Introduction

Currently, strong re-oligotrophication processes are evident in many deep and shallow lakes (Jeppesen et al. 2005; Koehler et al. 2005; Stich and Brinker 2010), and are associated with an expansion of submerged macrophyte coverage and changes in community structure towards charophytes. Charophytes prosper under oligotrophic conditions and are replaced by tall-growing angiosperms during eutrophication (Chambers 1987; Ozimek and Kowalczewski 1984). Charophytes are generally the “gold standard” in littoral areas, since their bottom-dwelling structure and capacity to act as nutrient traps (Kufel and Kufel 2002) do not generally interfere with anthropogenic use of lakes (van Nes et al. 2002). Factors accounting for their reappearance after re-oligotrophication, however, are not well understood.

Competition for resources such as nutrients, light or space, strongly influence plant–plant interactions, and along with allelopathy or susceptibility to herbivores or pathogens, may affect individual plant performance and thus result in changes in community structure (Bornette and Puijalon 2011; Le Bagousse-Pinguet et al. 2012). Tall- and fast-growing species are considered strong competitors because they may form canopies and interfere with light availability for understory plants. Myriophyllum spicatum L. is found in various aquatic systems exhibiting a wide range of environmental conditions such as nutrient availability, salinity and light (Barko 1983). We recently showed that the outcome of interactions between M. spicatum and two pondweeds strongly depended on the structure and composition of the macrophyte community, the level of herbivory and the presence of chemical defences (Le Bagousse-Pinguet et al. 2012). Factors accounting for the decline of canopy-forming and potentially invasive species such as M. spicatum are less well understood (Trebitz et al. 1993).

Several studies report the recolonisation and expansion of charophytes and the replacement of angiosperms after re-oligotrophication (Blindow 1992; van den Berg et al. 1998b). Even early in the season, Chara globularis Thuill. showed the highest biomass under low nutrient conditions compared to Elodea canadensis (Planch.) H. St. John and Potamogeton pectinatus L. (Bakker et al. 2010). While M. spicatum is known to take up nutrients via roots and shoots, sediment nutrient uptake for Chara has been considered negligible (Kufel and Kufel 2002). However, this capacity might have been underestimated since a new cultivation technique showed good growth of three Chara species when only sediment phosphorus was provided (Wuestenberg et al. 2011). Further, recent findings suggest that polar auxin transport is involved in long-distance transport of nutrients from rhizoids to shoots in Charales (Raven 2013). Charophytes may have more efficient nutrient uptake systems at low nutrient availability (Bakker et al. 2010; Wuestenberg et al. 2011), allowing them to outcompete tall-growing aquatic angiosperms. Competition for bicarbonate might further explain the dominance of charophytes. The competitive advantage of Chara over P. pectinatus was explained by decreased DIC concentrations and increases in pH within Chara stands (Hidding et al. 2010; van den Berg et al. 2002).

At present, the littoral zone of Lower Lake Constance is widely covered by well-established charophyte stands (mainly Chara globularis, C. aspera Deth. ex Willd., also C. contraria A. Braun ex Kütz., C. tomentosa L.), probably as a result of nutrient depletion in the lake due to water treatment efforts of all adjacent states. Tall-growing angiosperms, among them M. spicatum, are declining and presently only occur in sparse stands among charophytes, in contrast to the situation about a decade ago (Gross and Kornijow 2002).

Based on the above-mentioned changes in macrophyte community structure, we hypothesise that high densities of Chara negatively affect the physiological performance and thus functional traits of M. spicatum. To determine the interaction between both species, we conducted an outdoor mesocosm experiment using the response surface design. The response surface model, also called complete additive design, represents the best choice for investigation if competition strength depends on the density of both the associated and the focal species since densities are varied independently (Inouye 2001). Higher densities of the smaller species are used to compensate for different growth forms or plant size (Connolly 1987). We measured standard growth parameters, leaf chlorophyll content and plant stoichiometry in both species, and for M. spicatum additionally biomass allocation in roots or shoots and autofragmentation.

Methods

Experimental design

To investigate the impact of variable plant density on intra- and interspecific interactions of C. globularis and M. spicatum, we conducted a response surface design experiment in the outdoor mesocosm area at the Limnological Institute, University of Konstanz, Germany (47°41′42 N, 9°11′37E). Both species were planted at two densities, in mono and mixed cultures yielding eight treatments. Twenty mesocosms were distributed among the eight treatments, with three replicates each for the mixed cultures, and two for each density of the monocultures. We allowed for both above- and below-ground interaction, given that Chara might also be able to compete for sediment nutrients.

