Impact of drying and re-flooding of sediment on phosphorus dynamics of river-floodplain systems.

One of the consequences of human impacts on floodplains is a change in sedimentation leading to enhanced floodplain aggradation. Thus, accumulated sediments rich in nutrients might interfere with floodplain restoration. In this study we investigated the phosphorus release behavior of sediments from shallow backwaters of an isolated floodplain of the Danube River situated east of the city of Vienna with the aim to understand the effects of changes in dry/wet cycles on established floodplain sediments. In the light of restoration plans aiming at increased surface water exchange with the river main channel, the response of sediments to frequent alternations between desiccation and inundation periods is a key issue as changes of sediment properties are expected to affect phosphorus release. In order to determine the effect of changing hydrological conditions on internal phosphorus loading, we exposed sediments to different dry/wet treatments in a laboratory experiment. Total phosphorus (TP) release from sediments into the water column increased with increasing duration of dry periods prior to re-wetting. Partial correlation analysis showed significant positive correlations between ΔTP and ΔNH(4)(+) as well as between ΔTP and ΔFe(3+) concentrations (Δ refers to the difference between the final and initial concentration during the wetting period), indicating that enhanced mineralization rates leading to a concomitant release of NH(4)(+) and TP and the reduction of iron hydroxides leading to a concomitant release of Fe(3+) and TP are the mechanisms responsible for the rise in TP. Repeated drying and wetting resulted in elevated phosphorus release. This effect was more pronounced when drying periods led to an 80% reduction in water content, indicating that the degree of drying is a major determinant controlling phosphorus release upon re-wetting. The reconnection of isolated floodplains will favor fluctuating hydrologic conditions and is therefore expected to initially lead to high rates of phosphorus release from sediments.

Introduction

Since the 1950s nutrient loading has been a major threat to the stability of freshwater ecosystems worldwide, with an expected tendency to worsen in the future (Millenium Ecosystem Assessment, 2005). Dominant anthropogenic inputs of phosphorus include point source emissions from industrial and municipal waste water treatment plants and diffuse inflow from agriculturally used areas. Inputs of phosphorus into the Danube river system rose from 45 kt y− 1 in 1955 to 115 kt y− 1 in 1990 due to an increase in the use of detergents and an expansion of sewerage. From 1990 to 2000 phosphorus emissions dropped down to 68 kt y− 1 due to technical progress in waste water treatment and the replacement of phosphorus in detergents (Kroiss et al., 2005).

Phosphorus introduced into aquatic systems is either subject to short-term storage mediated by plants and algae or to long-term storage or permanent retention mediated by sediment deposition (including organic components) or adsorption and precipitation (Boers et al., 1998; Golterman, 2004; Reddy and DeLaune, 2008). Within the Danube river basin 53% of phosphorus emitted into the water is retained in small water bodies along the downstream transport to the Black Sea (Kroiss et al., 2005). River floodplain systems including riparian zones play a major role in nutrient retention, especially during high flows as shown for the Dutch Rhine by Van der Lee et al. (2004). For the availability of phosphorus in riparian zones, shallow water bodies and other wetland types less influenced by riverine dynamics, the process of internal loading from the sediments (i.e. vertical exchange across the sediment–water interface due to changes in redox conditions in different sediment layers) has the potential to play a major role depending on the flow path and the status of the sediment surface (Hoffmann et al., 2009; Søndergaard et al., 2003; Zak and Gelbrecht, 2007).

The frequency and pattern of connectivity between the main river and the water bodies of the floodplain determine whether external or internal processes control phosphorus dynamics. Lair et al. (2009) reported that sediments of floodplain areas with high connectivity levels to the main channel contain large amounts of inorganic phosphorus due to allochthonous inputs, whereas sediments of areas with lower surface water connectivity contain large amounts of organic phosphorus due to local autochthonous production. The retained amount of phosphorus in connected side-arms of the Danube is linked to the discharge pattern in the river (Hein et al., 2005). In the floodplains of the Atlantic coastal plain (USA) floodplain sites disconnected from the main river displayed lower accumulation rates regarding the sediment and nutrients (Noe and Hupp, 2005). Thus, floodplains cut off from surface water exchange with the main river are no longer able to act as systems of quantitative and wider scale importance for nutrient retention. The implementation of restoration projects with the aim to increase hydrologic connectivity could lead to a regain of this function (Naiman et al., 2005). During surface water connection phosphorus is introduced into the former isolated system directly altering equilibrium conditions for phosphorus exchange within the floodplain. In consequence, phosphorus release at the sediment–water interface (“internal loading”) in the floodplain is influenced by lateral exchange processes between the floodplain and the main river, as also shown for backwaters subject to a controlled water enhancement scheme reducing phosphorus remobilization from (anoxic) sediments (Bondar-Kunze et al., 2009).

While the main channel and deeper parts of floodplain backwaters remain permanently inundated, riparian zones and shallow areas are periodically falling dry and get re-wetted depending on their topography and the surface flow paths. Water level fluctuations and desiccation are expected to have a strong effect on internal phosphorus loading by altering redox conditions of sediments, pH, physical properties of minerals (Aldous et al., 2005; Baldwin, 1996; Baldwin and Mitchell, 2000; Darke et al., 1996; Lijklema, 1980; Sah et al., 1989) and the composition and activity of the microbial community (Batzer and Sharitz, 2006).

