremely rich fens have moderately to highly calcareous water, though usually somewhat less than some British or continental sites, and have high or very high pH values. They are almost entirely restricted to calcareous areas.
THE RELATION BETWEEN WATER AND PEAT CHEMISTRY
The surface waters of mires originate from precipitation (including meltwater) and, in minerotrophic sites, from water that has been in contact with mineral soil, on or near the surface or percolating at some depth. In some mires the water has also passed lakes or rivers. Nevertheless, the mire surface water rarely comes exclusively directly from these primary sources. It has normally been in contact with living vegetation, notably bryophytes and microorganisms, and with litter and non-living peat (Malmer 1993; Malmer et al. 1994). Between the peat and the mire water there is an intrinsic, near-equilibrium. The ions (mainly cations) adsorbed on the peat substances are quantitatively dominant, the free ions dissolved in the surface water occurring in much smaller amounts.
The peat substances tend to accumulate bivalent ions (Ca2+ and Mg2+) and release H+, whereas living cells tend to take up K+ and NH4+ selectively, but less Ca2+ than the dead organic matter. The hydrogen ions in both peat and water are produced in at least two ways, by exchange from living plants, notably bryophytes, and from incomplete oxidation of litter and peat, largely in the acrotelm.
The great overlap between both vegetational categories and water-chemical values and their poor covariation suggest that vascular mire plants are more dependent on the chemical conditions in the peat substrate than in the water. This may be less evident for bryophytes (e.g. Vitt & Chee 1990), which grow on the surface and mostly have a direct contact between green cells and water.
The quantitative dominance of peat substances in relation to the small amounts of dissolved matter in the water leads to near stability of hydrochemical conditions. Thus dilution by rain or melt-water does not have a very long-lasting effect and pH and cation concentrations change less with weather and season than might be expected, except for the strong but temporary dilution at snowmelt. On the other hand, the long drought periods, which occur in certain summers, will cause the surface water to dry out, often completely. The ionic balance of the remaining free water, if any, may then be greatly disturbed. The shrinkage caused by the drought may even allow air to penetrate deep into cracks in what is normally regarded as the catotelm.
In his excellent paper on mires in interior Canada, Vitt (2000) discusses the origin of peatland acidity. He emphasizes the interaction between water and living Sphagnum, to some extent disregarding processes occurring in the peat, which may have an important role (cf. Clymo 1984). For example, on permafrost palsas, in northern mires, the acidity is even totally independent of Sphagnum, which is practically absent. The originally minerotrophic palsa surface peat becomes practically ombrotrophic through leaching and oxidation.
THE ROLE OF MAGNESIUM
Magnesium has an effect due to its higher proportion in the precipitation in areas close to maritime coasts reflected in the increasing Ca/Mg-quotient of ombrotrophic waters further inland. As Mg is bivalent, it will become strongly adsorbed to the peat and thereby cause the higher pH seen in maritime areas, even in ombrotrophic peat and water. This may very likely be a main cause of the lower acidity of many bogs in Britain and especially coastal Ireland (unpublished determinations by E. Gorham in 1956) and also in coastal Pacific North America (pH 4.2–5.1 in SE Alaska; Sjörs 1984), and would partly explain the much richer flora of ombrotrophic bogs there (Neiland 1971; Sjörs 1985; Vitt et al. 1990; Malmer et al. 1992), although the high Pacific floristic biodiversity and the milder climate may also contribute.
Gorham et al. (1984, 1987), Wheeler & Proctor (2000) and Vitt (2000) point out the rarity of observations of intermediate values of water pH. They argue that this is due to a shift between buffer systems at about pH 5.5, with pH regulation being taken over above this value by the hydrocarbonate buffer. However, the actual acidity of mire water is strongly regulated by the degree of neutralization (base saturation) of the acid substances in the peat. This is different from conditions in lake water of low humus content, where hydrocarbonate ions usually predominate in the buffer system.
Very low pH values abound in our data (Fig. 4), with a maximum frequency at pH 3.8, due to many studies in or near ombrotrophic sites and in poor fens. Intermediate values for pH are, however, also fairly common, especially at a pH about 6 and often combined with quite low Ca. There is a continuous variation in occurrence, albeit with a clear minimum around pH 5.3, i.e. a little lower than in Vitt (2000) and Wheeler & Proctor (2000), but almost exactly as in Gorham et al. (1984, 1987).
The pH frequency is, however, too much dependent on site selection to have real statistical significance. Although very poor and acid mires prevail in most of central and northern Sweden, botanists may have a selective bias for species-rich fens with a high biodiversity, making the latter over-represented in relation to their area, and the ‘less interesting’ less rich fens under-represented. The selection of sample sites can have an influence on the magnitude and sharpness of the minimum, but hardly on its position along the pH scale.
