Wetlands in flux: looking for the drivers in a central European case

Study area

The Biebrza wetlands lie in the Biebrza River valley in north-eastern Poland (Fig. 1) in the Podlaska Lowland. They comprise undrained floodplains, marshes and fens surrounded by a post-glacial landscape with ice-pushed hills, moraines and outwash plains. The near-natural character of the wetlands is reflected in a regular pattern of plant communities which run the length and breadth of the river valley. Although a number of its tributaries have been trained and turned into drainage canals, the river is unregulated, meandering throughout the length of the valley and with many oxbow lakes. In spring, extensive areas are flooded by the river, snowmelt, rain and groundwater (Chormański et al. 2011; Keizer et al. 2014).

Open image in new window

The riparian vegetation consists of reed-beds and the floodplain contains highly productive rich fen (Wassen et al. 2006). These are relatively species-poor tall sedge, grass and herb vegetations. Further from the river are abundant low-productivity sedge fen types containing many Red List species. In an intermediate belt, beyond the seasonal river flood zone and where the calcareous groundwater from the moraines does not reach the fen surface, poor fen and transitional fen types are present, fed mainly by rainwater. The importance of conserving the immense ecological value of the Biebrza wetlands has been recognized internationally and nationally. The Biebrza River valley is designated as a wetland site of global significance in the Ramsar Convention, it belongs to the Natura 2000 EU Ecological Network (PLH200008) and has the status of national park (Biebrza Nationa Park), covering an area of 59,800 ha.

Hydrological and climatic data

The hydrological and meteorological data were obtained from the Institute of Meteorology and Water Management (IMGW). The dataset covered the period 1962–2012 for daily records of precipitation, minimum and maximum air temperature, snow depth, and for the period 1951–2012 for river discharge. The reference evapotranspiration was calculated with the Penman–Monteith equation (Allen et al. 1998), with data registered in one station for the period 1960–2009 for a daily resolution for the growing season i.e. 1 April to 30 September. The values of reference evapotranspiration reflect evapotranspiration from a reference surface with no water stress, thus are related to atmospheric conditions and do not reflect the actual evapotranspiration in the study area, which depends on soil moisture and vegetation conditions too. Figure 1 shows the locations of meteorological stations and water gauges.

The discharge time series were obtained from two water gauges: Burzyn and Sztabin. The Burzyn gauge is located near the catchment outlet (the confluence of the Biebrza and Narew) thus the discharge here is driven by factors characteristic for the whole catchment (7000 km²), whereas the Sztabin gauge drains the upstream part of the Biebrza catchment (700 km²), which is the most extensively managed region of the study area. The catchments of the two water gauges contain similar proportions of peat soil: 16% (Burzyn) and 17% (Sztabin).

Soil-moisture complexes maps

Soil-moisture complexes group soils with similar air–water conditions in the vadose zone of the soil profile (Okruszko 1998). They are named after the water regime (Okruszko and Ilnicki 2003): wet (A), periodically wet (AB), moist (B), periodically moist/periodically drying out (BC), drying out (C), periodically dry (CD) and dry (D). Soil-moisture complexes are generally used for selecting optimal management practices on grasslands and are thus good indicators of water conditions in soil.

In this study two maps of soil-moisture complexes were analysed (Fig. 2): a historical map, based on a survey in 1970 (Liwski et al. 1984) and a more recent map based on a survey in 1999 (Szunieiwcz et al. 2004). Both maps were registered in the same geographical coordinate system, and digitized, yielding a new map with polygons showing the intersection of the source maps. The attributes of this map allowed the magnitude and direction of the changes to be investigated. The maps overlapped for 54,035 ha: approximately 75% of the Biebrza River valley area.