Adsorption and desorption of phosphorus from minerals is redox- and pH-sensitive: for instance, the phosphate binding capacity mediated by Fe minerals decreases with increasing pH value in a pH range between 5 and 8 due to the competition of hydroxyl ions for binding sites (Ahlgren, 2006; Lijklema, 1980; Reddy and DeLaune, 2008). At redox potentials below 300 mV or oxygen concentrations below 0.1 mg l− 1 ferric hydroxides are reduced to ferrous iron by bacteria leading to a release of phosphate that was occluded in the hydrated coatings of the ferric hydroxide (Golterman, 2004; Reddy and DeLaune, 2008). During desiccation the altered pore water chemistry and intensified degradation processes result in rising concentrations of phosphate in the pore water (Aldous et al., 2005; Golterman, 2004). During exposure to air, ferrous sulfides present in the sediment are oxidized to amorphous ferric (oxy)hydroxides having a large surface area and a very high affinity for phosphorus. However, prolonged periods of drought accelerate mineral aging resulting in a higher degree of cristallinity of ferric (oxy)hydroxides and a decrease in binding sites for phosphorus (Baldwin, 1996; Lijklema, 1980).

Desiccation does not only affect sediment and pore water chemistry but is also known to have a strong effect on the microbial community: bacterial activity decreases linearly with decreasing water content (Orchard et al., 1992; West et al., 1992) and obligate anaerobic bacteria (e.g. iron and sulfur reducing bacteria) are killed during extended drying or form resting stages (Baldwin and Mitchell, 2000; Lynch and Hobbie, 1988). The loss of obligate anaerobic microbes inhibits the reduction of ferric hydroxides and sulfate. Consequently, phosphate remains adsorbed onto ferric hydroxides.

In summary, the effects of drying and re-wetting of sediments on phosphorus dynamics can be ascribed to changes in the bacterial community as well as to changes in water chemistry, especially at the sediment water interface. While there is much evidence that hydrological changes have strong and long lasting impacts on the process of internal loading of phosphorus from different sediment types in wetlands (e.g. Aldous et al., 2005; Zak and Gelbrecht, 2007), for floodplain sediments the short term effects of changes in hydrology are not well known.

Therefore we conducted a laboratory experiment addressing the following questions:

How does the length of drying periods—leading to a significant reduction of the water content—affect the phosphorus release behavior of sediments?

How does repeated drying and wetting affect the phosphorus release behavior of sediments?

Thus, the objective of this study was to assess the impact of changing hydrological conditions on internal phosphorus loading by sediments of isolated backwaters. We expected phosphorus release to increase with increasing periods of drought as well as with repeated drying and wetting periods. In our laboratory experiment the effect of the lateral exchange between the floodplain and the main river on vertical exchange (“internal loading”) was considered by using water from the Danube River for the wetting of sediment cores taken from backwaters. The objective of this study was to assess the impact of changing hydrological conditions on internal phosphorus loading by sediments of isolated backwaters.

Material and methods

Study site

The floodplain of the Lower Lobau (48°10′N 016°30′E), stretching to an area of 915 ha, was disconnected from the main channel of the River Danube by flood protection measures by the end of the nineteenth century. Today, water supply is mainly mediated by groundwater inputs as the “Schönauer Schlitz” (interruption of dam, see Fig. 1) represents the only residual site providing limited surface water connection to the main river. Sedimentology of the water bodies of the Lower Lobau, which are isolated from each other by cross bars or fords, is determined by their distance from the “Schönauer Schlitz” reflecting decreasing hydrological connectivity: The thickness of the fine sediment layer decreases and the organic content of the sediment increases with increasing distance from the “Schönauer Schlitz” (Reckendorfer and Hein, 2006).

Fig. 1

Study area Lower Lobau and the sampling sites: “Hanslgrund” (H), “Schönauer Wasser” (S1, S2), Danube water used for wetting treatment (D). Surface water connection between the Danube River and the sampling sites is only possible when water flows back through the inlet (“Schönauer Schlitz”) interrupting the flood protection levee.

Although altered in many respects, the Lower Lobau still exhibits a high nature value and was thus declared a UNESCO biosphere reserve in 1977, designated as a “Wetland of International Importance” after the Ramsar Convention in 1982 and finally became part of the National Park Donau-Auen founded in 1996. Besides its importance as a nature conservation area, the Lower Lobau is subject to small-scale fishing, serves as a drinking water source for the city of Vienna during periods of high demand and is used for recreation (Hein et al., 2006a). In order to reduce siltation processes threatening the valuable water bodies of the Lower Lobau, plans to enhance surface water connectivity to the river Danube and to improve integration between currently separated basins have been developed and implemented in parts of the area (Managementplan Nationalpark Donau-Auen, 2009-2018, 2009). The implementation of a controlled surface water exchange program for the Lower Lobau—like already realized for the Upper Lobau—with the aim to increase the surface water supply from different sources (river main channel, upstream floodplain section) and thus, connectivity, to maintain habitat heterogeneity, to protect endangered species and also to maintain ecosystem services like drinking water supply and recreation (Baart et al., 2010; Hein et al., 2006b) is currently being tested.