Each mesocosm consisted of a 1-m-tall Perspex® tube (diameter = 14.5 cm, total volume 16 L). Tubes were arranged side-by-side at a southeast facing, 2-m-high wall in a 2-m-wide passage between larger concrete mesocosms, and tied with ropes to a wooden frame. Every tube was connected to a flow-through system (20 L h−1) consisting of a water reservoir (120 L), a circulating pump (Universal pump 1250A, Eheim, Deizisau, Germany), filters (coarse- and fine-pored polyurethane panels, diameter = 31 cm), a UV lamp (Aqua Cristal UV-C 5 W Series II, JBL, Neuhofen, Germany), a heater (T+H EHK CE 1108, Gerstetten, Germany) and the regulated inflow of water from the institutes’ Lake Constance water circulating system. The control unit (Technical workshop, University of Constance) maintained a constant temperature of 19 °C.

We chose this set-up to reflect in situ in the calcareous conditions in Lake Constance as closely as possible. The flow-through system avoided strong photosynthesis-driven changes in pH and alkalinity often found in closed containers (Hidding et al. 2010). We measured inorganic carbon a few times during the experiment, yielding values between 1.8 and 2.2 mM bicarbonate. Our sampling sites are located in “Rheinsee”, the part of Lower Lake Constance where the river Rhine flows through, and water exchange is high (Rossknecht 2006). Minimum bicarbonate concentrations are 1.8 mM, reaching up to 2.8 mM in winter. The pH ranges from 8.0 to 8.6. Thus our mesocosms represented the field situation in terms of seasonal inorganic carbon availability. The chosen water temperature reflected values recorded in early summer in the littoral zone of Lower Lake Constance. To reduce the effects of lateral incident light, we covered the lateral open site of all tubes with black gauze, yielding irradiances at approx. 20 cm water depth comparable to that in the lake (150–1,200 μmol photons m−2 s−1, depending on sunshine).

Each mesocosm consisted of a 20-cm-high base tube, on which an 80-cm extension was placed, connected by an O-ring seal and bolts. This allowed placing of pre-cultured pots (see below) into the base before adding the extension. Pots had a diameter of 13 cm and were two-thirds filled with a mixture of sand, clay (10 %) and long-term substrate fertiliser-concentrate (5 %, TetraPlant® CompleteSubstrate, Tetra, Melle, Germany), covered by a 4-cm layer of original sediment from Lake Constance. The pots remained in the tubes throughout the experiment.

At our field sites Chara reappears during May before M. spicatum and rapidly establishes high biomass densities of 400–600 g m−2 (ash-free dry mass; AFDM), while only individual tufts of M. spicatum with maximum shoot length of 20 cm are visible by the end of June (Richter, unpub. data). Thus, we planted C. globularis 3 weeks in advance mimicking field conditions. We used apical shoots from an established stand, originally from Lower Lake Constance, grown in one of the permanent outdoor mesocosms connected to continuous lake water through-flow. Shoots of approx. 4 cm and 3 nodes were planted one-third into the sediment, and pots were precultured in their tube bases at a water depth of 80 cm. The final starting densities were 0, 6 and 18 shoots per pot (corresponding to 0, 500 and 1,500 plants m−2), but we planted two-thirds surplus to allow for mortality or weak growth. On 1 July we removed excess shoots; established shoots had a height of 11.2 ± 0.7 cm (mean ± SE), but had not yet formed a dense mat.

Myriophyllum spicatum originated from Lower Lake Constance and was transferred to the outdoor mesocosm in order to adapt to the experimental conditions 3 days before the experiment was started. Due to the low mortality of M. spicatum observed in previous experiments, we planted no surplus shoots. Apical shoots were cut into four nodes (about 4 cm long) and planted at densities of 0, 1 and 4 shoots per pot (0, 100 and 300 plants m−2). Additional shoots of both species were taken to determine the initial dry weight for further growth analyses. Unfortunately, at the beginning of the experiment, four M. spicatum plants were added to both C. globularis pure stands with the highest density resulting in two additional replicates of the highest mixed culture combination. Therefore, only seven treatments were finally used in the experiment.