Open image in new window

Vegetation maps

To discover vegetation changes in the Biebrza valley, two vegetation maps were compared (Fig. 3): a map published by Palczynski (1984) based on data collected in the 1970s and early 1980s and a vegetation map made by Matuszkiewicz et al. (2000) based on a survey in 1997–1999. Changes were detected similarly to the analysis of soil-moisture complex maps. Since we were interested in the river valley only, we excluded from the analysis dry sandy dunes, high moraines and a large forested bog (Czerwono Bagno) on the northern edge of the valley in the middle basin. The maps overlapped for 45,040 ha, i.e. approximately 62% of the Biebrza River valley area. For the generalization of vegetation types of both maps into common vocabulary, see Online Appendices S1, S2.

Open image in new window

Land use and water management

We compiled a database of historical information from published and unpublished sources, oral stories from locals and archive data of land ownership, agricultural practices and management measures (Wassen and Okruszko 1994; Wassen and Mioduszewski 1999). From this information we derived spatially explicit information about changes in land use and water management practices in the Biebrza catchment between 1960 and 2000. Agricultural practices in the catchment surrounding the valley were increasing in intensity and fertilizer gifts after Poland joined the EU in 2004 (GUS 2015). Since this was after the period for which we could compare soil moisture and vegetation data (up to 2000) we focused only on water and land use changes in the valley up to the year 2000. We constructed a map depicting water management measures, land management by farmers and nature management in the national park (Fig. 4).

Open image in new window

Statistical modelling

The temporal changes in our dataset were analysed using statistical tests. First, linear monotonic trends in the time series were highlighted with the Mann–Kendall test (Mann 1945; Kendall 1975). The sign of the S statistics obtained from this test indicates whether the trend is positive or negative. Next, we calculated the slope (a) of the linear regression line which was fit to the time series. The slope gives information analogous to the S statistics but is more straightforward and frequently used. For both S and a we calculated the p values to analyse if the statistics are significant. When mentioning in the Results section that a statistic result was significant it means that the p value < 0.05.

Hydrological and climatic data

Major changes during the last 50 years are clearly visible in air temperature, reference evapotranspiration and river discharge characteristics; precipitation and snow depth remain stable.

The minimum temperatures show significant increasing trends for January, April, July and August (Fig. S1). The trend in maximum temperatures is similar, but also with a significant positive trend in March (Fig. S2). The greatest rises in temperature are in January and April (0.081 °C year⁻¹) for the maximum temperatures and in January (0.096 °C year⁻¹) for the minimum temperatures.

The only months showing significant trends in precipitation are January and February (increasing trend) and November (decreasing trend) (Fig. S3). The biggest increase in monthly precipitation is in January (0.339 mm year⁻¹); the biggest decrease in monthly precipitation is in November (− 0.379 mm year⁻¹).

Monthly reference evapotranspiration figures show significant increasing trends in April, May and August (Fig. S4). During the middle and end of the growing season (June, July and September) no significant trends occur. The reference evapotranspiration increases to 0.339 mm year⁻¹ for May.

The monthly mean snow depth does not show any significant trend (Fig. S5). However, the time series shows a general tendency of decreasing snow depths at the end of the winter (February and March).

The mean yearly river discharge for the Burzyn gauge increases significantly by 0.202 m³s⁻¹ year⁻¹, mostly in the winter half-year (Fig. S6). By contrast, the Sztabin gauge has a stable mean discharge for all seasons, with insignificant trends (Fig. S6). The minimum annual discharge at both locations increases during the winter half-year and for the Burzyn cross-section for the whole year too. Maximum discharges decrease, but the trends are significant only for Sztabin for the yearly period and the winter half-year. The summer half-year characteristics for both cross-sections do not show significant maximum discharge trend statistics.