For the collection of sediment samples to be used in our experiment, water bodies differing in surface water connection to the river Danube and in the organic content of sediments were chosen as these characteristics were assumed to be major determinants for phosphorus remobilization. Sediment samples were taken from two different shallow backwaters called “Schönauer Wasser” and “Hanslgrund” (H, Fig. 1, Table 1). Within “Schönauer Wasser”, two sites differing in organic content (S1, S2, Fig. 1, Table 2) were chosen. Surface water connection to the river Danube is prevailing on 137 da− 1 at backwater S and on 4 da− 1 at backwater H, respectively. Tritthart et al. (2011) calculated patterns of connectivity based on a hydrologic model for the investigated floodplain area: The average duration of a connection event was more than 30 consecutive days in the lower part of the floodplain (S1, S2) and less than 2 days in the upper section of the floodplain (H). The average duration of a period of disconnection was less than 10 consecutive days in the lower part of the floodplain (S1) and almost 90 days in the upper reach (H).

Table 1

Chemical characteristics of surface water in the study area (S, H, D) according to monitoring data from 2010 (median ± interquartile range, n = 9) and of water from the river Danube treated with anion exchange resins (DAE) used for the experiments (median ± interquartile range, n = 5).

	S	H	D	DAE
PO43 − [μg l− 1]	0.59 ± 0.44	17.79 ± 20.96	27.00 ± 17.00	6.10 ± 10.40
TP [μg l− 1]	27.11 ± 5.78	62.00 ± 25.11	54.00 ± 18.00	6.40 ± 8.70
NH4+ [μg l− 1]	7.51 ± 9.60	13.69 ± 30.91	18.50 ± 7.25	15.40 ± 7.00
NO2− [μg l− 1]	0.58 ± 1.53	1.04 ± 0.51	10.00 ± 4.00	4.80 ± 0.40
NO3− [μg l− 1]	93.75 ± 53.70	130.11 ± 28.01	1780.00 ± 570.00	769.5 ± 267.0
Fe3 +[μg l− 1]	–	–	–	37 ± 29.75
pH	8.30 ± 0.30	8.00 ± 0.29	8.10 ± 0.10	9.65 ± 0.37

Table 2

Mean sediment composition of the three sampling sites regarding organic content, amorphic iron, and different phosphorus fractions (mean ± standard deviation, n = 12).

	S1	S2	H
Organic content [%]	10.1 ± 3.6	7.0 ± 0.8	19.0 ± 5.7
Amorphic iron [g kg− 1]	3.97 ± 0.93	3.60 ± 0.41	1.66 ± 0.27
TP [mg kg− 1]	543.1 ± 94.9	548.3 ± 31.6	580.8 ± 70.3
OP [mg kg− 1]	159.1 ± 33.8	155.4 ± 49.5	196.5 ± 71.3
IP [mg kg− 1]	391.5 ± 18.0	385.6 ± 17.9	389.3 ± 19.3
Water soluble P [mg kg− 1]	0.8 ± 0.3	0.9 ± 0.3	2.7 ± 1.6
Mn- and Fe- bound P [mg kg− 1]	50.7 ± 19.9	74.7 ± 13.0	50.8 ± 17.0
Ca- and Mg- bound P [mg kg− 1]	354.3 ± 23.8	343.7 ± 17.7	371.2 ± 20.0

Siltation processes at backwater S are mainly generated by allochthonous input from the river Danube. Siltation processes at backwaters farther upstream like backwater H are the result of autochthonic production (Hein et al., 2006a). All sampling sites are exposed to drought for several weeks during summer/autumn at low flow periods.

Field sampling

Sampling of sediment cores was carried out in July 2010, when all sampling sites were water saturated. Per sampling site, twelve sediment cores with a diameter of 15 cm, a length of 10 cm and a fresh weight of approximately 900 g were taken using a PVC tube. The sediment cores were transferred into plastic vessels without disturbing the surface layer. During the two-hour transport to the laboratory the sediment samples were kept in the dark, water saturated and at a constant temperature. Water from the river Danube (D, Fig. 1) was collected 30 cm below the surface.

Analyses of sediment samples

From each of the 36 sediment cores, 150 g subsamples for analysis of organic content, amorphous iron and phosphorus fractions were taken from the edge of each core to minimize any disturbance of the upper sediment layer immediately after field sampling and kept deep-frozen for 4 weeks. Organic content was determined by the loss on ignition method (450 °C, 4 h). Amorphous iron was extracted by adding 30 ml of a solution containing 16.2 g of (COONH4)2·H2O and 10.9 g of (COOH2)·2H2O l− 1 to 2–2.5 g of sediment (Loeb et al., 2008; Schwertmann, 1964). The iron concentration was determined according to US Standard Methods 3500-Fe D and DIN 38406-E1-1 using a Hitachi U-2000 photometer. Inorganic phosphorus (IP), organic phosphorus (OP), water soluble phosphorus, Fe- and Mn-bound phosphorus, Al-bound phosphorus and Ca- and Mg-bound phosphorus were determined according to the protocol of Ruban et al. (2001). Each fraction was extracted separately and at least 3 g of dry sediment for each extraction was used. The sediment with reagent was shaken for 16 h and centrifuged at 3000 rpm for 15 min at 20 °C. Total phosphorus (TP) was extracted by adding 10 ml of HNO3 (65%) to 0.3 g–0.4 g of sediment and putting it into a microwave (CEM MarsXpress, 10 min heating-up to 175 °C). Afterwards, phosphorus concentrations were determined by continuous flow analysis (CFA) by a Systea 3rd Generation Continuous Flow Analyser (Alliance instruments) according to DIN EN ISO 15681-2 (Eaton and Franson, 2005).

