SEARCH

SEARCH BY CITATION

Keywords:

  • Disturbance;
  • flooding regime;
  • submergence;
  • survival;
  • floodplain zonation;
  • suspended load;
  • light attenuation

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  • 1
    Summer floods whose severity is affected by flooding duration, submergence depth and underwater light availability, have a large impact on the zonation of riparian plant species.
  • 2
    We analysed the range and variability of these flooding components in the River Rhine and quantified their effects on the ability of Arrhenatherum elatius, Achillea millefolium, Rumex acetosa and Rumex crispus, to survive periods of submergence under experimental conditions.
  • 3
    Survival characteristics were used to model species’ lower distribution boundaries for extreme and average floods and were compared with the current field distribution. Different light conditions were simulated by implementing three scenarios of suspended load.
  • 4
    Extreme deep Rhine floods are characterized by very low median light transmission levels (i.e. below 0.5%). The largest survival responses in the experiment were observed at such low levels (0.4–3.5 µmol m−2 s−1). Strong effects of light were found in R. crispus and A. millefolium, but responses were weaker in A. elatius and R. acetosa. Submergence depth also affected survival, but not as strongly as light.
  • 5
    For the flood intolerant species (A. millefolium and A. elatius) the average flood was predicted to have little effect on field distributions under normal light conditions. However, their actual field distributions in 2000 corresponded to the predicted lower boundaries in the extreme years. This suggests that extreme years determine the distributions of these species for many years. The suspended load scenarios significantly modified the predicted lower boundaries in both extreme and average years, implying that plant lower distribution limits may be significantly shifted upwards or downwards depending on the suspended load of the river system.
  • 6
    The predicted lower boundaries of the intermediately tolerant R. acetosa and the highly tolerant R. crispus for both extreme and average years were below the actual field distribution in 2000. This suggests that their current distribution is only partly influenced by major flood disturbances and that other factors, either proximate or historical, may play a prominent role.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Disturbances caused by occasional extreme events, e.g. fire, storm, drought or flooding, are an important determinant of the distribution patterns of organisms, and a main element of ecological theory (Grime 1979; Southwood 1988; Lytle 2001). In habitats subjected to disturbance, the distribution of species is determined both by survival of given disturbance events and succession in the periods between such events, as influenced by dispersal, recruitment and competition. The direct responses of species to a disturbance are crucial, because surviving organisms not only determine the short-term post-disturbance patterns, but also dictate much of the successional trajectory between disturbances (Turner et al. 1998).

Disturbance regimes are defined by the frequency and intensity of disturbance. Because a disturbance of a given intensity may have severe effects on some species while affecting others to a much lesser degree, studying disturbance regimes requires consideration of disturbance intensity in relation to individual species responses (Turner et al. 1998). Our study focuses on hydrological factors that determine the intensity of summer floods in relation to plant survival and distribution in river floodplains. Extreme floods during the growing season, such as the 1993 Upper Mississippi flood (Sparks et al. 1998) and the Rhine summer floods in the 1970s and 1980s (Brock et al. 1987; Van de Steeg 1984), result in an upward deflection of the lower distribution boundaries of plant species, and may even eliminate species from the floodplain. Timing and duration of a flooding event are considered the main components for the survival of flooding (Hosner 1960; Gill 1970; Klimesova 1994; Toner & Keddy 1997). Extreme floods often result in complete submergence and underwater light availability then becomes an additional factor, determined by the load of suspended sediment and submergence depth. Experiments have demonstrated that the degree to which riparian plants survive a given period of submergence is determined by light intensity (Adkins et al. 1990; Blom et al. 1994; Siebel et al. 1998; He et al. 1999; Nabben et al. 1999). Depth of the water column may affect survival via underwater light availability, but might also affect plant survival independently from light, through hydrostatic pressure (Adkins et al. 1990).

This information suggests that underwater light availability and depth of submergence may play an important role in the survival of periods of submergence. However, little is known of their significance in riverine environments. For example, it is as yet unclear to what extent the lower distribution limits of riparian plant species depend on the turbidity of the water during flood, compared to other components such as duration. As light conditions may affect the underwater photosynthesis of some riparian species more than others (Vervuren et al. 1999), light may also affect species rankings of survival and zonation. This calls for a quantification of plant survival responses in relation to light and depth, and detailed information on these factors during periods of flooding to allow extrapolation to field conditions. The objectives of this study were therefore: (i) to analyse the range and variability of the flooding components (duration, depth and underwater light availability) of a natural river regime; (ii) to quantify the effects of underwater light availability and submergence depth on the survival of plant species under experimental conditions; and (iii) to assess the importance of these components by modelling the lower distribution boundaries of the species under study in relation to the conditions during historical extreme floods and under different scenarios of suspended load.