Soil-moisture complexes

Between 1970 and 2000 although most (25,345 ha) soil-moisture complexes remained stable, in 17,221 ha the soil-moisture complexes shifted towards higher moisture, and in 11,471 ha the soil moisture decreased (Table 1). The biggest change towards moister complexes was from drying out (C) to periodically moist/periodically drying out (BC): it occurred in 7201 ha (13.3% of overlap). The biggest change towards drier conditions was from drying out (C) to periodically dry (CD) and comprised 6997 ha (12.9%). Nevertheless, the most stable complex was C: its area remained unchanged at 17,145 ha (31.7% of the overlap of the maps). Zooming in on the spatial patterns of soil moisture changes (Fig. 5) it becomes clear that especially the area close to the source of the Biebrza River, became wetter between 1976 and 2000, as also generally occurred in the middle basin, where large areas of drying areas changed into wetter types. In the lower basin the picture is more diverse. In 1976 most of the area contained drying soils. In 2000 the belt along the river changed from drying out (C) into even drier types classified as CD and D. Further from the river the scattered picture of 2000 shows that intermediate dry soils became wetter whereas elsewhere a shift towards dryer types occurred (cfr. Figs. 2, 5). Taking into account that the soil complexes A and AB are the only ones which led to the peat forming process, the scale of changes show more halting the decay of peatlands then restitution of mires.

Table 1

Changes in soil-moisture complexes between 1970 and 2000

From ► To	Area (ha)	Trend	Relative change (%)
C ► C	17,145	.	31.7
C ► BC	7201	+	13.3
C ► CD	6997	−	12.9
B ► B	4207	.	7.8
C ► B	3358	+	6.2
CD ► CD	2269	.	4.2
BC ► B	2180	+	4.0
CD ► C	2087	+	3.9
B ► C	1995	−	3.7
BC ► C	1049	−	1.9
BC ► BC	1041	.	1.9
B ► A	751	+	1.4
B ► AB	625	+	1.2
A ► A	476	.	0.9
CD ► BC	396	+	0.7
B ► BC	377	−	0.7
BC ► CD	332	−	0.6
B ► CD	326	−	0.6
AB ► A	231	+	0.4
BC ► AB	225	+	0.4
AB ► B	220	−	0.4
AB ► AB	207	.	0.4
A ► B	123	−	0.2
C ► AB	84	+	0.2
CD ► B	69	+	0.1
A ► C	21	−	0.0
AB ► C	19	−	0.0
BC ► A	14	+	0.0
A ► CD	6	−	0.0
AB ► CD	5	−	0.0
A ► AB	0	−	0.0

A—wet, AB—periodically wet, B—moist, BC—periodically moist/periodically drying out, drying out—C, CD—periodically dry. Trend symbols: “+” wetter, “−” drier, “.” stable. Relative change shows percentage of changed area in relation to the area of all soil-moisture complexes

Open image in new window

Vegetation

The reed-bed area (Phragmition) increased by approximately 346 ha, to 114% of its size in 1984 (Table 2). In that year the alliance was hardly present in the upper part of the Biebrza basin, but it now forms a continuous belt throughout the valley. This belt was already present in the middle and lower basins, but in the approximately 20-year interval between the maps it widened (Fig. 3).

The area under Molinietalia meadows of the Molinion and Filipendulion alliances increased markedly in the analysed area (Table 2). Molinion increased by 1640 ha (+ 90%) and Filipendulion by 425 ha (+ 177%). The “meadows and pastures not covered by other associations” mostly include land under intensive agriculture; their total area has remained stable (Table 2).

The total area under tall sedge communities (Magnocaricion alliance) shows an overall loss of 225 ha, but it increased in the middle basin (Fig. 3). However, the area under Caricion diandrae (the low-growing sedge-moss communities) has almost doubled since 1984, gaining 1147 ha. Most of this gain was achieved at the expense of Calamagrostion neglectae in the lower basin.

The class of Caricion demissae has increased its cover by 169% of the 1984 area (Table 2). This gain occurred in the lower basin, where Calamagrostion neglectae was replaced; in the middle basin the class lost ground to Caricion diandrae.

