Long-Term Floodplain Vegetation Dynamics after a Permanent Water Level Regulation

Abstract

This chapter is a review of studies on vegetation dynamics on a freshwater delta ecosystem after a permanent water level drawdown. Floodplain wetlands are globally threatened and have long been recognized as sensitive to changes in the hydrological regime. The original floodplain vegetation types were left “hanging” over the new water level, and secondary successions were immediately initiated both in the original vegetation and on the newly exposed sediments. Permanent transects were established running from the original vegetation toward the new water’s edge. Repeated analyses were conducted to document plant dynamics over a period of 33 years. The focus has been to analyze the dynamics of the decomposition of the original vegetation, the development of new vegetation on exposed sediments, and the varying fertility of plants. The plants’ responses are discussed in relation to effects of important environmental factors and why the original vegetation has not been reestablished after the regulation.

1. Introduction

A permanent regulation of a lake water level will have irreversible effects on the nearby floodplains. I have investigated vegetation dynamics after the regulation for 33 years, and the results are reviewed and discussed in the present article. Data collection was carried out in different parts of a freshwater delta in the Vestland County, SW Norway. The studies have focused on several aspects of vegetation changes both within the original vegetation and on the exposed sediments where new plants immigrated. Dynamic long-term trends have been studied with a focus on both individual species and on the development of secondary successions initiated after the regulation.

The near-shore areas of lakes and rivers are biologically important interfaces between terrestrial and aquatic ecosystems. The floristic structure of the floodplains is primarily regulated by the hydrological regime. Numerous studies have identified flood events as an important factor influencing plant distribution along elevation gradients on floodplains. Floods processes shape and maintain the spatial and temporal heterogeneity of floodplain areas and their plant communities [1]. Floodplain wetlands have long been recognized as sensitive to disturbances such as land reclamation and river regulation, which are globally threatened ecosystems. For example, 80–90% of the floodplains in Europe have been lost to agriculture and urban development [2].

Areas influenced by variable degrees of flooding have frequently been separated into three main zones: sublittoral, eulittoral, and epilittoral (supralittoral) [3, 4]. The littoral zones of undisturbed lakes may form narrow or broad fringing wetlands, with helophytes sorted by their tolerance to variations in the water level.

Two different succession processes were initiated after the regulation: a retrogressive (degradation of the original vegetation) and a progressive one (development of new vegetation on exposed sediments). In all areas, the plant populations will become dynamic and initiate long-time directional changes [5].

Littoral vegetation types usually include populations of many different species, which are dynamic because the individual plants have different physiologies, life cycles, and reproduction strategies. The vegetation in any area is naturally dynamic, with annual (seasonal) variations in growth, reproduction, and abundance. In areas influenced by environmental changes, long-term directional trends in the species composition will also be initiated. Dynamics of the behavior of plants in a community or study plot can include changes in species size, number of taxa (biodiversity), abundance (cover percentage), distribution, and fertility over time [6, 7]. To my knowledge, studies of plant dynamics over a long-term are few, with the exception for studies on tree species. The dynamics of herbs and graminoids are highly different because all aboveground biomass decay during the autumn and new shoots must be developed during spring and summer. Therefore, shoot biomass (cover) can be highly variable both from year to year and on the long term. My previous studies have quantified changes in the vegetation with focus on the following topics:

• The dynamics of species during the degradation of the original vegetation types

• The dynamics of the establishment of species on newly exposed sediments

• The dynamics of flower establishment during the secondary succession

• The dynamics of flower coverage in the degraded original vegetation

• The dynamics of the development of new vegetation types on exposed sediments

• The dynamics of changes in total species coverage and diversity during the development of new vegetation

• The dynamics of the development of new vegetation in areas where the original vegetation had been degraded

• The dynamics in species turnover and rate of floristic changes during the secondary successions

The results are discussed in relation to essential environmental variables and previous studies.

2. Study area

The investigations have been conducted on the Myrkdal delta (60° 40´N, 6° 28´E) located at the northern end of lake Myrkdalsvatn (229 masl.) in SW Norway. The delta extends approx. 700 m inward from the north end of the lake. The part of the delta closest to the lake is approx. 600 m wide and is divided into sections by several rivers and canals (Figure 1). The inner parts of the delta had for several 100 years been used for agricultural purposes. Most of these areas were more than 1 m above the original summer water level. The lower parts of the delta were originally dominated by Equisetum fluviatile and Carex vesicaria growing both above and below the water level.

Figure 1.

Photo (taken in 1978) of the lower parts of the Myrkdal delta before the regulation.

More than 90% of the delta sediments consisted of particles smaller than 125 μm. The pH was mostly around 5.0, the loss on ignition was around 3%, and the base saturation was mostly between 14 and 25% [8].

2.1 The water level regulation

The summer lake water level was permanently lowered by approx. 1.4 m in August 1987. In addition to the lowering, the outlet was extended to reduce water level fluctuations. The main reason for the regulation was to improve and expand the areas used for grass production. Frequently the water level increased to more than 3 m above the normal water level before the regulation. After the regulation, the area of cultural fields increased, and forests gradually developed dense stands close to the new lake shoreline (Figure 2).

Figure 2.

Recent map of the lower parts of the Myrkdal delta after the regulation. Yellow color indicates cultural fields and light green color indicates forested areas (From Norgeskart.no, the National Map Agency).

2.2 Hydrology and climate after the regulation

Hydrological data were made available from NVE (Norwegian Water Resources and Energy Directorate). A limnigraph had recorded water levels continuously in the lake. Based on daily measurements, mean June water level, mean summer water level (defined as mean level during May, June, and July), and mean annual water levels were calculated (Figure 3). The durations of the flood periods in percentage of the year were estimated at different elevations both before and after the regulation. The elevation of all study plots was related to the elevation scale on the limnigraph using a leveling instrument.