Experimental setup

During the experiment, the sediment cores were kept in two-liter plastic vessels and incubated in the dark at a temperature of 30 °C in the drying cabinet. Three sediment cores per sampling site were randomly assigned to one out of four dry/wet cycles (Fig. 2).

Fig. 2

Scheme displaying four different dry/wet cycle patterns (A, B, C, D) simulated in the drying cabinet at a constant temperature of 30 °C. Wetting periods are consecutively numbered within each dry/wet cycle pattern.

The first wetting periods of dry/wet cycles A, B, C and D were simulated after 100 h, 200 h, 300 h and 400 h of drying. This setup was chosen in order to evaluate the effect of different degrees of drying on phosphorus release upon re-wetting. In order to find out if repeated drying and wetting phases have an impact on phosphorus release, multiple wetting periods were simulated within dry/wet cycles A and B: Dry wet/cycle A was composed of three drying periods of 100 h each, alternating with three wetting periods of 100 h each. Dry wet/cycle B was composed of two drying periods of 200 h each, alternating with two wetting periods of 100 h each.

The duration of wetting periods was set to 100 h because in experiments carried out by Turner and Haygarth (2001) maximal phosphorus release attributed to lysis of bacterial cells in soils dried at 30 °C for 7 days occurred within the first 3 days of re-wetting. Furthermore, according to Reddy and DeLaune (2008), the reduction of iron and concomitant phosphate release starts after the reduction of the entire nitrate pool which was observed to be completed 3 days after inundation in soils of the Lake Apopka Marsh. We assume this process to be similar in floodplain ecosystems as a negative correlation between the concentration of nitrate and the point of time at which phosphorus release started was observed for river riparian sediments by Surridge et al. (2007). For wetting, water from the Danube River was used. The water was filtered using double Whatman GF/F filters and stored at 4 °C. Two days before its addition to the sediment cores, the water was treated with anion exchange resins (DOWEX Marathon A; 200 g per 20 liters of water) in order to reduce concentrations of NO3−, NO2− and PO43 − by approximately 50% and to thereby render the effect of phosphorus release from the sediment more apparent (Table 1). Sediment cores were filled with 1 l of treated river water.

Wetting periods were terminated by withdrawing the water from the plastic vessel with a peristaltic pump. After 100 h of wetting, surface water samples for the determination of TP, PO43 −, NO3−, NO2− and NH4+, and Fe3 + were taken and filtered through Whatman GF/F filters. Unfiltered samples were used for TP determination. Prior to being used for the experiments the water was analyzed for the parameters mentioned above.

Analyses of water samples

Nutrient concentrations were determined according to Eaton and Franson (2005), complying with ÖNORM EN ISO 15681-2 (PO43 −), DIN EN ISO (NO3− and NO2−) and DIN EN ISO 11732 (NH4+) by continuous flow analysis using a Systea Alliance instrument.

Total P was digested with persulfate according to Eaton and Franson (2005) and subsequently analyzed following the same procedure as for PO43 −. Concentrations of Fe3 + were determined according to DIN 38406 E1 (US-Standard Methods 3500-Fe D) using Lange LCK 320 cuvette test and Hach DR-2800 portable photometer.

Statistical analysis

Statistical analysis was conducted using SPSS 16.0 for Windows. In the analyses, the differences between the concentrations of TP, PO43 −, NO3−, NO2−, NH4+, and Fe3 + in the water column after100 h of wetting of the sediment cores and of the water used for the initiation of the wetting period were calculated (the prefix Δ marks that this difference was used).

Kruskal–Wallis tests were performed in order to find out, if sampling sites differed regarding their organic content and phosphorus fractions. When the Kruskal–Wallis test showed significant results, a Mann–Whitney test was carried out for pairwise comparisons of the three sampling sites.

For statistical evaluation of the impact of drying prior to re-wetting, data from the first wetting periods of dry/wet cycle patterns A, B, C and D (Fig. 2) were analyzed. First, a Kruskal–Wallis test was performed in order to determine if the dry treatment prior to re-wetting and/or the sampling site affected the release or uptake of TP, PO43 −, Fe3 +, NH4+, NO3− and NO2− upon re-wetting. As the factor “sampling site” did not have a significant effect on the change in TP, NH4+ and Fe concentrations, further statistical analysis was conducted without splitting data according to sampling sites. Mann–Whitney test was carried out for the pairwise comparison of dry treatments. Partial correlation considering the dry treatment and the sampling site as covariates was performed in order to evaluate relationships between TP and NH4+ as well as between TP and Fe3 +. In order to statistically evaluate the effect of repeated drying and wetting, data from multiple wetting periods simulated within dry/wet cycle patterns A and B (Fig. 2) were analyzed.