Vegetation changes in Biebrza valley show various magnitude and direction (Table S1). Roughly 60–70% of the area of most communities was lost to other communities. The community with the highest relative change is Calamagrostion neglectae, which lost 71% of its 1984 area to other communities. The most stable community is forest, which is currently also the most abundant in the study area. Its area changed by only 13%, but an important share of this (5%) was a change into half-open shrub-forest, which is an arbitrary change that might be attributable to differences in the mapping methodology.

A large part of the area of Calamagrostion neglectae changed into diverse other communities, most of which consist of half-open shrub-forest, Caricion demissae, Caricion diandrae, Molinion, Magnocaricion and forest; each of these changes covers about 3% of the study area (Table S1).

Hydrological and meteorological variables

The minimum and maximum air temperatures show a positive trend in the spring and summer months, which directly determines an increasing reference evapotranspiration. These trends could be expected to influence river discharge in summer, but do not, since on average the river discharge increases from year to year both for Burzyn and Sztabin (Fig. S6). The peak discharges (during the spring flood) are decreasing, which is related to decreasing snow depths (Fig. S5) from November to March. The importance of snow for hydrological processes in the Biebrza catchment has been shown by hydrological modelling (Berezowski et al. 2015). As their study shows, discharge in Burzyn is sensitive to changes in the spatial distribution of snow cover.

The centre of mass of the yearly hydrograph, which indicates the period of the highest discharge (i.e. the spring flood), occurs earlier and earlier each year (Fig. S7). This shift is also probably related to the decreasing snow depths throughout the winter resulting in snow melting progressively earlier at the end of the winter.

Similar findings were presented by Ignar et al. (2011), who developed a hydrodynamic model of river flow for calculating annual floods in the Lower Biebrza River Basin. Their results show that the maximum flood area, flood volume and mean water depth decreased during 1965–2000, but not statistically significantly.

Soil-moisture complexes

An increase in soil moisture occurred predominantly in areas where conservation measures have been implemented (Fig. 5). The first such measure, implemented since 1994, is the prohibition of the removal of aquatic plants from the river. This measure is clearly related to river flow velocity and other hydraulic parameters (Verschoren et al. 2017). Its biggest impact is in the upper basin, where the valley is narrow and dominated by peat soils. The aquatic vegetation reduces the flow velocity and thus raises the water level in the river and the soil moisture content in the surrounding ecosystems. This area (and central part of the Middle Basin) remain the only ones where peat forming process still take place. An important measure was abandonment of drainage ditches in some parts of the river valley. These ditches have silted up and become vegetated, curtailing water discharge. Next measure was the construction of dams in most drainage canals in the lower and middle basins, which delayed the runoff from the valley to the river), thereby preventing the valley from drying out in water-deficit periods. Increase of the soil moisture protects bigger area from peat decay but still does not bring Biebrza Wetlands to their original status. This is a part of the broader problem of future of the wetlands in the boreal region. The described herein management measures were successfully applied also in other wetlands for increasing water table; however, they do not always lead to full restoration of wetlands features such as carbon sequestration, vegetation composition and peat properties (Wilcox et al. 2006; Schimelpfenig et al. 2014).

Vegetation

The comparison of the two maps revealed two types of vegetation change (Fig. 3), both leading to a loss of the open sedge-moss communities.

The first is vegetation change resulting from succession in a similar pattern as reported for other wetlands (Rosenthal 2010). Such changes are expressed in the valley in two ways: (1) invasion by reed-beds and tall sedges (Phragmition alliance) and (2) increase of forested area. Both are believed to result from dereliction of management in the valley (Palczynski 1984). However, the decrease in the area under dense bushes is not what one would expect as the effect of abandonment of management practices. It could, be a result of generalizing the two maps; the area under dense bushes in the Pałczyński map will have been overestimated if the Pałczyński associations that were thought to be bushy were not dense enough to qualify as. And this would be more in line with findings from other research in the area (Palczynski 1984; Schmidt et al. 2000).