We used species and hydrological data from the River Rhine. Flooding duration and depth were studied using daily water levels of 1950–99. An empirical light attenuation model was constructed to determine underwater light availability as a function of suspended load and depth. Four perennial riparian plants species with different flooding tolerances were selected for the submergence experiments: a grass, Arrhenatherum elatius (L.) J. & Presl., an Umbellifer, Achillea millefolium L., and two dock species Rumex acetosa L. and Rumex crispus L. Lower distribution boundaries were predicted on the basis of survival in the experiment, using a lower boundary model, and compared to the distributions of the species in the field.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

flooding data

Rhine discharge regime and the flooding gradient

The River Rhine (length 1320 km; drainage area 185 000 km2) originates in the Swiss Alps, flows north along the border with France, and then passes through Germany and the Netherlands, before discharging into the North Sea. At Lobith, near the Dutch–German border, the daily river discharge for the period 1950–99 varied between 665 m3 s−1 and 11885 m3 s−1 (mean: 2200 m3 s−1). Peak discharges in lower sections of the Rhine are caused by excess rainfall in the mid-section of the basin, which occurs predominantly in winter (Friedrich & Müller 1984).

At Lobith, riparian plants occur predominantly in the area between the mean river water level, c. 10 m (for the period 1950–99), and the top of the winter dyke, c. 16 m (all elevations refer to sea level). Throughout this paper, two elevations, 11 m and 13 m, are used as reference levels. The low elevation, 11 m, is the approximate lower boundary of perennial, non-pioneer plant growth. The river bank below 11 m is characterized by an open vegetation dominated by softwood tree species and summer annuals. The mid-level elevation of 13 m roughly represents the lower boundary of high-floodplain grassland vegetation, characterized by flood-sensitive species (see Sykora et al. 1988; Blom et al. 1996; Van de Steeg & Blom 1998).

Duration of summer-floods

The water level data consisted of daily records of the Rhine main channel at the Lobith gauging station for the period 1950–99 (source: Waterdata-desk, Rijkswaterstaat, Lelystad). Over this period of time, water level statistics such as the yearly mean showed no obvious trends. Flooding within the period 1 May−31 September was classified as summer flooding. Duration of flooding was determined by counting the number of days between the first and the last days on which the water level exceeded a given elevation (i.e. flood duration was slightly underestimated because flooding extended up to a day on either side of these measurements). All calculations were carried out using sas software (SAS 1989a).

Underwater light profiles and suspended load

Underwater light profiles were measured in order to construct a light attenuation model. To cover a wide range of suspended loads, measurements were carried out both in the River Rhine and in handmade suspensions in the laboratory. Light was measured as photosynthetic photon flux density (PPFD) using a cosine-corrected underwater quantum sensor (LI-COR, model LI-192SB; Lincoln, Nebraska, USA) connected to a quantum-photometer (LI-COR, model LI-185B; Lincoln, Nebraska, USA). Directly after measuring a light profile, a water sample was taken for determination of suspended material. The sample was filtered over Whatman filter paper, and dried to constant weight at 105 °C.

For the laboratory measurements, fine sediment (silt and clay) was obtained from a sediment trap, which was part of a set-up for washing soil from roots. Variable amounts of sediment were suspended in a basin with tap water, creating a range of suspended loads. PPFD was measured every 5 cm to a depth of 0.5 m and carried out directly after the sediment had been suspended to avoid a decrease in turbidity due to precipitation of the sediment. The light source was a metal halide lamp (600 W Osram Super NAV), which provided 450 µmol m2 s−1 just below the water surface. Measurements in the field were conducted in the River Rhine near the city of Nijmegen on days with a cloudless sky during a flooding period between 17 and 31 May 1999. These measurements were carried out on the downstream side of a pontoon, where suitable conditions existed (i.e. a flat water surface and high flow rates, yielding a well-mixed water layer). There was no shading from the pontoon. All measurements were carried out in the middle of the day (12:00–14:00 h CET). The solar PPFD just below the water surface was around 800 µmol m2 s−1 on each day. PPFD was measured every 5 cm to a depth of 1 m (total depth was > 2.5 m). To account for fluctuations, light profiles in the field were determined three times directly after each other. The average PPFD for each depth was used for further analysis.

The light profiles were described with an exponential decline function (Lambert–Beer; eqn 1)

  • image(eqn 1)

where PPFDdepth is the depth-dependent downwelling PPFD, PPFD0 is the surface-level PPFD, and Kd is the attenuation coefficient for downwelling light. Kd values for each profile were determined by regression, using the sas procedure nlin (SAS 1989b). The resulting set of data was used to find a relationship between Kd and suspended load (see Results). Empirical, linear relationships between Kd and the concentrations of dissolved and suspended material have been described in other studies (e.g. Van Duin et al. 2001).

experiments

Plant species

In the experiments, four common Rhine floodplain plant species were studied: Arrhenatherum elatius, Achillea millefolium, Rumex acetosa and Rumex crispus. The submergence tolerance of three of these species was examined in a submergence experiment (Blom et al. 1994) and shown at high light availability to be A. elatius (intolerant) < R. acetosa < R. crispus (tolerant).

Seeds of the four species originated from the Dutch Rhine river area (A. elatius: Rijswaard, Neerijnen; A. millefolium: Zaltbommel; R. acetosa: Rijswaard, Neerijnen; and R. crispus: Bemmelse Waard). After germination, seedlings were grown in 380-cm3 plastic pots in a greenhouse for 8 weeks. The substrate consisted of a sieved mixture of one volume of potting soil and three volumes of a sandy clay originating from a floodplain near Nijmegen. Plants were watered daily with tap water. To prevent nutrient deficiency, 65-cm3 nutrient solution (see Vervuren et al. 1999) was given twice a week, from the third week onwards. A minimal PPFD of 150 µmol m2 s−1 at plant level was provided by a combination of metal halide (600 W Osram Super NAV) and sodium lamps (600 W SON-T) with a 16-h photoperiod.