Water analysis

Conductivity, pH, dissolved oxygen concentration and temperature in the tubes were measured weekly at midday 20 cm below the water surface with a Universal Pocket Meter (MultiLine® P4, WTW, Weilheim, Germany) in one random replicate per mono and mixed culture of high density (four tubes). Water samples for the analyses of nitrogen and phosphorus were taken at the end of the experiment from one replicate per treatment (n = 7) and stored in acid-washed glass bottles in darkness at 4 °C overnight. We filtered twice, 50 mL per sample through Tuffryn® Membrane Filter disks (0.2 μm, diameter = 25 mm, Pall Corporation, Ann Arbor, USA). Soluble reactive phosphorus (SRP) and nitrate were determined from the filtrates and the particulate fractions (Ppart and Npart) from the respective filter disks; the sum of soluble and particulate P or N yielded total P or N. Samples for Ppart were digested in potassium peroxydisulphate/sodium hydroxide at 121 °C for 60 min, and SRP was measured with the molybdate-ascorbic acid method. Determination of P and N followed standard procedures (Greenberg et al. 1995), and were performed with a Technicon® AutoAnalyzer II (SEAL Analytical, Norderstedt).

Plant growth analysis

On August 24, 2009, we harvested the plants, 54 days after the start. We determined the maximum height of the C. globularis stands, but could not discern individual plants. The species were placed separately in plastic bags and transported on ice to the laboratory. We carefully washed the plants in tap water to remove all loose deposits on their surface. We recorded the number and length of main and lateral shoots and autofragments of M. spicatum. After recording the fresh weight of roots and shoots of M. spicatum, roots were dried at 80 °C and shoots were lyophilised. Dry shoots were separated into leaves and stems. We calculated the leaf-mass fraction of M. spicatum (LMF) by separately measuring leaves, stems and roots (Cornelissen et al. 2003). The root/shoot mass ratio for M. spicatum indicates resource allocation to above- or below-ground biomass. We analysed the dry-matter content (DMC) by dividing the aboveground dry mass by the fresh mass of each sample.

The relative growth rate (RGR) was calculated with the equation

$$ {\text{RGR}} = \left( {{\text{ln dm}}_{\text{t2}}-{\text{ln dm}}_{\text{t1}} } \right)/{\text{d}} $$

where dmt1 is the mean aboveground dry mass of individuals (g dm) at the beginning of the experiment, dmt2 the dry mass of individuals at the end of the experiment, and d the duration of the experiment (Hunt 1982). It was impossible to separate C. globularis into individual plants at the end of the experiment, thus the results were calculated based on the initial number of planted shoots to reflect the individual growth increase. The RGR remains the same, irrespective of whether we calculate with individual plants or the total number.

We ground leaves of M. spicatum and entire C. globularis shoots to a fine powder in a ball mill (Fritsch, Idar-Oberstein, Germany). Ash-free dry mass (AFDM) was determined after incineration for 5 h at 550 °C. We determined the tissue content of carbon, nitrogen and phosphorus (C, N, P) and calculated the molar stoichiometry for C/N and N/P. The C and N contents were measured with an elemental analyser EURO EA (series 3000, HEKAtech, Wegberg, Germany) and the content of P was determined by digesting 1 mg plant powder as mentioned above (see “Water analysis”). Leaf nitrogen concentration (LNC) and leaf phosphorus concentration (LPC) are the total amounts of N and P per unit of dm or AFDM of M. spicatum leaves, and per unit of AFDM of C. globularis. For Chara, dm is not suitable as a reference because of lime accumulation on the surface: the dry mass of charophytes may consist of about 60 % CaCO3, much higher than usually found in other submerged macrophyte species (van den Berg et al. 2002). We present results for C, N and P in M. spicatum based on dm and AFDM. Usually M. spicatum shows less-firm carbonate encrustations on their surface, and most references show the results based on dm. We additionally used AFDM to have comparable results to C. globularis. We also determined the leaf chlorophyll content of M. spicatum (Porra et al. 1989).