Figure 3.

Variation in the lake water level during the study period shown in relation to the new mean summer water level (184 cm). Mean water levels for the whole period was denoted M. The variation in mean water levels were estimated each year and denoted Wa.

Water level data and the vegetation conditions before the regulation have shown that populations dominated by E. fluviatile grew in areas where the flood duration lasted more than approx. 26% of the year. C. vesicaria grew where the flood duration was originally between approx. 10 and 26% of the year. Shrubs, trees, and grass species dominated where the flood duration was less than approx. 10%. After the regulation, the flood duration at mean summer water level was approx. 25%, and at the mean June water level, the flood duration was approx. 12%.

Mean annual rainfall in the area was approx. 1300 mm. Mean annual temperature was 4.4°C, mean July temperature was 13.5°C, and mean December temperature was −5.0°C.

Temperature data during the study period were interpolated from nearby meteorological stations recorded at Norwegian Meteorological Institute from 1987 to 2020. Summer temperatures (mean of June and July) varied between years from 12.4 to 17.6°C, with a mean of 14.5 ± 1.3°C. Highest values were in 2018 and lowest in 1993, 1996, and 2015. There was a significant linear increase of 2.1°C (r = 0.8, p < 0.01) in the summer temperature during the study period.

3. Review of the sampling methods

This article summarizes the results of investigations conducted between 1987 and 2020, where the sampling methods have been described. Several distinct aspects of vegetation ecology in different parts of the delta have been studied. The delta was first visited in August 1987, 1 month after the regulation. Three permanent transects were established for long-time monitoring in different parts of the delta. The transects covered different elevations above the new summer water level, and the survey periods could vary because different study topics were initiated and ended at different times. The study plots (1 m long and 0.5 m broad) were analyzed in late July or early August. In some analyses, data from four nearby plots were combined into ten 4 x 0.5 m large plots where the mean species coverage and the mean elevation were calculated. The species’ coverage was visually estimated as a percentage. The number of species recorded in the study plots (species richness) was counted, and the total coverage of all species in each of the plots was summed.

The main transect (T1) covered an area from the original E. fluviatile vegetation toward the new water level [8, 9]. T1 included 40 plots situated at elevations from 7 to 97 cm above the new mean summer water level (Table 1). In time, plants were only established above the new summer water level. T1 therefore does not include the lower parts of a potential full littoral zonation. T1 also included eight plots in the lower part of the original E. fluviatile vegetation, which was situated 30–75 cm below the original summer water level. The transect was surveyed a total of 19 times between 1987 and 2020, most often at one-year intervals. Variations in the number of flowers produced by C. vesicaria and Carea rostrata were counted from 1993 until 2020.

Limn	Z1	H1	H2	F% 1	F% 2
394		62
	Epilittoral
		32		<10%
364		24	174	10%
	Eulittoral
332	WL 1	0	147
		−29	121	26%
304		−30	120	>26%
	Hydrolittoral
		−57	97		< 1%
		−75	75
255					<10%
	Exposed
			15
					12%
			7		13%
184	WL 2		0		25%
			−5		>25%
	Aquatic
124			−60

Table 1.

Schematic overview of the elevation (cm) of the littoral zones and mean summer water level on the delta before (Z1) the regulation.

WL1 indicates the original summer water level and WL2 indicates the new summer water level. Elevation (H) and duration of flooding (F) of the zones were estimated both in relation to the situation before (H1) and after the regulation (H2). Duration of flooding (in percentage of the year) associated with the elevational distribution of the littoral zones before (F% 1) and after (F% 2) the regulation.

Transect T2 was established in an original C. vesicaria-dominated vegetation zone [10]. T2 included 15 plots situated from 24 cm above to 29 cm below the original summer water level. Variation in species coverages was investigated from 1988 until 2002, and the variation in the number of flowers produced by C. vesicaria was counted from 1993 until 2002.

Transect T3 was established in the lower part of the exposed delta sediments. T3 included 14 plots situated from 15 cm above to 5 cm below the new summer water level. Variation in species coverages was investigated from 1988 until 2002, and the variation in the number of flowers produced by C. vesicaria and C. rostrata was counted from 1993 until 2001 [11].

The vertical distribution of the studied vegetation zones was estimated in relation to the scale of the lake limnigraph. Elevations of the transect plots were given in relation to both the old and the new mean summer water level (Table 1).

Data collected in the various investigations were analyzed using statistical methods. Long-term trends in species cover and flower production changes were analyzed by regression analyses [10]. Trends in the development of new vegetation types were explored by ordination/gradient analyses. Ordination techniques are used in ecology to describe long-term relationships between gradients in species composition patterns and underlying environmental factors. Variation in species study plot composition during the study periods will be related to time (number of years) or elevation. DCA (Detrended Correspondence Analysis) is an unconstrained ordination method where environmental factors were included as passive variables. DCA can be used to find trends in data with large standard deviation [9]. The rate of change (often denoted ROC) or species compositional floristic turnover is a measure of the speed at which a variable changes over a specific period.

Long-term trends in floristic change in data with relatively little standard deviation were explored by PCA (Principal Component Analysis). DCA and PCA are multivariate statistical techniques widely used by ecologists to find long-term trends or gradients in ecological studies. The various methods that were used are described in the results section.

The degree of plant fertility is generally assumed to be more sensitive to the environmental conditions than their vegetative structures. To elucidate this topic, the number of C. vesicaria and C. rostrata shoots producing flowers was counted in study plots from 1993 onward. The plants produce new shoots from underground rhizomes each year, and a varying proportion of these may produce flowers. The number of shoots with flowers was counted in the 0.5 m² plots at each survey.