Experimental set-up

The effects of PPFD and submergence depth on plant survival was studied in two experiments, referred to as the light experiment and the depth experiment, respectively. The experiments were carried out in a heated greenhouse in large basins (1.85 m diameter) filled with tap water. PPFD levels in the light experiment were varied by covering basins with up to three layers of well-ventilating shade cloth, rather than by suspension to avoid problems with precipitation over the long duration of the experiment. Light was provided by sodium lamps (600 W SON-T; photoperiod 16 h). The PPFDs at the top of the pots under these conditions were 0.4, 3.5, 17 and 135 µmol m2 s−1. Water depth in the light experiment, measured from the top of the pot to the water surface, was 65 cm. In the depth experiment depth levels were 40, 80, 120 and 160 cm and plants were submerged in total darkness. Total dissolved inorganic carbon in the basins was 2 mol m−3 with pH 7.9, resulting in a CO2 concentration of 58 mmol m−3 (measurements and calculations as described in Vervuren et al. 1999).

A total of 300 plants per species were grown prior to each experiment. After removal of plants with retarded growth, the total number of plants of each species varied between 216 and 275 per experiment. These were evenly divided over the four treatment levels. The light experiment with A. elatius, A. millefolium and R. acetosa was conducted in spring 1996. The depth experiments with these species and both the light and the depth experiment with R. crispus were carried out in the winter of 1996/1997. In particular R. crispus displayed elongation of petioles and leaf blades, but the elongated leaves never reached the water surface. In the 40-cm depth treatment, leaves of all species were kept submerged by means of a mesh screen.

The water temperature remained constant through time and was comparable between experiments (range 17.4–18.8 °C; measured at pot level). To check whether hypoxic or anoxic conditions would develop during the experiment, the oxygen concentration was measured in the various basins on several occasions. The lowest concentration recorded at pot level was 6.5 mg L−1, indicating that levels remained within the well-oxidized range. Rumex crispus had to be submerged for a very long time before plants started to die, allowing strong algal development in the basin at the highest PPFD. This species/treatment combination was therefore omitted from the experiment. Water remained clear throughout the experiment in all other basins.

Groups of usually 10 plants were removed from the basins at various time-intervals, and were allowed to recover under conditions similar to those for their initial growth. Plants were presumed dead when above ground parts had died and no regrowth occurred within 2 weeks.

Survival analysis

Survival analysis was carried out using the sas procedure lifereg (SAS 1989b). Survival data were fitted to a Weibull model (eqn 2).

  • image(eqn 2)

where G represents the survival probability over time (duration of submergence), and σ (ranging within 0–1) and µ are the function parameters which determine the shape of the distribution and the overall survival time, respectively. The median lethal time (LT50) was computed on the basis of µ and σ (eqn 3):

  • LT50 = eµ+σ log(log(2))(eqn 3)

Standard errors of lt50 were computed using the standard errors of the Weibull parameters provided by sas lifereg (eqn 4):

  • image(eqn 4)

A likelihood-ratio test (Cox & Oakes 1984), based on the log-likelihoods given by sas lifereg, was used to determine the statistical significance of treatment and species effects.

For each species, the relationship between the Weibull parameter µ and PPFD was described by a hyperbolic tangent function (see Vervuren et al. 1999), using the sas procedure nlin (SAS 1989b).

modelling of species lower boundaries

A model was constructed that combined the survival responses in the light experiment with the duration of flooding and the underwater light availability during three extreme floods (1970, 1980 and1983) and one average Rhine flood (1984). See Appendix for details on modelling procedures. The daily suspended load, required to determine underwater light transmission, for the selected floods was predicted using a second order polynomial relationship between river discharge and suspended load for the period 1989–99 (n = 3926, P < 0.0001, R2 = 0.81; sas procedure nlin). Daily readings of suspended load for the River Rhine were only available for this period. The surface PPFD, used to derive underwater light intensity from transmission, was assigned a constant value of 38.5 mol m−2 day−1, which is the mean of May–July 1971–82, De Bilt, the Netherlands (derived from Velds (1990), using a factor of 210 kJ mol−1 photons to convert irradiance to PPFD (Nobel 1991)). The underwater PPFD was calculated using the light attenuation model (see Results section).

Due to strong fluctuations in depth and suspended load, these calculated PPFDs often contained extreme values. The median was chosen as a summary measure of the PPFD of a flooding period as it is relatively insensitive to skewed distributions. The median PPFD was used to derive species-specific values of the survival parameter µ from the results of the light experiment. Together with flood duration, µ yields a survival estimate following the Weibull model, for a given flooding period and elevation. Weibull parameter σ was assigned a constant value of 0.158, the average across species and treatments in the light experiment (n = 15, SD: 0.045).

The model was run for elevations ranging from 11 m to 15 m with a step size of 0.2 m. To summarize the resulting survival vs. elevation curves, the elevations corresponding to 95%, 50% and 5% survival were determined. This was done through linear interpolation between the 0.2 m elevation levels. Three suspended load scenarios were implemented: 0.1 ×, 1 × and 10 × normal suspended loads. The final results of the analysis consisted of three boundary elevations (95%, 50% and 5%), for 48 combinations of species (4 ×), scenarios of suspended load (3 ×) and flooding periods (4 ×).