Statistical analysis

Data were analysed with Microsoft Excel and IBM SPSS Statistics 19 (SPSS Inc., USA). Among the numerous ways to calculate the outcome of competition via indices (Weigelt and Jolliffe 2003), competition coefficients measuring interspecific relative to intraspecific effects are considered valuable (Jensen 1987). Multiple regression analyses are commonly used to elucidate the main factors of competition (Spencer and Rejmanek 1989; van den Berg et al. 1998a). We chose to analyse our data with stepwise linear regression to assess the effects of the response surface treatments on the tested parameters for both species at each density combination. The effects of increasing densities were calculated for the RGR of both species, the shoot length of M. spicatum and maximum height of C. globularis.

We used the model:

$$ {\text{Pe}}_{\text{indMs}} = {\text{Pe}}_{0} - {\text{D}}_{\text{Ms}} \times {\text{f}}_{\text{Ms}} - {\text{D}}_{\text{Cg}} \times {\text{f}}_{\text{Cg}} $$

Spencer and Rejmanek (1989) had used the reciprocal equation. Pe indicates the performance, D the planting density of M. spicatum (Ms) and C. globularis (Cg), and the regression coefficients f are measures of intra- and interspecific competition. The coefficients provide quantitative estimates of the reduction in growth of an individual of M. spicatum due to the addition of conspecifics or of individuals of C. globularis (Spencer and Rejmanek 1989). The same model was used analogously for C. globularis.

In the stepwise regression analysis, density was used as explanatory variable, with AFDM, C, N, P contents, chl a and b content and ratios as response variables, and also to examine the changes in biomass allocation patterns of M. spicatum (LMF, root/shoot ratio and DMC) with and without competition. We used the partial F-test to accept or reject a predictor variable. Density was entered into the equation when the probability of the F-value was lower than 0.05 and removed when greater. The significance of the partial regression coefficients was tested by the Student’s t test, corrected by the sequential Bonferroni test, using a base significance level of 0.05. The relation between chlorophyll concentration and N or P content in M. spicatum leaves was analysed using Pearson′s correlation. We used a regression 3D panel of RGR of C. globularis and of RGR, length and number of autofragments of M. spicatum for graphical visualisation (SigmaPlot 11.2.0; Systat, Chicago, USA).

Results

Water chemistry

The flow-through system accounted for stable physicochemical water parameters during the experiment. Weekly measurements at midday showed a conductivity of 276.3 ± 0.3 μS cm−1, pH 8.24 ± 0.01, oxygen concentration 8.7 ± 0.1 mg L−1 and temperature 19.0 ± 0.1 °C, with no differences between mono and mixed treatments or over time. We also did not observe any difference in total phosphorus and nitrogen concentrations (F-test: F 2,6 = 1.81, P = 0.276 and F 2,6 = 2.694, P = 0.182, respectively), which were in the range of present-day values for Lake Constance. Total P was 8 ± 0.2 μg L−1 and total N 757 ± 11 μg L−1 at the end of the experiment. The low variance of approx. 2 % indicates the homogeneity of macronutrients in all mesocosms, even though we only measured one replicate per treatment. Also, soluble nutrients did not differ between treatments (SRP 1 ± 0.3 μg L−1, nitrate 716 ± 1 μg L−1; F-test, F 2,6 = 1.071, P = 0.424 and F 2,6 = 0.202, P = 0.825, respectively). We performed no analyses of the sediment nutrient concentrations, but analysed the nutrient content of the plants as a proxy of nutrient uptake (see below).

Chara globularis

We separately analysed plant performance and functional traits for both species. The results for C. globularis are summarised in Table 1. The final biomass was 1041 ± 52 g dm m−2 at low and 1456 ± 85 g dm m−2 at high density. The RGR (Fig. 1a) declined with increasing density by approx. 23 %, documented by the negative intraspecific competition coefficient. Myriophyllum spicatum had no influence on C. globularis biomass development; therefore this variable was rejected in the stepwise linear regression. The stand heights reached 50–55 cm and did not differ between treatments. Ash and dry-matter content did not vary among treatments. Also, the C, N and P contents and resulting C/N and N/P molar ratios were unaffected.