For dry/wet cycle A—including three different wetting events—a Friedman test was performed for comparisons regarding TP, NH4+, and Fe3 +. When the Friedman test resulted in significant differences, a Wilcoxon test for pairwise comparisons was conducted in order to evaluate which of the three flooding events differed from each other.

For the dry/wet cycle B—including two different wetting periods—a Wilcoxon test was performed for testing differences regarding TP, NH4+ and Fe.

Results

Sediment characteristics

Sediment samples of the three sites did not differ regarding TP (Kruskal–Wallis test, n = 45, χ2 = 3.366, df = 2, p = 0.186), OP (Kruskal–Wallis test, n = 45, χ2 = 3.913, df = 2, p = 0.141) and IP (Kruskal–Wallis test, n = 45, χ2 = 1.157, df = 2, p = 0.561).

IP accounted for 60% of TP in “Schönauer Wasser 1” (S1) and “Schönauer Wasser 2” (S2) and for 50% in “Hanslgrund” (H) (Table 2).

The three sampling sites significantly differed from each other in organic content, with the highest values in H (Tables 2 and 3). Sampling site S2 exhibited significantly higher contents of Mn- and Fe-bound P than the sampling sites S1 and H in (Tables 2 and 3). Sampling site H exhibited significantly higher contents of amorphic iron and Ca- and Mg-bound P and significantly lower contents of water soluble phosphorus than sampling sites S1 and S2 (Tables 2 and 3).

Table 3

Results of Mann–Whitney test, pairwisely comparing sediment composition of the three sampling sites (n = 12). TP, OP, and IP are not included as Kruskal–Wallis test did not show any significant differences among sampling sites for these phosphorus fractions.

Pairwise comparison of	Statistical parameters	Organic content	Amorphic iron	Water soluble P	Mn- and Fe-bound P	Ca- and Mg-bound P
S1–S2	Whitney U	40.5	90	84	41	73
	p	0.003	0.351	0.237	0.003	0.101
S1-H	Whitney U	28	0	25	109	64
	p	0.0005	0.000003	0.0003	0.885	0.044
S2-H	Whitney U	21.5	0	28	31	34
	p	0.0002	0.000003	0.0005	0.001	0.001

Impact of drying periods on the composition of the overlying water

The impact of different degrees of drying on nutrient release upon re-wetting was evaluated by comparing wetting periods A1, B1, C1, and D1 (Fig. 2) that occurred after 100 h, 200 h, 300 h, and 400 h of drought, respectively. The drying periods corresponded to the gravimetric water content of the sediment ranging from 30% after 100 h to 0% after 400 h of drying (Table 4). The changes in the concentrations of TP, Fe3 +, NH4+, NO3−, and NO2− in the water column upon re-wetting were highly significantly influenced by the duration of the preceding drying period (Table 5), whereas the sampling site did not have any significant effect (Table 6). In consequence, further analysis regarding ΔTP, ΔFe3 +, ΔNH4+, ΔNO3−, and ΔNO2− was conducted by pooling together data from all sampling sites.

Table 4

Mean and standard deviation of gravimetric water content of sediment after different periods of drying (n = 9).

Period of drying [h]	Gravimetric water content [%]
0	51.81 ± 1.30
100	29.44 ± 3.49
200	11.15 ± 3.54
300	3.54 ± 2.43
400	0

Table 5

Kruskal–Wallis test with the “dry treatment” (100 h/200 h/300 h/400 h of drying, see Fig. 2) as a group variable (n = 36).

	ΔTP	ΔPO43 −	ΔFe3 +	ΔNH4+	ΔNO3−	ΔNO2−
χ2	27.789	4.216	25.063	29.222	30.611	11.537
df	3	3	3	3	3	3
p	0.00004	0.239	0.000015	0.000002	0.000001	0.009

Table 6

Kruskal–Wallis test with the “sampling site” (S1, S2, H, see Fig. 1) as a group variable (n = 36).

	ΔTP	ΔPO43 −	ΔFe	ΔNH4+	ΔNO3−	ΔNO2−
χ2	2.252	13.781	2.934	0.221	0.011	2.254
df	2	2	2	2	2	2
p	0.324	0.0001	0.231	0.896	0.994	0.324

In contrast to other nutrients measured in the water column, the concentration of ΔPO43 − significantly differed among sampling sites (Table 6) with H showing significantly higher concentrations compared to S1 and S2. As ΔPO43 − did not show any significant difference among dry treatments (Table 5) and as this work focuses on the effect of dry/wet cycles on nutrient release, ΔPO43 − was not considered in further statistical analysis.

The longer the previous drying period lasted, the higher were ΔTP, ΔFe3 +, and ΔNH4+ during re-wetting (Fig. 3). Significant differences between each of the drying treatments were observed regarding ΔTP, ΔFe3 +, and ΔNH4+ (Table 7). Partial correlation analysis considering the dry treatment and the sampling site as covariates in order to avoid a spurious correlation showed strong and highly significant positive correlations between ΔTP and ΔNH4+ (r = 0.744, df = 31, p = 0.000001) as well as between ΔTP and ΔFe3 + (r = 0.441, df = 31, p = 0.01) (Fig. 4).

Fig. 3

Boxplots displaying ΔTP (a), ΔNH4+ (b), and Δ Fe3 + (c) in the water column after 100 h of wetting for four different drying treatments: A1 = 100 h, B1 = 200 h, C1 = 300 h, D1 = 400 h of drying at 30 °C previous to wetting (n = 9). Δ refers to the difference between the concentration at the end and the start of the wetting period. Data of all sampling sites were pooled (n = 9).