The model was written in Base sas 6.12 (SAS 1989a). Source program is available upon request.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

duration of summer floods

The shape of flood duration curves (Fig. 1) varies considerably between extreme floods. As a result, flood duration at low- and mid-level floodplain shows little relationship with peak height. The 1983 flood was the highest summer flood of the Rhine recorded in the 20th century, but flooding lasted a considerably shorter time at mid-level elevations (between 11.5 m and 13 m) than it did during the floods of 1970 and 1980. Around 12 m, the duration of flooding during the 1983 flood was even comparable to the average flood of 1984. All floods were characterized by sudden increases in flood duration at a level between 12 m and 12.5 m.

image

Figure 1. Relationship between flood duration and Rhine floodplain elevation (Lobith station) during three extreme summer floods (1970, 1980 and 1983) and one average summer flood (1984). Duration was based on the flooding period associated with the highest summer water level. Other flooding peaks within the same year or days occurring outside the summer period (1 May−31 September) were excluded. The mean summer exceedance curve of 1950–99 is plotted for comparison.

Download figure to PowerPoint

Extreme summer floods are rare and occur erratically. The recurrence intervals between extreme summer floods may vary from a few years to over a decade. At 13 m, extreme floods (lasting between 11 and 22 days) occurred in 1965, 1970, 1980, 1983 and 1987, but on no subsequent occasion.

underwater light availability

The attenuation coefficient Kd showed a linear relationship with suspended load (n = 18, P < 0.0001, R2 = 0.996; sas procedure reg). Because the data derived from field measurements matched the pattern of the laboratory data, regression was conducted on the combined data. Inserting the linear function in the Lambert–Beer equation yielded an empirical light attenuation model (Appendix, eqn 2) that enabled prediction of transmission from depth and suspended load. Figure 2 shows curves of equal underwater light transmission (isophotes) as a function of depth and suspended load, predicted by the light attenuation model.

image

Figure 2. Underwater light transmission curves in relation to depth and suspended load, determined by the light attenuation model.

Download figure to PowerPoint

Figure 3 shows the water levels, suspended load and daily transmission at the reference elevations (11 m and 13 m) during a single year. As may be expected, peak levels in suspended load correlate with the peak height of flooding. The highest floods are characterized by long periods of deep flooding and rather constantly low transmission levels, despite fluctuations in suspended load. This accords with predictions by the model (Fig. 2) that at greater depths (e.g. > 1.5 m), light transmission is below 1% even at baseline levels of suspended load (c. 30 mg L−1). High transmission levels mostly occur during short, shallow floods or at the end of the flooding period when water levels have sufficiently lowered.

image

Figure 3. Rhine daily water level, suspended load and underwater light transmission during a one-year period (1 October 1998–31 September 1999) for two reference levels, 13 m (lower boundary of flood-sensitive grassland) and 11 m (lower boundary of perennial, non-pioneer plant growth). Transmission is calculated using the light attenuation model.

Download figure to PowerPoint

To compare light transmission between periods of flooding, the light transmission during periods of flooding was integrated using the median of the daily transmission values. Figure 4 shows the median light transmission plotted against the median depth of all floods in the period 1989–99. The median transmission shows a clear relationship with the median depth. Beyond a median depth of 1 m, virtually all floods have a median transmission below 1%. For a given median depth (particularly in a range between 0 m and 1 m) the median transmission is usually lower at 13 m than at 11 m. This is due to the fact that at higher floodplain levels, the peak values of suspended load often dominate the flooding period.

image

Figure 4. The median underwater light transmission during flood of all flooding periods of the Rhine between 1989 and 1999 in relation to the median depth of submergence, for low-level (11 m) and mid-level (13 m) floodplains. Daily transmission was calculated on the basis of suspended load and depth, using the light attenuation model. ○, winter; •, summer.

Download figure to PowerPoint

submergence experiments

The survival data of the species fitted well to the Weibull distribution (Fig. 5). Increasing the depth of submergence had a negative effect on plant survival in all species, whereas increasing the light intensity had a positive effect in all species. Both light and depth effects were statistically significant (Table 1).

image

Figure 5. Survival of the four plant species in the depth and light experiments. Submergence depth levels were 40, 80, 120 and 160 cm. Survival data were fitted to a Weibull distribution using the sas procedure lifereg (see Methods). Note the different scales of the submergence-duration axes.

Download figure to PowerPoint

Table 1.  Main effects and interactions of species and treatments in the light and depth experiments. Log-likelihoods were calculated using the sas lifereg procedure. Light effects were tested on the basis of log-transformed PPFDs
TestExperiment2 × difference log-likelihoodsd.f.Probability
  • 1

    Only PPFD levels 0.4, 3.5 and 17 µmol m−2 s−1 are included.