Table 1 Results of stepwise linear regression of growth parameters and functional traits of Chara globularis based on the model PeindCg = Pe0 (± 1 SE) − DCg × fCg (± 1 SE) − DMs × fMs analogous to the model presented in “Methods” section, and with Pe representing the respective performance parameters as listed below
Full size table
Fig. 1
figure 1

Effect of Chara globularis density on a the RGR of C. globularis (d−1), b the RGR of Myriophyllum spicatum (d−1), c the length of M. spicatum main shoots (cm) and d the number of M. spicatum autofragments. The coefficients of determination (R 2) and statistical evaluation of the surface planes for the different variables can be found in Tables 1 and 2, respectively. N = 14 for panel a, and N = 16 for all other panels

Full size image

Myriophyllum spicatum

The results for M. spicatum are summarised in Table 2. Growth was strongly affected by C. globularis, even at low densities, indicated by the negative interspecific competition coefficients. We found a much lower biomass at high C. globularis density with reductions in RGR of 42 % (Fig. 1b), and reductions in shoot length of up to 70 % (Fig. 1c). No intraspecific negative effects were observed. All monospecific treatments produced autofragments, while this dispersal mechanism was almost completely inhibited in the presence of C. globularis (Fig. 1d).

Table 2 Results of stepwise linear regression for growth parameters and functional traits of M. spicatum based on the model PeindMs = Pe0 (± 1 SE) − DMs × fMs (± 1 SE) − DCg × fCg analogous to the model presented in “Methods” section, and with Pe representing the respective performance parameters as listed below
Full size table

Based on dry mass, the C and N content of leaves at high densities of C. globularis decreased by 11 and 30 %, respectively, resulting in an increase of the C/N ratio by 33 %. The leaf phosphorus content was low (0.21–1.40 mg g−1 dm), but not different between treatments. Thus, the molar N/P ratio of M. spicatum leaves was not affected. The lower chl a content resulted in a slight decline of the chl a/b ratio by 9 % at high C. globularis densities. Total chlorophyll concentration correlated with leaf nitrogen content (Pearson′s correlation: r = 0.498, P = 0.025, n = 16), but not with leaf phosphorus content (r = 0.025, P = 0.464, n = 16).

The ash content was exceptionally high, ranging from 71 % in monocultures to 75 % at high densities of C. globularis, but differences among treatments were not significant. If we calculate the C, N, P and chl content based on the AFDM, the effects on C and chl disappear, but the decline in N remains. Neither high density of M. spicatum nor C. globularis influenced the leaf-matter fraction or root/shoot ratio of M. spicatum. In contrast, the dry-matter content of M. spicatum increased dramatically by up to 43 % at high C. globularis densities.

Leaves of the autofragments had high C, N and P contents when compared to main shoots, with no difference among treatments (327 ± 8 mg C g−1 dm; 24 ± 2 mg N g−1 dm; 2 ± 0.2 mg P g−1 dm; n = 4). We could not determine differences between mono and mixed cultures because only one single, small fragment was found at low C. globularis density, with C, N and P concentrations of 425, 44 and 4 mg g−1 dm, respectively.

Discussion

Predicting changes in submerged macrophyte community composition following changes in environmental conditions is challenging, but also valuable for several reasons. The knowledge about plant community structure allows us to assess possible ecological interactions and ecological functions of the species present, and aids in the decision on possible management options, specifically if the lake offers economic and recreational benefits for the region. Fortunately, at least some lakes have returned to a good ecological status, which provide good growth conditions for submerged macrophytes (Jeppesen et al. 2005; Koehler et al. 2005; Stich and Brinker 2010). Usually, tall-growing macrophytes outcompete charophytes (Blindow 1992), but recent efforts in nutrient reduction offered new opportunities for charophytes (Stewart and Lowe 2008; Schmieder et al. 2006), which are more competitive under clearwater and hardwater conditions than in turbid waters.

We started this project because our field observations indicated strong nutrient limitation in M. spicatum (Gross 2009), associated with an increase in charophytes, particularly C. globularis. We developed an experimental set-up mimicking field conditions as closely as possible, and were able to elucidate key factors and effects of the interaction among this tall-growing macrophyte and Chara. Although both species are well known for their allelopathic effects on phytoplankton (Wium-Andersen et al. 1982; Gross et al. 1996), allelopathic interactions are very unlikely both in the lake and in our experiment due to the high water-exchange rates. Water chemistry and nutrient concentrations remained constant during the course of the experiment, and did not differ between treatments. Similar conditions occur in our field sites, because this part of Lower Lake Constance is strongly influenced by the through-flow of the river Rhine. Using the response surface experimental design (Inouye 2001), we revealed that not only the mere presence of C. globularis but also its density is important for the observed M. spicatum decline. We thus conclude that in addition to the re-oligotrophication of Lake Constance, the decline of M. spicatum in the littoral is significantly enhanced when Chara is present.