Table 7

Mann–Whitney test for pairwise comparison of drying treatments (A1 = 100 h, B1 = 200 h, C1 = 300 h, D1 = 400 h of drying at 30 °C prior to re-wetting). Data of all sampling sites are pooled (n = 9).

		A1–B1	A1–C1	A1–D1	B1–C1	B1–D1	C1–D1
ΔTP	Whitney U	7	0	0	6	0	13
	p	0.005	0.001	0.001	0.002	0.004	0.015
ΔFe3 +	Whitney U	24	0	0	9	3	16
	p	0.144	0.0003	0.0003	0.005	0.004	0.031
ΔNH4+	Whitney U	8	0	0	7	39	6
	p	0.005	0.0004	0.0004	0.003	0.0000001	0.002

Fig. 4

Scatterplots showing the relationship between ΔTP and ΔNH4+ (a) and the relationship between ΔTP and ΔFe3 + (b) upon re-wetting. Three different sampling sites and four different dry treatments are included. Highly significant positive correlations were observed for (a) r = 0.744, df = 31, p = 0.000001 and (b) r = 0.451, df = 32, p = 0.007 (partial Spearman correlation, covariates = dry treatment, sampling site).

Impact of repeated drying and wetting cycles

Within the dry/wet cycle pattern A (Fig. 2), including drying periods of 100 h alternating with wetting periods of 100 h, ΔTP differed significantly between the first and the third wetting event (Wilcoxon, n = 9, Z = − 0.169, p = 0.018) as well as between the second and the third wetting event (Wilcoxon, n = 9, Z = − 0.169, p = 0.018) (Fig. 5). No significant differences regarding ΔTP were observed between the first and the second wetting event (Wilcoxon, n = 9, Z = − 0.169, p = 0.866). ΔTP showed a five-fold increase in the third wetting event (mean = 57.20 μg l− 1) compared to the two preceding ones (means = 11.59 μg l− 1, 10.58 μg l− 1, respectively). The change in the concentration of Fe3 + differed significantly among wetting events of dry/wet cycle A (Friedman, n = 9, χ2 = 6.889, df = 2, p = 0.032), but pairwise comparison of wetting events by Wilcoxon did not show any significant results. The change in NH4+ did not differ significantly among wetting events of the dry/wet cycle A (Friedman, n = 9, χ2 = 0.750, df = 2, p = 0.687).

Fig. 5

Boxplots showing ΔTP (1), Δ NH4 (2), and ΔFe3 + (3) after 100 h of wetting for different wetting events of dry/wet cycle A (a) and B (b). Data of all sampling sites are pooled (n = 9).

Within the dry/wet cycle pattern B (Fig. 2), including drying periods of 200 h alternating with wetting periods of 100 h, wetting periods significantly differed regarding ΔTP (Wilcoxon, n = 9, Z = − 2.192, p = 0.028) as well as ΔNH4+ (Wilcoxon, n = 9, Z = − 2.192, p = 0.028) (Fig. 5). The means of both variables showed a four-fold increase in the second wetting event (ΔTP = 352. 09 μg l− 1, ΔNH4+ = 10022.46 μg l− 1) compared to the first one (ΔTP = 80.01 μg l− 1, ΔNH4+ = 2640 μg l− 1). No significant differences among the wetting events of dry/wet cycle B were observed regarding ΔFe3 + (Wilcoxon, n = 9, Z = − 1.362, p = 0.173).

Discussion

The impact of drying on nutrient release upon re-wetting was evaluated by comparing changes in concentrations of TP, NH4+ and Fe3 + in the overlying water after 100 h of re-wetting among treatments differing in the preceding drying period (A1, B1, C1, D1; see Fig. 2). The duration of wetting periods was set to 100 h because in experiments carried out by Turner and Haygarth (2001) maximal phosphorus release attributed to lysis of bacterial cells in soils dried at 30 °C for 7 days occurred within the first 3 days of re-wetting. Furthermore, according to Reddy and DeLaune (2008), the reduction of iron and concomitant phosphate release starts after the reduction of the entire nitrate pool which was observed to be completed 3 days after inundation in soils of the Lake Apopka Marsh. We assume this process to be similar in floodplain ecosystems as a negative correlation between the concentration of nitrate and the point of time at which phosphorus release started was observed for river riparian sediments by Surridge et al. (2007).

In our experiments, the release of TP, NH4+, and Fe3 + to the water column after 100 h of re-wetting increased with increasing duration of the preceding drying periods (Fig. 3). The release during the wet phase of the experiment has been included and has not been determined separately. Drying periods differing in length corresponded to varying gravimetric water content of the sediment ranging from 30% after 100 h to 0% after 400 h of drying (Table 4).

Phosphorus release mediated by mineralization of organic matter

The relationship between NH4+ and TP was characterized by a positive, highly significant correlation (Fig. 3), indicating that mineralization processes could be an important mechanism responsible for the rise in TP release. Due to our sampling intervals, a rise in PO43 − could not be measured presumably because of rapid uptake by microbes and benthic algae. Therefore PO43 − could only be detected as part of the total phosphorus pool.