4. Results and comments

During the study period, the frequency and coverage of all species were highly dynamic. The number of occurrences of the most common species recorded in T1 during the study period is shown in Table 2. All species can be regarded as common in most freshwater wetlands.

Species	Abb.	No
Agrostis stolonifera	A.sto	117
Alopecurus geniculatus	A.gen	13
Calamagrostis phragmitoides	C.phr	316
Callitriche palustris	C.pal	22
Calttha palustris	C.pal	52
Carea rostrata	C.ros	474
Carex canescens	C.can	95
Carex echinata	C.ech	12
Carex leporina	C.lep	79
Carex nigricans	C.nig	35
Carex vesicaria	C.ves	548
Deschampsia cespitosa	D.ces	336
Eleocharis acicularis	E.aci	7
Filipendula ulmaria	F.ulm	41
Galeopsis bifida	G.bif	8
Galium palustris	G.pal	339
Juncus filiformis	J.fil	263
Phalaris arudinacea	P.aru	592
Ranunculus repens	R.rep	477
Ranunculus reptans	R.rept	35
Salix nigricans	S.nig	21
Sparganium angustifolium	S.ang	23
Subularia aquatica	S.aqu	26

Table 2.

The most common species recorded in transect 1, with full names and abbreviations.

No = number of occurrences in the 760 plots during the study period.

4.1 Decomposition dynamics of the original vegetation types

As a result of the regulation, the original vegetation zones were left “hanging” above the new water level. It had to be expected that these zones in time would disappear on site. Before the regulation, the eulittoral zone was dominated by C. vesicaria (T2), and the hydrolittoral zone was dominated by E. fluviatile (T1). The zones had wide distributions in the lower parts of the delta. Both species are clonal, and the populations were monocultures. Interesting questions would then be how long the species would survive in their original positions, and how the dynamics of the plants’ decomposition developed. T1 and T2 were monitored for several years [10, 11, 12].

The C. vesicaria zone was situated between a zone dominated by Phalaris arundinacea on the upper part and E. fluviatile on the lower part. T2 consisted of 15 plots situated between 24 cm above and 29 cm below the original summer water level and 174–121 cm above the new summer water level (Table 1). The plots were surveyed from 1988 until 2002.

The coverage reduction of C. vesicaria started from the edges (upper and lower part) of the original zone [10]. Mean plot coverage decreased linearly from 94.4 to 1.3% (R² = 0.96) for 15 years. The coverage reduction was, however, variable between the study plots. In the middle part, C. vesicaria maintained a coverage of 100% until 8 years after the regulation. Fourteen years after the drawdown, the coverage was 40% in the middle part of the zone, while it had disappeared or had a coverage lower than 20% at the edges. After 15 years, only a few scattered shoots were recorded [10, 11].

The coverage of E. fluviatile was studied in eight permanent plots situated at the lower part of the original E. fluviatile zone [11]. The plots were situated between 30 and 75 cm below the original summer water level and 120–75 cm above the new summer water level. The coverage reduction was variable from year to year, but the mean coverage decreased in general linearly from 39.8% in 1988 to 3.3% after 14 years (R² = 0.81) when E. fluviatile had disappeared almost completely. However, the occurrence of scattered shoots showed that the rhizomes were still alive 33 years after regulation [13].

The results show that the coverage reduction of the two species was dynamic, but both had disappeared almost completely 15 years after the regulation. C. vesicaria retained a 100% coverage for 8 years after drawdown in the center of the population. The mean coverage of E. fluviatile was higher than expected in 1995 (year 8), probably because of a high water level that year (Figure 3).

This may indicate that E. fluviatile was more dependent on periods with high water level than C. vesicaria and therefore more sensitive to the drawdown. This is surprising since the regulation represented a greater water level change on C. vesicaria than on E. fluviatile since the former was originally situated at a higher elevation.

4.2 Dynamics of species immigration on the exposed sediments

The immigration of plants on the exposed sediments was highly dynamic both in relation to time and space (elevation above the water level). Already 1 month after the regulation, several plants were established on the exposed sediments. During the first few years, the sediments were dominated by bryophytes [8, 14]. The bryophyte richness was high, especially with many acrocarpous species. The coverage of the dominant bryophytes was highly dynamic during the first years. Blasia pusilla dominated the first year, frequently with a plot coverage of 70%. The next years, Polytricum commune increased gradually until it reached a mean coverage higher than 40% after year 4. Also, the Drepanocladus species increased gradually and reached a peak during year 5. Later, the bryophyte coverage decreased strongly, and most disappeared as the coverage of vascular plants increased.

Several vascular plants were also quickly established on the sediments, but their coverage was low the first month after the regulation in 1987 [8, 9, 14]. Dominants were Alopecurus geniculatus, Subularia aquatica, Ranunculus reptans, Callitriche palustris, Eleocharis acicularis, and Sparganium angustifolium. These species were mainly found in the six first years after the regulation. Agrostis stolonifera was also common until year 8, and Juncus filiformis was common until year 10.

The coverage of the tall plants gradually increased, and most bryophytes and the small vascular plants disappeared after year 5. Frequency of vascular plants in the 40 plots in T1 during the study period is shown in Table 2, and their coverages have previously been described [9].

Mean coverage of the most common species established along T1 was highly dynamic during the study period (Figure 4). The most rapid invader was Deschampsia cespitosa, which had a plot coverage between 0 and 80% the first year after the regulation. C. vesicaria and C. rostrata needed 4–5 years to reach high coverages. C. vesicaria had high coverage during two periods: between years 3–10 and years 29–33. P. arundinacea and C. phragmitoides reached 100% cover after approx. 10 years. The coverage of P. arundinacea increased over the whole period and had a mean coverage of approx. 50% from year 13 until year 33. P. arundinacea and C. vesicaria dominated during the last study period, but they were mainly separated to elevations above or below elevations of 50 cm, respectively (Figure 5). The mean coverage of C. phragmitoides, D. cespitosa, and Ranunculus repens decreased during the last study period. F. ulmaria was sporadically present and reached a coverage of 40–70% between year 20 and 26.