Effect of treatmentLight1 36.62< 0.001
Depth 36.83< 0.001
Difference between speciesLight1185.23< 0.001
Depth477.43< 0.001
Treatment × SpeciesLight1326.46< 0.001
Depth 62.09< 0.001

The shape of the relationship between survival and treatment was studied on the basis of the median survival time, LT50 (Fig. 6), derived from the Weibull parameters (see Methods). The relationship between LT50 and depth was similar among species. In general, LT50 decreased 2- to 3-fold between the lowest and the highest inundation depth (e.g. in R. acetosa, LT50 is 40 days at 0.4 m depth, and 22 days at 1.6 m depth). The relationship between survival and light intensity differed more strongly between species, both with regard to the overall effect and to the shape of the relationship. In general, the increase in LT50 between the lowest and highest light intensity is 3- to 5-fold. The effects of light are observed even at a very low range (0.4–3.5 µmol m−2 s−1). The largest effects of light are observed in R. crispus. At 0.4 µmol m−2 s−1, the LT50 in this species was 88 days, compared to 641 days at 17 µmol m−2 s−1. For comparison, between 0.4 µmol m−2 s−1 and 17 µmol m−2 s−1 the LT50 in R. acetosa increased from 28 days to 58 days.

image

Figure 6. Median lethal time (LT50) and Weibull parameter µ vs. PPFD and depth of submergence. LT50 was derived from the survival parameters µ and σ of the Weibull distribution (see Fig. 5) using a value of 0.158 for σ. Curves describing the µ vs. PPFD data of each species were fitted using a hyperbolic tangent function. Bars represent 95% confidence intervals.

Download figure to PowerPoint

Note that the tolerance ranking of A. millefolium and A. elatius changed with light intensity. Otherwise, the differences between species in tolerance to complete inundation are consistent across conditions.

simulated lower boundaries

The submergence-tolerant species R. crispus was predicted to survive any extreme flood down to 11 m (Fig. 7); the lowest elevation of the floodplain suitable for perennial non-pioneer species. For the flooding-intolerant species, A. millefolium and A. elatius, the extreme years showed predicted lower boundaries ranging from mid- to high-floodplain elevation. Rumex acetosa showed an intermediate response, with lower boundaries lying between low- and mid-level elevation in the extreme years.

image

Figure 7. Simulated lower distribution boundaries of four riparian plant species in three extreme years (1970, 1980 and 1983) and one average year (1984). The three scenarios of suspended load correspond to the three bars within each year (0.1 times, 1 times and 10 times normal concentrations). The transitions within each bar correspond to 95%, 50% and 5% survival. See Methods for details. The areas in the background represent the range between the 5th and 95th percentile of the distribution of the species, as determined on the basis of 123 relevées made at 6 floodplain sites of the River Rhine in the year 2000 (H. M. van de Steeg & W. van Eck, unpublished data). Note: the 5th percentile of the actual distribution of R. crispus is associated with an elevation of exactly 11 m.

Download figure to PowerPoint

The lower boundaries of the intolerant species A. millefolium and A. elatius predicted for the extreme years (under the normal suspended load scenario) corresponded with the actual field distribution recorded in 2000. In contrast, the predicted lower boundaries of the more tolerant R. acetosa and R. crispus occurred substantially below the actual lower boundaries.

The model predicted that the average year 1984 had no influence on either R. crispus or R. acetosa between 11 mand 15 m. For the intolerant species lower boundaries were predicted at c. 11 m under the low suspended load scenario.

Overall, varying the suspended load resulted in considerable shifts of the predicted lower boundaries. The effects were most pronounced in the light-sensitive species Achillea millefolium. The ranking of the lower boundaries of Achillea millefolium and A. elatius depended on suspended load scenario and year, with A. elatius reaching to lower elevations in dark scenarios and A. millefolium in light scenarios. An exception to this was the 1983 flood.

The 1983 summer flood held record values of water level (Fig. 1) and discharge, with concomitantly high suspended loads (161 mg L−1 on the day of the highest water level, compared to 84 mg L−1 for 1980). As a result, the light scenarios for this flood fell into a low light range; below the point where A. millefolium is more tolerant than A. elatius (Fig. 6). Because A. millefolium is very intolerant to submergence in this range, even the short submergence duration at high floodplain level was predicted to have a large impact (Fig. 7). The flooding tolerance of A. elatius is relatively high in this low light range, and is very unresponsive to light (Fig. 6). As a consequence, for 1983, the predicted lower boundaries of this species occurred at a relatively low level, and the suspended load scenarios had virtually no effect.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

the flooding gradient

The implications of responses to submergence are determined by the range and variability of the key disturbance components, flood duration and underwater light availability, along natural flooding gradients. Extreme floods are usually described in terms of peak water level (or discharge), but this does not provide information on the intensity of such floods along the flooding gradient. Our analysis of River Rhine data shows that there is considerable variation in the relationship between peak water level and duration. For example, the 1983 summer flood in the lower Rhine reached an extremely high level, but, overall, flood duration was less extreme than during the 1980 flood (Fig. 1). An important similarity between the different floods is the sudden increase in flood duration at the transition from mid- to low-level elevation. This is a direct consequence of the shape of the river basin, i.e. a wide floodplain above a relatively narrow channel, and may explain the rapid changes in species composition observed around these elevations (see Sykora et al. 1988).