Chara globularis

Growth of C. globularis was never affected by the presence of M. spicatum, even at the highest densities of the angiosperm. However, we observed strong intraspecific competition, comparable to that observed for C. aspera in a similar competition study with Potamogeton pectinatus, where nutrient depletion in the water column may have been the limiting factor (Hidding et al. 2010). Although in our study nutrients in the water column were very low, down to the detection limit for SRP, the constant water turnover provided about 25 mg SRP per mesocosm during the experiment, which is sufficient to account for the phosphorus content found in C. globularis tissue. Although competition via sediment-based nutrients was previously considered negligible for Chara (Kufel and Kufel 2002), recent findings indicate that Chara can grow when only sediment phosphorus is provided (Wuestenberg et al. 2011). Good growth at low nutrient availability indicates that Chara might have more efficient nutrient uptake systems than many tall-growing angiosperms (Bakker et al. 2010; this study).

This is further supported by the fact that the biomass development of C. globularis in our study was comparable to field data in Lower Lake Constance (Schmieder et al. 2006; Richter and Gross, unpub. data) and that ash content as well as and N and P concentrations were similar to different natural systems (Blindow 1992; Kufel and Kufel 2002). Thus the environmental conditions were appropriate for C. globularis in our system. A possible reason for the decline in RGR at higher C. globularis densities could have been competition for space, since lateral growth was limited by the tubes and longitudinal growth interferes with sufficient light availability for the lower shoots in the charophyte mats.

Myriophyllum spicatum

As predicted, and expected from our field observations, interspecific competition resulted in strong negative effects on M. spicatum. High densities of C. globularis strongly reduced the RGR and total length of M. spicatum and basically prevented dispersal by autofragmentation. In addition to these indications of reduced fitness, we also observed significant effects on plant structural quality, specifically its toughness and nutrient content. We noticed an exceptionally high ash content of 71 % in the mono- and 75 % in the mixed cultures, although this slight increase was only marginally significant. A possible effect of C. globularis on the ash content could have been masked by the already high ash-content of the plants used in the experiment. During recent years, we also observed an increase in the ash content in M. spicatum from Lower Lake Constance from 40 to 70 % (Gross, unpub. data). Unfortunately, only a few studies report ash contents for M. spicatum; normal values seem to be well below 30 % (Grace and Wetzel 1978). Myriophyllum alterniflorum in three Spanish ponds had ash contents between 12 and 23 % (Fernandez-Alaez and Fernandez-Alaez 2011), but this species does not use bicarbonate like M. spicatum, making a comparison difficult. What exactly caused the increase in ash content remains unclear. We did not observe carbonate encrustations on M. spicatum plants comparable to those on C. globularis, and also other loose deposits should have been washed away during the cleaning process. Rather, we consider that it may be due to changes within the cell wall of the plant, perhaps changes in silica content, but this remains to be verified by further studies.

The proportion of dry to wet biomass (dry-matter content, DMC) is considered an important plant trait, both for terrestrial and aquatic plants (Garnier et al. 2004; Elger and Lemoine 2005). The DMC of M. spicatum increased significantly by 43 % in the presence of C. globularis, while intraspecific competition had no effect. A higher DMC is associated with higher tissue toughness, and can also be caused by herbivory (Elger and Lemoine 2005). The higher DMC may reflect the need for more structural tissue (Hyldgaard and Brix 2011), which supports our findings on the ash content. We consider that leaf DMC is comparable to the DMC of the entire shoot for submerged macrophytes, because the DMC consists of structural components such as cell-wall structure and mineral inclusions (Wilson et al. 1999). Growth in water requires minimal investment in structural components, thus resulting in a rather low DMC of freshwater macrophytes compared with terrestrial leaves (Elger and Willby 2003). Thus, the difference between leaf and stem DMC in M. spicatum should be negligible. The functional trait leaf DMC is used as an approximation of leaf tissue density, mostly leaf constituents such as lignin, fibre and silica contents, and negatively correlated with nutrient content and thus also RGR (Elger and Willby 2003; Garnier et al. 2004). All measured traits for M. spicatum indicate that this macrophyte is severely stressed in Lower Lake Constance.