Figure 4.

Mean coverage of the most common species in all the 40 study plots estimated each study year. Species abbreviations are explained in Table 2.

Figure 5.

Mean coverage of the most common species in the 10 study plots with different elevation above the mean summer water level estimated during the whole study period. Species abbreviations are explained in Table 2.

The lowest part of the original zone dominated by E. fluviatile was situated at a 31 m distance from the lake summer water level. After the regulation, the rhizomes began to migrate downward toward the new water level. After 13 years, it had moved 22 m, which gave a mean migration of approx. 1.7 m/year. E. fluviatile had almost completely disappeared in T1 after 18 years [13].

The 40 plots in T1 were divided into 10 elevation zones above the mean summer water level. Mean coverage of the species was estimated in each of these zones. The species’ coverages were dynamic also in relation to elevation, indicating that the species responded differently to the distance above the mean summer water level during the study period (Figure 5). The coverage of C. vesicaria increased toward low elevations, and it became more stable at elevations lower than approx. 40 cm. The coverage of P. arundinacea was the highest between 50 and 70 cm, while the coverage of C. phragmitoides was the highest around 80 cm. A. geniculatus, J. filiformis, R. reptans, E. acicularis, and S. aquatica were all most common below 40 cm.

The study shows that the invasion of species on the exposed sediments was rapid. A prerequisite for this was that the sediments consisted of fine-grained material and had good moisture conditions and that large seed banks were normally present in floodplain sediments [15]. The pioneer plants were mostly a mixture of annuals, aquatic plants, and helophytes. These were over the first years distributed over most of all transect plots. Later, clonal helophytes and grasses became dominants. D. cespitosa, C. vesicaria, and C. rostrata were the first dominants followed by C. phragmitodes and P. arundinacea. Over most of the study period, only C. vesicaria and P. arundinacea retained a high coverage. But the coverage of all species was highly dynamic. Most species responded in time differently to the distance above the mean summer water level, becoming separated to specific elevations. After 2001 (year 14), the species distribution had become stable in T1 with the following weighted elevation optima: C. vesicaria at 31 cm, C. rostrata at 32 cm, P. arundinacea at 60 cm, C. phragmitoides at 76 cm, and D. cespitosa at 87 cm [11]. Before the regulation, C. vesicaria populations were found from 24 cm above to 29 cm below the summer water level (Table 1). This shows that C. vesicaria had not developed populations in relation to the water level as before the regulation.

4.3 Dynamics in flower production

The number of C. vesicaria and C. rostrata shoots producing flowers was counted in 26 plots in T1 from 1993 to 2020. During this period, both water levels and air temperatures varied strongly [12]. The study plots were situated between 7 and 59 cm above the mean summer water level. In plots situated higher than 60 cm, almost no shoots produced flowers.

In most plots, C. vesicaria produced more flowers than C. rostrata. The largest mean number of shoots with flowers in the plots was 43 for C. vesicaria and 30 for C. rostrata. The largest numbers found in all 26 plots throughout the study period plots were 344 (2019) and 212 (1994). The variation in the mean number of flowering shoots was highly dynamic during the study period (Figure 6). Except for the first 4 years (1993–1996), C. vesicaria produced more flowering shoots than C. rostrata. The mean annual number of flowering shoots during the study period was 2.1 for C. rostrata and 3.5 for C. vesicaria. There was no significant trend in flower production during the study period. Both species produced almost no flowers in 1995 and 2016, which were years of high water levels (Figure 3). The species showed almost identical relative annual variations in flower production, with a significant correlation (r = 0.62) [12]. This indicates that both species were affected by the same environmental factors although the degree of flowering was different.

Figure 6.

Dynamics in the mean number of C. vesicaria and C. rostrata shoots producing flowers in the eight plots between 1993 and 2020.

The significance of water level and air temperature for the variation in flower production was investigated statistically. Different values for annual water levels and temperatures were tested. The flowering of C. vesicaria was significantly related to the mean summer water level (average for June and July) and the summer temperature (average for June and July), both estimated on data from the current year and the previous year. The flowering of C. rostrata showed no significant relationship to these factors. The production of C. vesicaria flowers was not significantly related to its coverage degree [12].

The number of flowering shoots of C. vesicaria and C. rostrata was also counted at low elevations (T3) from 1993 until 2001 [11]. T3 included 10 plots situated from 15 cm above to 5 cm below the mean summer water level. Also, in these plots, the production of flowering shoots was highly dynamic, but the results were different from the data in T1. In T3, C. rostrata produced more flowers than C. vesicaria most years. The mean number of flowering C. vesicaria shoots was 10.1, and it was 14.5 for C. rostrata. During the study period, the number of C. rostrata shoots producing flowers increased significantly, while C. vesicaria showed no significant trend.

The number of flowering C. vesicaria shoots was also counted during the degradation period in T2 from 1993 until 2002 [11]. The average coverage of C. vesicaria decreased with time, but the number of flowering shoots was dynamic. Relatively few flowering shoots were recorded in 1995, which was a year with high water level (Figure 3). The number of flowering shoots was highest in 1996 (the highest number in a plot was 35). Ten flowering shoots were found in the middle of the transect, but almost none in either the upper or the lower plots in 1996. After 2002, no flowering shoots were found. This shows that shoots with flowers were recorded up to 15 years after the regulation.