The analysis of underwater light availability showed that deep (extreme) Rhine floods, i.e. floods of over 1 m depth, are characterized by very low median transmission levels, ranging from 0.5% to as low as 1 × 10−7% (Fig. 4). However, even in this low range, light availability may have large effects on plant survival, as a simple calculation illustrates. When assuming an average summer PPFD of 38.5 mol m−2 day−1, 0.5% transmission corresponds to a PPFD of 0.19 mol m−2 day−1. The largest effects of light in our submergence experiment, as discussed below, were observed almost entirely below this level (Fig. 6). Thus, light transmission levels as low as 0.05%, which occur regularly during extreme floods (Fig. 4), are likely to have a positive effect on the chances of riparian plant species surviving submergence.

effect of light and depth on the ability to survive submergence

Both submergence depth and underwater light intensity significantly affected the survival of the selected riparian species in the submergence experiments. There was little difference between the species regarding the magnitude of the depth effect. On average, the median lethal time (LT50) at a depth of 1.6 m was roughly half the value at 0.4 m, which is rather similar to the effects found in rice cultivars by Adkins et al. (1990). The cause of this depth effect is unclear. It may be hypothesized that, with increasing depth, gas-filled intercellular spaces may become smaller in size or gradually become filled with water. This would probably hamper internal ventilation (see Jackson & Armstrong 1999) and thereby accelerate plant decay.

Underwater photosynthesis relieves the low partial pressure of oxygen and the carbohydrate shortage resulting from submergence (Crawford & Brändle 1996). Photosynthesis curves are characterized by a maximum slope below the light compensation point (e.g. Björkman 1981), and this may explain the observation that the largest increases in the median survival time occurred at very low light intensities (0.4–3.5 µmol m−2 s−1; Fig. 6). For comparison, Sand-Jensen & Madsen (1991) described an average light compensation point of 6.9 µmol m−2 s−1 in relation to the growth of some submerged freshwater macrophytes. The gentle slope of the LT50-PPFD curve in the range from 17 µmol m−2 s−1 to 135 µmol m−2 s−1 in our experiment suggests an early saturation of the light response, possibly as a result of co-limitation by CO2 (see Madsen & Sand-Jensen 1994).

There were large differences between the species with regard to the effect of light on survival (Fig. 6). A particularly large response to light was observed in R. crispus. Between 0.4 µmol m−2 s−1 and 17 µmol m−2 s−1, the LT50 of this species increased from about 3 months to almost 2 years. The extreme tolerance to dark-submergence of R. crispus is probably related to an economical anaerobic energy metabolism of the tap-root (see Nabben 2001). The weakest light-responses were observed in A. elatius and R. acetosa. Because A. millefolium had a lower dark-tolerance than A. elatius, but showed a relatively large light-response, the tolerance rankings of these intolerant species reversed across the applied light range. Consistent with the results of the light experiment, Vervuren et al. (1999) showed that R. crispus was able to maintain underwater photosynthetic capacity during at least 30 days of submergence, whereas in A. elatius underwater photosynthesis collapsed within a few days.

In conclusion, the experimental results suggest that underwater light availability, but not the pressure-related effects of water depth, may differentially affect plant species survival and hence species zonation along flooding gradients.

survival of flooding and riparian plant zonation

There have been few attempts directly to translate responses of species to disturbance factors in experiments to their field distributions along disturbance gradients. Here we used the experimental responses of the four species to model their lower distribution limits for a range of historical Rhine summer floods. A rigid test of the validity of our model predictions would require repeated field observations on species distributions before and after large flood-disturbances. However, such information is rare, which is a general constraint for studying large infrequent and unpredictable disturbances (Turner & Dale 1998). Our modelling must therefore be regarded primarily as a method to examine the implications of the experimental results and to generate new hypotheses about the role of flooding components in the natural environment. There are some field observations, however, that enable us to evaluate our model.

In years of extremely high floods and for scenarios of unaltered suspended loads, the modelled lower boundaries of A. millefolium and A. elatius were found at high elevation (Fig. 7). In contrast, the average year predicted far lower distributions. The actual field distributions in 2000 corresponded with those predicted in the extreme years of more than a decade earlier rather than with the average years, suggesting that extreme years determine the distributions of these flood-sensitive species for many years. The apparent slow colonization of lower floodplain elevations could be the result of limited dispersal or microsite limitation. The predicted lower boundary patterns of A. millefolium and A. elatius varied considerably among the extreme years (each having typical duration and depth characteristics), and among scenarios of underwater light availability. This demonstrates that the impact of a given flood depends on the interplay between the overall tolerances of the species, their responsiveness to light, the duration of flooding, and the underwater light intensity as determined by flooding depth and suspended load. In most cases, A. millefolium was found below A. elatius under the most favourable light conditions, but the reverse was true under the darkest conditions. Thus, the reversal of tolerance rankings in the light experiment was reflected in the lower boundaries predicted for different suspended load scenarios. Rivers strongly vary in the baseline level of suspended load. Within a classification of American rivers by Meade & Parker (1985, cited in Walling & Webb 1992), the River Rhine, with a mean suspended load of approximately 30 mg L−1, belongs to the lowest class. Other rivers may possess suspended loads 10 times this level; similar to our high suspended load scenario, or even higher. Within river systems, suspended sediment loads may decrease (Belt 1975) or increase (Edwards 1969) as a result of alterations by man. Our model results suggest that plant lower distribution limits may be significantly shifted upwards or downwards depending on the suspended load of the river system.