In calcareous landscapes, the presence of acid peat often results from accumulation, separating this peat from subjacent or surrounding calcareous waters (Sjörs 1963; Gorham et al. 1984; Vitt 2000). The argument that the comparatively short time needed for this process will lead to a relative shortage of sites with intermediate pH values is less valid in boreal Sweden, where most peatlands have developed more gradually by paludification on more or less sloping siliceous terrain, with still prevailing soligenous seepage. However, in a study of the change in pH over a 50-year period on Skattlösbergs Stormosse (Gunnarsson et al. 2000), sites with pH of 5.3–5.8 showed the largest degree of change, supporting the idea that mires might remain in this interval only for a relatively short time, also because of their low buffering capacity (Gorham et al. 1984, 1987; Vitt 2000). Despite the frequency minimum at about pH 5.3, the vegetational gradient seems continuous in our area.
THE DISTINCTION OF OMBROTROPHY
We agree with Økland et al. (2001) that ombrotrophic bog sites should be recognized separately in the field despite the frequent occurrence of transitions to weakly minerotrophic conditions, and despite the impossibility of defining a mineral soil water limit everywhere on the same chemical values or plant indicators. We think that, except for highly oceanic areas, the distinction of ombrotrophic bogs is the least ambiguous in the mire classification, but most clearly defined from hydrological rather than hydrochemical or botanical criteria. Besides, ombrotrophic bogs cover vast areas in Fennoscandia and elsewhere, and have distinct botanical and ecological conditions, which within each separate region are similar everywhere, to a unique degree among plant communities (Rydin et al. 1999). Therefore we do not wish to include ‘poor fen’ among ‘bogs’, as advocated by Wheeler & Proctor (2000).
CHANGES IN MIRE WATER CHEMISTRY
The reported historic determinations were made during a 30-year period, which ended 30 years ago. Although air and precipitation acidification in Sweden increased from low to appreciable during the determination period, larger changes have probably occurred subsequently, with the sulphur component recently starting to decrease (but the nitrogen deposition remaining high). There were some botanical indications of change as early as about 1947 (Thygesen 1997). The old material has now a documentary value and can be used as a reference. It should, however, be mentioned that the influx of sulphate and nitrate ions has increased only slightly in the north Swedish areas.
Under conditions of increased sulphur and nitrogen deposition, pH might decrease, as seen for instance in Hedenäs & Kooijman (1996), Thygesen (1997) and Gunnarsson (2000). Increased H+ concentrations might release other cations attached to the peat substances and therefore produce increased Ca concentrations in some mire waters. For minerotrophic sites, increased weathering of mineral soils may also contribute more metal cations. In a study on Skattlösbergs Stormosse we actually found increased Ca concentrations in 1995, for all water samples, compared with water samples taken from exactly the same localities in 1945 (Fig. 5). Although the samples in 1995 were taken under rather low groundwater conditions and on only one occasion, we think that this sample was fairly representative. At the same time the conductivity of most samples had increased, which also has been seen in surface mire waters from Åkhultmyren, southern Sweden (Gunnarsson et al. 2002).
The water chemistry of the lagg fen at Ryggmossen mostly showed increased Ca concentrations and decreased pH, between 1947 to 1951 and 1993 (Thygesen 1997). However, there were also some cases with reduced Ca concentration, probably because water samples from hydrologically different mire types will show different patterns. Very acid poor fens and bogs seldom show any reduction in pH.
The increased deposition of nitrogen and sulphur on mires will have a multiple effect on growing plants, and increase both the Ca- and N-availability. The increased availability of N and CO2 may cause increased growth of vascular plants but decrease that of Sphagnum (Gunnarsson & Rydin 2000; Berendse et al. 2001). The changes have mainly had a negative effect on the species diversity at the landscape level, many fen vascular plants and rich fen bryophytes having disappeared or decreased in occurrence during the last 50 years (Hedenäs & Kooijman 1996; Gunnarsson et al. 2000; Gunnarsson et al. 2002). On the other hand, species that are not exclusively mire plants have increased on several mires (Gunnarsson & Rydin 1998; Frankl & Schmeidl 2000; Gunnarsson et al. 2002). The mires in southern Sweden will probably show large and rapid changes during the next 50-year period (less so in central and north Sweden), which shows the need for monitoring these ecosystems.
We thank Ingvar Backéus, Håkan Rydin, Jon ågren and three referees for valuable comments on earlier versions of the manuscript, and Teemu Tahviainen for remarks on variation in pH. Hugo Sjörs also thanks more than 20 former companions in the field.