Comparisons between the number of flowering shoots in the three investigated transects were different with major differences from year to year. The mean numbers of flowering C. vesicaria and C. rostrata shoots between 1993 and 2001 were 4.5 and 4.1 in T2 and 10.1 and 14.5 in T3, respectively. Obviously, the number of flowering shoots of both species varied with distance to the summer water level. The results indicate that drawing conclusions from a few years of research and from only one area can give contradictory results.

4.4 Long-time population dynamics of C. vesicaria and P. arundinacea coverage

Population dynamics is a field within ecology where changes in species occurrence (quantity or number) and distribution patterns are investigated over time. Whether there are significant long-term trends can be investigated using statistical methods.

C. vesicaria and P. arundinacea dominated the transect study plots over most of the study period, but their coverage varied both with elevation and time after the regulation. Their long-term coverage dynamics was therefore interesting to study by statistical methods [16].

Variations in mean coverage of C. vesicaria and P. arundinacea in the 40 study plots over 33 years are shown in Figure 7. The boxplots show that the coverage of both species was highly dynamic. The increase of C. vesicaria coverage started earlier than for P. arundinacea. During the last 4 years, both species had high coverage but at different elevations (Figure 5).

Figure 7.

Boxplots showing the variation in the mean coverage of C. vesicaria (red columns) and P. arundinacea (blue columns) in the forty transect plots during the study period.

Over time, the species’ coverage was gradually separated to specific elevations [16]. The 40 plots in T1 were divided into 10 consecutive elevation zones. The study plots were in this case 4 x 0.5 m large (2 m²), with mean elevations from 13 to 93 cm above the summer water level. The species’ mean coverages were estimated in each of these zones. During the study period, the coverage was highly dynamic as shown in Figure 8. P. arundinacea coverage was the highest at a mean elevation of 66.3 cm, while C. vesicaria had the highest coverage at low elevations. Over the two last years, their coverages were more equal, but they were mainly separated to below or over elevations of approx. 50 cm.

Figure 8.

Boxplots showing the variation in mean coverage of C. vesicaria (red columns) and P. arundinacea (blue columns) in ten transect plots with mean elevations between 92 and 13 cm above the mean summer water level during the study period.

Coverage variations from 1988 to 2020 at a mean elevation of 44.5 cm are shown in Figure 9. Trends in their dynamic mean coverages were statistically tested by different polynomial regression models. The long-term trends were significantly explained by third-degree functions [16]. At elevations between 37 and 56 cm, both species significantly followed third-degree (cubic) functions. In most of the other elevation levels, the trends were also significantly explained by third-degree functions, but the significance levels were often lower. The long-term dynamics mostly explained by third-degree functions are difficult to explain in terms of plant biology and ecology (see chapter 5.1). The exposed sediments situated less than 70 cm above the mean summer water level were dominated by P. arundinacea and C. vesicaria as shown in Figure 10.

Figure 9.

Mean coverage of C. vesicaria and P. arundinacea in plots situated at a mean elevation of 44.5 ± 1.1 cm from 1988 to 2020. The data show significant third-degree (cubic) distributions.

Figure 10.

Vegetation development on the exposed sediments 33 years after drawdown. Mainly C. vesicaria andP. arundinacea dominated areas less than 20–30 m from the summer water level. S. nigricans developed most areas situated more than 20–30 cm above the water level. The photo was taken in 2020, but later the changes were small. Vegetation development out into the water had been limited below the mean summer water level.

4.5 Development of new vegetation on the exposed sediments

The exposed sediments were rapidly invaded by species, as shown in Figure 4. It could be expected that both the floristic vegetation composition, the total coverage, and the species richness in the plots should be dynamic but become more stabilized during the last periods of the study period. The long-term trends during the secondary succession have previously been described by Refs. [8, 9].

Twenty-six species were recorded in the transect plots during the study period (Table 1), but many of them were sparse and disappeared after short periods. The mean coverage of species established in all study plots was estimated during the study period, and the coverage sum of the species is shown in Figure 11. The sum of the mean coverage of all species increased gradually during the first 13 years. The highest coverages were found during year 7–14, and thereafter, it decreased. This shows that the total plot coverages during the succession were highly dynamic. The total coverage was higher during the first year after the regulation than during year 2. This was mainly a result of the high coverage of the small pioneer plants along T1. During the study period, species richness was the highest during year 2 with 4–13 species in all T1 plots. After 13 years, the richness varied mostly between 1 and 6, with a mean of 4.2 ± 1.4. A quadratic model significantly described the long-term trend [9].

Figure 11.

Variation in mean coverage of the six most common species in transect T1 from 1988 until 2020. The sum of the mean species coverage in the forty plots has been estimated each study year. Species abbreviations are explained in Table 2.

Previous studies have shown that the variation in number of species can be highly variable during secondary successions [8]. Biodiversity may increase, decrease, or be the highest at intermediate succession stages, but there are few available long-term studies on the variation in species richness during a secondary wetland succession.

4.6 Rate of change during the secondary succession on exposed sediments

Changes in the floristic composition of the T1 plots were analyzed by DCA [8, 9]. The units are expressed in standard deviations (SD) of species turnover rate along floristic gradients [17].

The variation in floristic composition of the study plots along T1 each year and throughout the entire study period is shown in Figure 12. The vertical axis (DCA axis 1) shows the main difference between the plot floristic composition along T1 each study year. The compositional floristic turnover in relation to time showed a significant non-linear trend. During the first 15–16 years after the regulation, there was a strong change in the species composition. Over the last years, however, the mean DCA axis 1 plot scores showed relatively small floristic changes.

Figure 12.

Result of the DCA analysis. Variation in DCA axis 1 sample scores plotted against the study years after the regulation. Result of the lowess smoothing is shown.