In the second type of vegetation changes, management practices seem to have been continued or even intensified. Here the grass/sedge-moss communities of the Calamagrostietum neglectae and Caricion diandrae have been replaced by meadows of the Molinietalia order (mostly in the middle basin) or intensively used meadows (upper basin).

Looking at the role of management in the Biebrza valley, distinctions can be made among the three basins.

In the upper basin, peat analysis has revealed that open vegetation started to develop under the influence of permanent groundwater seepage (Wassen et al. 2002). Later, it seems that although hydrological conditions were favourable for shrubs, trees and reeds, regular mowing kept the fen vegetation open and low. Evidence supporting this hypothesis is the increase of forest, shrubs and Phragmition in the unmanaged areas of the upper basin.

The same can be said for the parts of the middle basin, where forest or Phragmition plant communities replaced the sedge-moss communities. But in other large parts of the middle basin the sedge-moss communities were replaced by Molinietalia meadows. The peat muck soil of these litter meadows is susceptible for shrub and tree encroachment (Palczynski 1984). So the fact that these areas changed from sedge-moss to Molinietalia vegetation and were kept open suggests that they were still being mown regularly.

The lower basin is different. Here the vegetation changes are smallest, though dereliction of management was also signalled here. It was the only basin in which the area under plant communities of the Calamagrostietum neglectae and Caricion diandrae remained stable. The lower basin of the Biebrza floods almost entirely every year, because of a fluvial cone at its southern end, which elongates the hydroperiod by acting as a dam. The hydroperiod and other flood parameters are known to drive the wetland vegetation composition (Murray-Hudson et al. 2015) which might explain the failure of shrubs and trees to colonize large parts of the lower basin. Plant communities of the Phragmition alliance can cope better with these floods than shrubs and trees. Moreover, Phragmition have expanded its cover as a consequence of the dereliction of management and extension of the dry belt around the river (Fig. 2).

As reported also in other studies, management can be a very important factor in preserving open fen vegetation (Güsewell 2003; Middleton et al. 2006; Sundberg 2012). That can be seen in the upper basin, where a relatively undamaged hydrological system alone could not preserve open vegetation without management (Wassen and Joosten 1996). However, management alone cannot preserve vegetation, as evidenced from the middle basin, where Molinietalia meadows replaced the Calamagrostietum neglectae and Caricion diandrae fen vegetation, though shrubs and trees did not encroach. In the lower basin, however, management was subservient to hydrology as, despite reduction in management, vegetation changes were small.

So management can be crucial to preserve open fen vegetation, although it need not be essential. Without proper hydrological conditions vegetation will change despite any management, although it seems that under sub-optimal hydrological conditions management enables vegetation to persist.

Management measures

For Biebrza we identified a number of management measures that may mitigate the effect of climate change. Firstly, building-up dams in the main river tributaries increases the adjacent soil moisture and promotes suitable conditions for certain vegetation types. Next, ceasing vegetation removal in large stretches of the river diminishes river discharge rates similarly to building river dams. Vegetation removal from the river bed was a practice aimed at increasing the water velocity and decreasing water levels in the river and adjacent areas for agriculture. Hence, desertion of vegetation removal maintains natural water levels and increases adjacent soil moisture. Effectively these two management measures have an important role in mitigating water shortages that may result from climate change; either too low precipitation, seasonal variation in snow pack patterns and enhanced evaporation.

A third measure is mowing which ceases succession and promotes early successional stages and hence prevents succession to later successional stages such as shrubs, bushes and forests. Important in relation to climate change is the effect on transpiration. Low growing sedges and grasses transpire less water than shrubs and trees (Grygoruk et al. 2014). Hence, mowing in the Biebrza valley effectively conserves water and increases water retention. It is important to realize that cattle grazing and harvesting hay for feed of livestock was a traditional farming practice in Biebrza for centuries and was nearly abandoned in the mid 20th century (Budka et al. 2013).