In contrast to the flood-sensitive species, the predicted lower boundaries of the intermediately tolerant R. acetosa and the highly tolerant R. crispus for both extreme and average years are below the actual field distribution in 2000 (Fig. 7). This suggests that the current distribution is only partly influenced by major flood-disturbances and that other factors, either proximate or historical may play a prominent role. Unlike our model results, where R. acetosa occurred substantially lower than A. elatius and A. millefolium, the actual lower boundary of R. acetosa recorded in 2000 was rather similar to those of A. elatius and A. millefolium. The field distribution in 2000 corresponds well with the distribution patterns of these three species in a lower Rhine tributary, 2 years after the extreme flood of 1983 (Sykora et al. 1988). Other site-related conditions (e.g. soil characteristics) may have considerable control over these data, and the limited number of sites for which information is available might be a reason for the discrepancy between the field situation and the predictions based on our experimental results.

The model predicted that, based on its tolerance to submergence, R. crispus may survive extreme floods even below the lowest elevations suitable for perennial herbaceous plant growth in the Rhine floodplain (Fig. 7). River engineering has dramatically altered the flow regimes of rivers world wide (Dynesius & Nilsson 1994). These changes often include increased dynamics at lower elevations, and a transition from shallow, lengthy floods to deep, relatively short floods (see Belt 1975; Brock et al. 1987; Sparks et al. 1998). The apparent ‘overkill’ in submergence-tolerance in R. crispus suggests that its evolution took place prior to river engineering when surviving long-term floods (that may no longer occur) was necessary, and that this may have been critical for its persistence at low-elevation sites. The extreme submergence-tolerance of this species may be an evolutionary fixed property.

Conclusions

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Extreme summer floods are a major disturbance to riparian plants. Flooding is a phenomenon with different components, each affecting plant survival. Our results show that the various components of extreme floods are poorly correlated and that plant species may respond differently to some of these components. In particular, underwater light intensity, regulated by submergence depth and suspended load, plays an important role in determining the likelihood of riparian plant species surviving periods of submergence. This effect is in addition to ‘classical’ components of flood disturbance, such as flood timing and duration. The impact of extreme flood-disturbances on species distributions cannot therefore be predicted solely from basic hydrological parameters like peak water level or discharge. Due to significant differences in the responsiveness of species to light and to the variability of underwater light availability during flooding, underwater light is an important differentiating factor for the zonation of riparian plants in river floodplains.

Discrepancies between model predictions and current field distributions of the species suggests that historical factors substantially influence current distributions. Despite the long-term absence of flooding events, with ample opportunities for re-invasion of low elevated sites, intolerant species remain at levels that have been set by historical extreme flood-disturbance events. Tolerant species may possess ‘excessive’ flood-tolerance, a situation that may reflect an evolutionary history in a disturbance regime that differs from the current, man-influenced regime.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