Previous DCA analyses have shown that the time (number of years after regulation) and the elevation above the summer water level explained equal parts of the succession development [8, 9, 18]. DCA axis 2 showed a significant linear compositional floristic turnover along the elevation gradient.

The variation in the species composition in relation to the water level appears to be constrained. After year 7, the floristic composition in all study plots situated in the lower part of T1 had almost the same DCA axis 1 plot scores (Figure 12).

The variation in the floristic composition along DCA axis 1 showed a significant polynomial change in the floristic composition in relation to time [8, 11]. After 15–16 years, the mean DCA axis 1 plot scores showed almost no long-term trend.

4.7 The dynamics of plant immigration in the original vegetation zones

It had to be expected that new vegetation types would gradually be developed as the original vegetation was degraded. Interesting questions would be to investigate the dynamics of the invading species and how long it would take before new stable vegetation types were established.

Species invasion in the 15 study plots originally dominated by C. vesicaria is shown in Figure 13. C. vesicaria retained a high coverage for several years, and the invasive species had a low coverage. After 15 years, C. vesicaria rarely occurred. P. arundinacea was established with a coverage of 30% after year 3. After 14 years, its coverage was higher than 70%. C. phragmitoides had a 60% coverage in year 9, but thereafter, it decreased. E. fluviatile was sporadically recorded with low coverage from year 2 to year 14. After year 12, P. arundinacea was dominant, and other species had low coverage [10].

Figure 13.

Dynamics of the plant invasion in the original C. vesicaria zone. Species abbreviations are explained in Table 2.

Variation in coverage of species immigrating in the original E. fluviatile zone is shown in Figure 14. D. cespitosa was established the year after the regulation, and it was recorded until year 29. C. phragmitoides was also quickly established, and after year 7, it became dominant for a period. After year 29, it was not recorded, and later P. arundinacea took over as the dominant species. C. vesicaria was recorded with low coverage from year 8 but with somewhat higher coverage in years 32 and 33.

Figure 14.

Dynamics of the plant invasion in the original E. fluviatile zone. Species abbreviations are explained in Table 2.

The development of new vegetation types in the original littoral zones was highly dynamic. The species richness was higher in the original E. fluviaile zone, but after 29 years, P. arundinacea became dominant in both zones.

4.8 Rate of vegetation change after the decomposition of the original vegetation types

Invasion of new species in the original vegetation zones was highly dynamic (Figures 13 and 14). Variations in the mean cover of all species along the transects in relation to time were analyzed by PCA (Principal Component Analysis) [17] with time as passive variables. The analyses describe the development during the secondary successions in the original vegetation types.

The rate of change of the mean coverage of species in all the transect plots during the development of new vegetation in the C. vesicaria zone is shown in Figure 15. The analysis shows a gradual change in the floristic composition during 15 years after the regulation. The plots were not studied after 2002 (year 15), but P. arundinacea had by then become the dominant species.

Figure 15.

Dynamics of vegetation development in the original C. vesicaria zone. The PCA analysis shows major species change in the fifteen study plots during the first 15 years after the regulation. The analysis was based on the mean coverage of all species in the 15 study plots from 1988 until 20. Later, the floristic changes were small. Result of the lowess smoothing is shown.

The rate of floristic change in all the transect plots during the development of new vegetation in the original E. fluviatile zone is shown in Figure 16. The analysis showed a gradual change in the floristic composition during the first 10–15 years after the regulation. Thereafter, the changes decreased, but there were still some variations.

Figure 16.

Dynamics of vegetation development in the eight plots in the original E. fluviatile zone. The analysis was based on the mean coverage of all species in the eight study plots from 1988 until 2020. After 10–15 years, the floristic changes were relatively small. Result of the lowess smoothing is shown.

The original coverage of C. vesicaria was 100%, while in the plots dominated by E. fluviatile, the coverage was only 40%. The invasion of other species in the E. fluviatile plots was therefore faster and more species-rich than in the C. vesicaria plots. In the C. vesicaria vegetation, the coverage sums of all plants increased to 200% in year 6 but thereafter decreased to 130% in year 15. The species richness varied mostly between 6 and 8 during the study period. In the E. fluviatile zone, the coverage sums of all plants varied between 70 and 130%, and the species richness varied between 19 and 7. In both populations, P. arundinacea, C. phragmitoides, and R. repens became common and often dominant. These trends appeared to be the same over most of the lower parts of the delta after the regulation. Although the vegetation and ecological conditions in the two types were different initially, both showed the same long-term trends after regulation, and new stable conditions were reached after approx. 15 years.

5. General discussions

Effects of seasonal water level drawdowns frequently employed for hydropower and flood control have widely been studied. Long-term vegetational effects of permanent lake water level drawdown have, however, hardly been studied previously.

The dynamic vegetation changes described in these investigations are effects of interactions between hydrological and plant physiological factors. It is easy to describe changes in plant distribution after a regulation, but to explain the results in terms of plant biology is difficult. The underlying mechanisms governing large-scale species turnover are ambiguous and poorly studied [19].

The changes in species’ coverage and distribution have also resulted in changes during the secondary vegetation successions. Patterns of vegetation succession after a drawdown depend on factors such as sediment characteristics, topography, short- and longtime changes in the water level fluctuation, and effects of wind and wave erosion [20]. These factors may all influence the biological and physical structures of the littoral zones. The primary factor that distinguishes wetlands from terrestrial landforms is that the wetland vegetation is dominated by helophytes adapted to the unique anoxic hydric soils. The plants have structures both above and below the soil surface, and the biomass is dynamic in both places.

5.1 Growth factors associated with plant coverage dynamics

Most plant species of wetland habitats expand through clonal growth, often forming dense, monospecific stands. Underground plant rhizomes continuously produce new shoots and root systems. Clonal propagation is the main mechanism of macrophyte reproduction and population growth, while sexual reproduction is often a subordinate strategy [21]. Neither water levels nor temperatures could explain the coverage variation in the study plots [16]. It is therefore likely that the plants’ growth responses control the cyclical variation in coverage.