Why did mineralization rates increase with increasing periods of drought? In a laboratory experiment, Turner et al. (2003) dried pasture soils for 7 days at 30 °C and found out that 95% of molybdate unreactive phosphorus (i.e. organic phosphorus and inorganic polyphosphates) released upon re-wetting could be related to the release of phospholipids and nucleic acids due to lysis of bacterial cells during the drying period or due to osmotic shock upon re-wetting. According to Turner and Haygarth (2001), 56–100% of phosphorus released after flooding of previously dried grassland soils (7 days, 30 °C) were part of the organic phosphorus pool. Furthermore, Turner and Haygarth (2001) found a positive correlation between water-soluble organic phosphorus and microbial phosphorus. Qiu and McComb (1995) estimated that air-drying of lake sediment killed 76% of the microbial biomass, resulting in a five-fold increase in dissolved phosphorus concentration upon flooding. Upon re-wetting, previously dried soils are characterized by a high activity of the newly establishing community of microbes (own unpublished data) and a high availability of fresh dead organic matter. These conditions are favoring high mineralization rates and in consequence the release of ammonium (Birch, 1960), as also shown for rewetted fens (Zak and Gelbrecht, 2007). Another cause for ammonium accumulation in sediments upon drying could be the fact that nitrification of ammonium is no longer taking place as nitrifying bacteria are much more susceptible to dehydration than ammonium producing bacteria (Blume et al., 2010). Taking these studies into account, gradual increases in TP and NH4+ release with increasing drying periods could be the result of the increasing numbers of lysed microbial cells serving as a substrate for those members of the microbial community that survived the stress of drying. Decreasing oxygen concentrations in the water column with increasing periods of drought prior to wetting might support this hypothesis as oxygen depletion is likely to occur when aerobic heterotrophic organisms show high metabolic activities and therefore have a high oxygen demand. The formation of anaerobic micro-areas in the sediment under low oxygen conditions in the water column could be another factor triggering NH4+ release from sediments as oxidation of NH4+ to NO3− (nitrification) is reduced under anaerobic conditions, resulting in the release of NH4+ following decomposition of organic matter (Naiman et al., 2005). Our results suggest that desiccated sediments of former isolated backwaters are acting as sources of nutrient release during the first phases of flooding events.

Phosphorus release mediated by reduction of iron hydroxides

According to Lijklema (1980), 50% of the phosphate binding capacity of sediments of shallow lakes and reservoirs in the Netherlands is mediated by amorphous iron species, exhibiting a specific surface area of 200–600 m2 g− 1 (Blume et al., 2010). Depletion of oxygen in the water column with increasing drying periods coincides with low redox potentials, favoring the concomitant release of phosphorus and Fe2 + from sediments due to the reduction of iron(III)hydroxides (Zak et al., 2004). Ferric iron is reduced in contact with bacterial enzymes and not indirectly in the presence of reducing metabolites at low redox potentials (Munch and Ottow, 1983). Amorphous iron(III) hydroxides are the only species of iron accessible to microbial reduction (Lovley and Phillips, 1986, 1987; Munch and Ottow, 1983). In our experiments concentrations of Fe3 + and TP in the water column increased with increasing periods of drought prior to re-wetting. The relationship between Fe3 + and TP was characterized by a significant positive correlation. We use Fe3 + as a measure for iron release because Fe2 + released from anoxic sediments is subject to rapid re-oxidation in the aerated water column and therefore largely present as Fe3 +. This was also confirmed by our experiment as the concentration of Fe2 + in the water column was nearly constant, whereas the concentration of Fe3 + varied among treatment groups. The reduction of iron(III) hydroxides takes place when redox potentials fall below 300 mV (Golterman, 2004). In the present study we did not measure redox potentials, but decreasing oxygen concentrations in the water column with increasing duration of drought periods co-occurred with elevated concentrations of Fe3 + in the water column.

Phosphorus release due to loss of phosphorus sorption capacity

Drying events lead to an increase in crystallinity of iron hydroxides and therefore result in a loss of the phosphorus binding capacity of sediments (McLaughlin et al., 1981). In agreement with this, Darke and Walbridge (2000) showed that iron oxides and hydroxides in high elevation sites of the Ogeechee river floodplain in the United States exhibited higher crystallinity than low elevation sites that are exposed to dry conditions for shorter time periods. Submerged sediments of a water storage reservoir in New South Wales (Australia) exhibited higher ratios of amorphous to crystalline iron and therefore higher phosphate adsorption capacities than wet sediments from the littoral zone and dry sediments from sites exposed to air (Baldwin, 1996). In a laboratory experiment conducted by Qiu and McComb (1994), drying of the lake sediment for 40 days at 20 °C resulted in an elevated phosphorus loading into the water column, which was also related to a loss of the sorption capacity of the sediment due to increased crystallinity of iron oxides and hydroxides. In another study Qiu and McComb (2002) exposed sediments of seven shallow lakes in Western Australia to air-drying and observed that 72% of the variation in phosphorus sorption capacity could be explained by the variation of iron crystallinity upon drying. Increasing TP release into the water column with more extended drying in our experiment could therefore be attributed to a gradual increase in the crystallinity of iron species. The loss of the sorption capacity is expected to particularly prevent the adsorption of additional phosphate released from lysed microbial cells or produced by mineralization of organic matter.