We calculated C, N and P as well as chlorophyll content for M. spicatum both on dry mass (dm) and ash-free dry mass (AFDM), because only a few studies explicitly refer to the latter. Values for C, N and P were in agreement or even lower than those found previously in M. spicatum in Lower Lake Constance (Gross 2009; Hempel et al. 2009). The very low leaf carbon content of 19–21 % based on dm was even reduced at high C. globularis densities. Yet, carbon content did not differ when calculated on AFDM. This suggests that carbon content based on organic biomass (AFDM) is a fixed value of approx. 40 %. Thus, the proportion of physiological active cell components to structural components (ash content, DMC) was negatively affected by C. globularis. In general, the elemental composition of organisms reflects the macromolecular composition of cells (Elser et al. 2011), which in turn reflects the integration of the various processes involved in nutrient uptake and assimilation.

The leaf nitrogen content based on dm of M. spicatum ranged between 0.8 and 1.6 %, and declined at a high density of C. globularis; the effect remained based on AFDM. The N content was even lower than found before in situ (Hempel et al. 2009), but was still above the critical level of 0.75 % suggested by Gerloff (1975). The leaf phosphorus content was low but not affected by C. globularis. Values were in agreement with field data (Hempel et al. 2009; Gross 2009) but much lower than found in other studies (Gerloff 1975; Carpenter 1980; Schmitt and Adams 1981). The N/P molar ratio in M. spicatum was high and ranged between 33 and 110 in the presence of C. globularis, indicating P limitation (Loladze and Elser 2011). Plants on soils poor in phosphorus increase the toughness of leaves (Hidaka and Kitayama 2011), again providing evidence for a strong interaction of the various cell components as stated above. The C/N molar ratio even increased at high C. globularis densities, indicating a stronger effect on nitrogen than on carbon. Changes in the C/N ratio might imply decreased nitrogen availability (Ventura et al. 2008). It is possible that M. spicatum did not invest much in the light-harvesting machinery and also limited carbon fixation since leaf N is correlated to the photosynthetic activity (Blindow et al. 2003; Wang et al. 2008). Shading by C. globularis might have resulted in the decreased productivity of M. spicatum. This is supported by the weak but positive correlation between N and total chlorophyll, and a lower chl a/b-ratio in the presence of C. globularis, indicating acclimation to lower irradiance (Lichtenthaler 1987). However, leaf-mass fraction (Cornelissen et al. 2003) was unchanged between treatments, suggesting that M. spicatum was not capable of investing in a proportional increase in leaf tissue.

High root/shoot ratios are typically associated with infertile environments (Chapin 1980). The high root/shoot ratio found for M. spicatum compared to other studies (Knauer et al. 2008) indicates low nutrient availability. However, the ratio was not affected by C. globularis density, in contrast to findings for P. pectinatus competing with Chara (Hidding et al. 2010). We consider two possible explanations for the enhanced nutrient limitation in the presence of C. globularis: shading might have reduced sediment nutrient uptake because low light causes a decline in N and P uptake (Gross 2003; Wang et al. 2008), or shoot competition for nutrients may have contributed to the lower leaf nutrient contents (Wang et al. 2008).

Leaf nitrogen content and dry-matter content affect components of the carbon and/or nitrogen cycles at the leaf, whole-plant and ecosystem level (Garnier et al. 2004): ecosystems composed of slow-growing species with low leaf nitrogen but high dry-matter content producing low litter quality will show slow decomposition rates. High C. globularis densities resulted in a lower tissue quality indicated by low leaf nitrogen and dry-matter contents in M. spicatum, suggesting effects at the ecosystem level. Based on our results, we strongly suggest more studies taking into account variables such as plant ash and dry-matter content, specifically when focusing on stoichiometry and physiological activity. Abscised M. spicatum fragments had higher leaf carbon, nitrogen and phosphorus contents than main shoots. These autofragments were well equipped for dispersal, but M. spicatum could not benefit from this vegetative dispersal mechanism in the presence of C. globularis. Competition for space and nutrient limitation might have reduced autofragmentation (Wang et al. 2008).

In conclusion, a striking result of our study is that C. globularis benefits from nutrient-poor conditions and also strongly affects the performance of M. spicatum and possibly other tall growing angiosperms (Hidding et al. 2010). Our results may apply to regions where M. spicatum is an invasive species and re-oligotrophication is possible thus offering new possibilities for its biological control via Chara (Nichols 1991).