The capacity of clonal plants to exchange resources and growth substances between interconnected shoots has long been held responsible for the ability of clonal plants to prevent shoot overproduction and for the general lack of density-dependent mortality and self-thinning [18, 22]. The long-term variation in plant coverage is therefore probably mainly a result of rhizome growth dynamics, and not of aboveground self-thinning [23].

Previous studies have shown that P. arundinacea produces dense crowns and prominent networks of vigorous underground rhizomes, allowing for aggressive vegetative spread on disturbed and moist sites [24]. Tillering young clones could cover an area of one square meter consisting of 100 tillers by the end of the first growing season. During the fourth year of its life, the partial clump did not form aboveground shoots and died out during the fifth year.

Aerial growth dynamics of P. arundinacea and C. vesicaria have been studied for 2 years [25]. Major differences were found, probably because of differences in underground clonal growth. The lifespan of the dominant species may explain their long-term coverage dynamics, but there are few studies on the life history of plants. The average lifespan of C. rostrata shoots varied from 1.5 to 6 years, with an annual shoot survival rate of 60% [26, 27, 28]. There are few available studies on C. vesicaria, but it can have an intensive long-term tillering [29]. Consequently, it is not immediately easy to explain why the coverage dynamics of P. arundinacea and C. vesicaria were significantly described by cubic functions.

5.2 Effects of the water level amplitude

Water level amplitudes and the duration of flooding periods have large effects on the ecological processes and patterns of littoral vegetation zonation [30, 31]. Flooding duration, which is a function of the water level amplitude, can be an important stress factor for plants, but this proposition has never been tested [32].

The number of days with flood impact decreases with increasing elevation above the normal water level [33]. Based on this, elevation can be used as a proxy for flood frequency. Water level fluctuation can be estimated in number of days or quantified on different time scales: daily, seasonal, during months, and during the study periods. The mean annual flooding duration can be estimated for the mean June water level (1 month) to be 8.2%, and the mean summer water level (3 months) to be 24.7%.

Harmful effects of reduced water level amplitudes have previously been demonstrated [34]. The flooding tolerance of angiosperms species determines their distribution along flooding gradients in river forelands [32, 35, 36]. The low diffusion rate of gases (and more particularly carbon dioxide and oxygen) leads to problems of carbon and oxygen supply [37, 38]. A major problem for many species will therefore be to survive long periods of flooding.

These studies indicate that short flooding periods are not critical; it is the long-term mean water levels that are most important. Experiments have shown that species inhabiting river foreland survived both 3 and 6 weeks of flooding and were classified as tolerant to flooding duration [37, 39]. The production of flowering shoots was significantly reduced during years with a high water level.

C. vesicaria, D. cespitosa, and P. arundinacea growing on a floodplain have their optima at different elevations relative to the water level [40]. C. vesicaria was found to be confined to the lowermost parts that were flooded for more than 40 days during the study period.

P. arundinacea has been found to be sensitive to summer flooding, and after 3 weeks, rhizome growth was reduced [41]. A 50-day flood period (ca. 14% of the year) adversely affected its biomass production [42, 43]. It was restricted to elevations less than 1 m above normal high water level and only 11.2 days of flooding during the growing season [44].

Results of the present investigation are consistent with previous results, which have shown that flooding determines the distribution of most floodplain plants. The prevalence of P. arundinacea and C. vesicaria, which were dominants, was found above or below approx. 50 cm above the summer water level, where the flood duration was around 10%. As indicated in Table 1, the transition between the epilittoral and the eulittoral zones had a flooding duration around 10%, and the transition between the eulittoral and the hydrolittoral zones had a flooding duration around 26%. The studies show that the elevational distribution of the littoral zones is associated by the average long-term water levels. The mean flooding duration had negligible effect on the species’ coverage, but it can have a decisive effect on the development of flowering shoots.

5.3 Effects of wind and wave erosion

Transect T1 was initially supposed to describe the vegetation development in the new sublittoral zone, but below the mean summer water level, only scattered shoots were periodically established. Previous studies have also shown that wind and wave actions can restrict the distribution of the littoral vegetation zones by regulating plant development on the within-lake shorelines [45, 46, 47, 48].

The reduced water level amplitude increased the erosion at the mean summer water level. This is probably because the water level fluctuations have become much smaller and more stabilized at the mean summer water level. Before the regulation, there were well-developed populations dominated by E. fluviatile down to approx. 60 cm below the summer water level (Table 1). Studies have shown that the new sublittoral vegetation was limited to small areas protected from erosion after the regulation. In delta areas that were protected from erosion, vegetation dominated by C. vesicaria was found below the mean summer water level, and E. fluviatile developed populations down to a water depth of approx. 1 m [11, 49].

5.4 The value of long-term studies

This review demonstrates the importance of studies in permanent plots over several years to document the dynamics of both individual plant coverage and the overall floristic composition of the study plots. Previous investigations have also emphasized needs for surveys of permanent plots. Any study of vegetation dynamics at the species level is impossible without permanent plots [34, 50, 51]. Long-term time series are necessary to better understand population dynamics because brief time series may lead to wrong conclusions, given large natural year-to-year variability [52].

Several ecologists have called for more studies on plant population dynamics because they are at the core of the ecological sciences but poorly represented in the current literature [53]. Explanation of why and how plant dynamics change in space and time is important from a fundamental scientific point of view and has become a cornerstone of modern ecological research and conservation biology [54, 55].