Within the dry/wet cycle pattern A (Fig. 2), including drying periods of 100 h alternating with wetting periods of 100 h, TP release was low compared to that of other treatment groups (Fig. 5). We assume that the dry/wet cycle A did not subject the microbial community to severe drying stress as gravimetric water content within 100 h of drying decreased only slightly from about 50% to 30% (Table 4). Therefore, within the dry/wet cycle A, we did not expect large inputs of TP and NH4+ due to the reduction of microbial biomass by dehydration. The slight increase in ΔTP in the third wetting period compared to the first and second wetting periods could be the result of an elevated number of lysed cells due to the fast alternation of dry and wet conditions. This is supported by a slight increase in ΔNH4+ and ΔFe3 +, which can be ascribed to a slightly enhanced mineralization activity and oxygen consumption by the surviving microbial community.

Within the second wetting period of the dry/wet cycle pattern B (Fig. 2), including drying periods of 200 h alternating with wetting periods of 100 h, ΔTP, ΔNH4+, and ΔFe3 + reached levels similar to that of sediments wetted after a drying period of 300 h (Figs. 3 and 5). We assume that dry/wet cycle B has caused severe drying stress to microbes as gravimetric water content decreased from about 50% to 10% within a drying period of 200 h (Table 4). In total, microbes were subjected to two drying periods lasting for 200 h each, interrupted by two wetting periods of 100 h each. Within a drying period of 300 h gravimetric water content decreased from 50% to nearly 0% (Table 4). Therefore, the intensity of drying stress was higher prior to re-wetting within the dry/wet cycle C, but was acting for longer time periods within the dry/wet cycle B prior to the second wetting period. In addition to the negative effects of extended drying, microbes had to deal with fast changes between conditions of severe dryness and re-wetting within the dry/wet cycle B. Similar pH values and O2 concentrations between wetting periods C1 and B2 further underline the fact that biota were affected in a similar manner by these different treatments.

In summary, repeated drying and wetting resulted in elevated phosphorus release. This effect was more pronounced when drying periods lasted for 200 h than when they lasted 100 h, indicating that the degree of drying is a major determinant controlling phosphorus release upon re-wetting.

Various management options aiming to increase the water area in the floodplain Lower Lobau and to intensify the surface water exchange are currently discussed. All of these measures will lead to changes in the frequency and amount of water level fluctuations affecting riparian areas including the water saturation of riparian soils. Thus, these riparian areas are going to face faster successions between periods of desiccation and inundation. In the light of the present results, restoration efforts leading to intensified surface water exchange are assumed to lead to enhanced phosphorus loading from sediments during the first phase in the floodplain, which could stimulate primary production and favor increased algal biomass. Sediments of the Lower Lobau are expected to be an important source of phosphorus, when the floodplain area is subjected to infrequent drying and re-wetting events compared to longer and frequent wet phases, largely independent of the quality of the source water entering the floodplain.

Longer phases of connection and continuous connectivity following restoration efforts are known to lead to a shift from an isolated floodplain dominated by internal nutrient cycling (autochthonous production) to a connected floodplain characterized by frequent and intense exchange of nutrients with the main river (Hein et al., 2004; Felkl, 2011; Lair et al., 2009). Therefore, considering long-term effects, a reconnection of the Lower Lobau is not expected to trigger eutrophication processes, but to enforce the export of developed algal biomass into the river system and thereby preventing the accumulation of sediments rich in organic components that represent a major source of phosphorus release. According to Lair et al. (2009) sediments of floodplain areas with high connectivity levels to the main channel contain large amounts of inorganic phosphorus due to allochthonous inputs. Furthermore, Baldwin and Mitchell (2000) suggested that the repetition of drying and wetting over longer time periods will select bacterial r-strategists that reproduce fast when favorable environmental conditions occur and can outlive unfavorable conditions by producing resting stages. Taking our results into account, this could lead to new equilibrium conditions between the water column and sediment surface as dead bacterial biomass was suggested to be the main driving factor responsible for elevated mineralization rates leading to oxygen depletion and consequential phosphorus release.

Conclusion

Drying events led to an enhanced phosphorus release upon re-wetting, which could be related to enhanced mineralization rates, enhanced reduction of iron hydroxides and a general loss in sorption capacity due to increased crystallinity of iron hydroxides.

The magnitude of internal phosphorus loading upon re-wetting was shown to rise with the degree of sediment dehydration: the longer the intermittent drying period lasted, the higher was the rise in phosphorus release compared to previous wetting events.

Fast alternations between periods of desiccation and inundation are assumed to cause severe stress to the microbial community, if the gravimetric water content of the sediment falls below 10% during the drying phase and thereby triggering internal phosphorus loading.

The reconnection of former isolated floodplains is going to favor fluctuating hydrologic conditions and is therefore expected to initially lead to high rates of phosphorus release from sediments.

Considering longer time periods and a continuous surface water exchange, restoration of hydrological dynamics is expected to inhibit any massive eutrophication processes in reconnected sites by enabling the export of material and algae from the floodplain to the river and thereby inhibiting long-lasting deposition of large amounts of organic matter that serve as an internal nutrient source for aquatic primary production.