The following persons made invaluable contributions to this study as undergraduate students: Simone Beurskens (experiments), Ankie Brock (light attenuation model and experiments), and Jorg Reuvers (analysis of hydrological data). We are indebted to Jos Michielse for assistance during experiments and field measurements; John Lenssen and Eric Visser for commenting on the manuscript; and Werner van Eck and Harry van de Steeg for field data on species distribution. We should like to thank Karle Sykora for providing the raw data of his 1988 paper. Data on water levels, river discharge and suspended load were kindly provided by the Waterdata-desk, RIZA, Lelystad, the Netherlands.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  • Adkins, S.W., Shiraishi, T. & McComb, J.A. (1990) Submergence tolerance of rice – A new glasshouse method for the experimental submergence of plants. Physiologia Plantarum, 80, 642646.
  • Belt, C.B.J. (1975) The 1973 flood and man's constriction of the Mississippi river. Science, 189, 681684.
  • Björkman, O. (1981) Responses to different quantum flux densities. Physiological Plant Ecology. I. Responses to the Physical Environment (eds O.Lange, P.Nobel, C.Osmond & H.Ziegler), pp. 57107. Springer-Verlag, Berlin.
  • Blom, C.W.P.M., Van de Steeg, H.M. & Voesenek, L.A.C.J. (1996) Adaptive mechanisms of plants occurring in wetland gradients. Wetlands: environmental gradients, boundaries, and buffers. Proceedings of an International Symposium, April 1994 (eds G.Mulamoottil, B.G.Warner & E.A.McBean), pp. 91111. CRC Press Inc, Boca Raton, Florida, USA.
  • Blom, C.W.P.M., Voesenek, L.A.C.J., Banga, M., Engelaar, W.M.H.G., Rijnders, J.G.H.M., Van de Steeg, H.M. & Visser, E.J.W. (1994) Physiological ecology of riverside species: adaptive responses of plants to submergence. Annals of Botany, 74, 253263.
  • Brock, T., Van der Velde, G. & Van de Steeg, H.M. (1987) The effects of extreme water level fluctuations on the wetland vegetation of a nymphaeid-dominated oxbow lake in The Netherlands. Archiv für Hydrobiologie, Beiheft Ergebnisse der Limnologie, 27, 5773.
  • Cox, D.R. & Oakes, D. (1984) Analysis of Survival Data. Chapman & Hall Ltd, London.
  • Crawford, R.M.M. & Brändle, R. (1996) Oxygen deprivation stress in a changing environment. Journal of Experimental Botany, 47, 145159.
  • Dynesius, M. & Nilsson, C. (1994) Fragmentation and flow regulation of river systems in the Northern third of the world. Science, 266, 754762.
  • Edwards, D. (1969) Some effects of siltation upon aquatic macrophyte vegetation in rivers. Hydrobiologia, 34, 2937.
  • Friedrich, G. & Müller, D. (1984) Rhine. Ecology of European Rivers (ed. B.A.Whitton), pp. 265314. Blackwell Scientific Publications, Oxford.
  • Gill, C.J. (1970) The flooding tolerance of woody species – a review. Forestry Abstracts, 31, 671688.
  • Grime, J.P. (1979) Plant Strategies and Vegetation Processes. John Wiley & Sons, Chichester.
  • He, J.B., Bögemann, G.M., Van de Steeg, H.M., Rijnders, J.G.H.M., Voesenek, L.A.C.J. & Blom, C.W.P.M. (1999) Survival tactics of Ranunculus species in river floodplains. Oecologia, 118, 18.
  • Hosner, J.F. (1960) Relative tolerance to complete inundation of fourteen bottomland tree species. Forest Science, 6, 246251.
  • Jackson, M.B. & Armstrong, W. (1999) Formation of aerenchyma and the processes of plant ventilation in relation to soil flooding and submergence. Plant Biology, 1, 274287.
  • Klimesova, J. (1994) The effects of timing and duration of floods on growth of young plants of Phalaris arundinacea L. and Urtica dioica L.: an experimental study. Aquatic Botany, 48, 2129.
  • Lytle, D.A. (2001) Disturbance regimes and life-history evolution. American Naturalist, 157, 525536.
  • Madsen, T.V. & Sand-Jensen, K. (1994) The interactive effects of light and inorganic carbon on aquatic plant growth. Plant, Cell and Environment, 17, 955962.
  • Nabben, R.H.M. (2001) Metabolic adaptations to flooding-induced oxygen deficiency and post-anoxia stress in Rumex species. PhD Thesis, University of Nijmegen, Nijmegen, the Netherlands.
  • Nabben, R.H.M., Blom, C.W.P.M. & Voesenek, L.A.C.J. (1999) Resistance to complete submergence in Rumex species with different life histories: the influence of plant size and light. New Phytologist, 144, 313321.
  • Nobel, P.S. (1991) Physicochemical and Environmental Plant Physiology. Academic Press Inc, San Diego, California.
  • Sand-Jensen, K. & Madsen, T.V. (1991) Minimum light requirements of submerged freshwater macrophytes in laboratory growth experiments. Journal of Ecology, 79, 749764.
  • SAS (1989a) sas Language and Procedures, Version 6, 4th edn. SAS Institute Inc., Cary, North Caroline.
  • SAS (1989b) sas/stat User's Guide, Version 6, 4th edn. SAS Institute Inc., Cary, North Caroline.
  • Siebel, H.N., Van Wijk, M. & Blom, C.W.P.M. (1998) Can tree seedlings survive increased flood levels of rivers? Acta Botanica Neerlandica, 47, 219230.
  • Southwood, T.R.E. (1988) Tactics, strategies and templets. Oikos, 52, 318.
  • Sparks, R.E., Nelson, J.C. & Yin, Y. (1998) Naturalization of the flood regime in regulated rivers – the case of the upper Mississippi River. Bioscience, 48, 706720.
  • Sykora, K.V., Scheper, E. & Van der Zee, F. (1988) Inundation and the distribution of plant communities on Dutch river dikes. Acta Botanica Neerlandica, 37, 279290.
  • Toner, M. & Keddy, P. (1997) River hydrology and riparian wetlands: a predictive model for ecological assembly. Ecological Applications, 7, 236246.
  • Turner, M.G., Baker, W.L., Peterson, C.J. & Peet, R.K. (1998) Factors influencing succession: lessons from large, infrequent natural disturbances. Ecosystems, 1, 511523.
  • Turner, M.G. & Dale, V.H. (1998) Comparing large, infrequent disturbances: what have we learned. Ecosystems, 1, 493496.
  • Van de Steeg, H.M. (1984) Effects of summer inundation on flora and vegetation of river foreland in the Rhine area. Acta Botanica Neerlandica, 33, 365366.
  • Van de Steeg, H.M. & Blom, C.W.P.M. (1998) Impact of hydrology on floodplain vegetation in the lower Rhine system: implications for nature conservation and nature development. New Concepts for Sustainable Management of River Basins (eds P.H.Nienhuis, R.S.E.W.Leuven & A.M.J.Ragas), pp. 131144. Backhuys Publishers, Leiden.
  • Van Duin, H.S., Blom, G., Los, F.J., Maffione, R.M., Zimmerman, R., Cerco, C.F., Dortch, M. & Best, E.P.H. (2001) Modeling underwater light climate in relation to sedimentation, resuspension, water quality and autotrophic growth. Hydrobiologia, 444, 2542.
  • Velds, C.A. (1990) Zonnestraling in Nederland; Klimaat van Nederland 3. Thieme/KNMI, Baarn.
  • Vervuren, P.J.A., Beurskens, S.M.J.H. & Blom, C.W.P.M. (1999) Light acclimation, CO2 response and long-term capacity of underwater photosynthesis in three terrestrial plant species. Plant, Cell and Environment, 22, 959968.
  • Walling, D.E. & Webb, B.W. (1992) Water Quality: I. Physical Characteristics. The Rivers Handbook, Vol. 1Hydrological and Ecological Principles (eds P.Calow & G.E.Petts), pp. 4872. Blackwell Scientific Publications, Oxford.