An essential question has been raised as to how long of a time series is necessary because brief time series may lead to wrong conclusions given large natural year-to-year variability. It has been shown that to identify long-term changes in abundance, on average 15.91 years of continuous monitoring were often required, but there was a wide distribution of estimated minimum times [52]. It was also shown that 72% of time series required at least 10 years of continuous monitoring to achieve a high level of statistical power.

Most of the plants examined in the delta studies showed major annual variations in their biomass (coverage). An essential question would then be how many study years would be necessary to describe the different long-term trends. The several topics investigated in these studies required different lengths of time to fully describe the long-term dynamics.

A period of 33 years appeared to be necessary to show that the long-term coverage dynamics of C. vesicaria and P. arundinacea significantly followed third-order functions.

The development of secondary successions should, in theory, over time result in stable floristic compositions. In the present studies, the species compositions in the study plots were stabilized after approx. 15 years. The degradation of the original vegetation types also required a survey period of 15–18 years.

The number of C. vesicaria and C. rostrata shoots producing flowers showed no significant trend during the whole survey period. Both species responded, however, with decreased flower production during years with high water levels. C. vesicaria responded also with significantly increased flower production during years with high summer temperatures. This shows that a study period of 27 years was necessary to establish significant data on factors of importance for C. vesicaria flower production.

5.5 Have new littoral zones been developed in accordance with the new hydrological regime?

It would be interesting to investigate if the secondary succession was stabilized during the study period and if a new vegetation zonation was recreated in accordance with the original conditions after the regulation. These questions are particularly important where rare or vulnerable habitat types are affected.

The regulation has changed the distribution of several vegetation types. Usually, during a secondary succession, the vegetation will gradually stabilize [56, 57, 58, 59]. The DCA analyses of the vegetation in T1 have shown that after approximately 15 years, no long-term trends in the distribution of the species were recorded. Another fact indicating that the ecological conditions were stabilized was that biodiversity no longer showed a long-term trend. The species richness in T1 was stabilized after approx. 15 years [8, 9].

But the littoral zonation has not been recreated as it was before the regulation. Wind and wave erosion over large areas have prevented eulittoral and sublittoral vegetation types to develop and reduced the extent of the vegetation dominated by E. fluviatile. Before regulation, it could dominate areas more than 100 m from the water edge, while at present, only scattered stands occur in protected areas. In undisturbed lakes, E. fluviatile may extend to sites more than 100 cm below the summer water level, while C. rostrata and C. vesicaria are often found 60 cm below the mean summer water level [60, 61]. The distribution of C. vesicaria was also reduced, but stands have developed in areas situated around the mean summer water level.

The stabilized water level fluctuations have decreased the areas affected by frequent flooding, which has benefitted the distribution of S. nigricans. Most tree species do not tolerate long periods of flooding. S. nigricans has largely expanded its range and now covers large parts of the lower part of the delta where forest did not exist before the regulation (Figures 1 and 2). Large parts of the delta situated 20–30 cm above the summer water level and approx. 30 m inland from the water edge are at present dominated by S. nigricans forests [11]. The forest field layer is mainly dominated by C. phragmitoides, P. arundinacea, and D. cespitosa, which do not tolerate long flooding periods (Figure 10).

Dense stands dominated by E. fluviatile were gradually developed in areas protected from wave erosion 10 years after the regulation. They grew from the mean summer water level down to a depth of more than 60 cm [11].

6. Conclusions

The investigations have shown major dynamics of both individual plant species coverage and the floristic composition of the new vegetation types after the regulation. Due to favorable substrates, the changes have proceeded very quickly. The distribution of both species and vegetation types was permanently changed compared to the state before the regulation. The littoral zones have shifted downward toward the new summer water level. The distribution of sublittoral and eulittoral vegetation has been reduced, and forests dominated by S. nigricans have become more widespread.

All the investigated species showed major coverage dynamics both annually and across long terms. The plants had periods of high coverage interspersed with periods of low coverage. But the trends showed little agreement between the species, and the lengths of periods with high and low cover were mostly variable. More specific results of the investigations are:

• The original delta vegetation dominated by E. fluviatile and C. vesicaria was gradually degraded and disappeared almost completely after approx. 15 years.

• After the regulation, E. fluviatile began to migrate downward toward the new water level from its original lowermost site.

• The number of C. vesicaria shoots producing flowers was reduced over time in the original vegetation. After 15 years, no flowering shoots were recorded.

• Sediments exposed after the regulation were completely dominated by vascular plants within 3 years. Study plot total coverage became higher than 100% after 7 years.

• Over time, most plants achieved their greatest coverage at specific levels above the summer water level.

• Most species, except C. vesicaria and C. rostrata, had their highest coverage 50 cm above the summer water level.

• At the end of the study period, C. vesicaria and P. arundinacea became dominants but at different elevations above the summer water level.

• The coverage variation of C. vesicaria and P. arundinacea throughout the study period significantly followed third-degree distribution trends at most elevations above the summer water level.

• Study plot species richness decreased strongly the first years after the regulation but was stabilized after approx. 15 years.

• The secondary vegetation succession became stabilized after approx. 15 years. Because of a more stable summer water level, vegetation did not develop in accordance with the original summer water level.

• In most years after the regulation, C. vesicaria produced more flowers than C. rostrata. The number of flowering shoots in both species was dynamic but significantly correlated during the study period. Both species increased the production of flowering shoots with decreasing elevation.

• The dynamics of C. vesicaria flowering shoots increased significantly with high summer temperature and decreased with high water level.

• The production of flowering shoots decreased during years with high water levels, while their coverages were not affected.

• Dense S. nigricans forest and thickets dominated by C. phragmitoides, P. arundinacea, and D. cespitosa today dominate most areas between the cultivated fields and the lake water’s edge.

• The present results show that the regulation has been negative because a valuable delta area has lost many of its original qualities.