• High radial oxygen loss (ROL) from roots of aquatic plants to reduced sediments is thought to deplete the roots of oxygen and restrict the distribution of those species unable to form a barrier to oxygen loss. Metal precipitates with high iron content (Fe-plaques) frequently form on roots of aquatic plants and could create such a diffusion barrier, thereby diverting a larger proportion of downward oxygen transport to the root meristems.
• To investigate whether Fe-plaques form a barrier to oxygen loss, ROL and internal oxygen concentrations were measured along the length of roots of the freshwater plant Lobelia dortmanna using platinum sleeve electrodes and Clark-type microelectrodes.
• Measurements showed that ROL was indeed lower from roots with Fe-plaques than roots without plaques and that ROL declined gradually with thicker iron coating on roots. The low ROL was caused by low diffusion coefficients through root walls with Fe-plaques resulting in higher internal oxygen concentrations in the root lacunae.
• By diverting a larger proportion of downward oxygen transport to root meristems in L. dortmanna, the presence of Fe-plaques should diminish root anoxia and improve survival in reduced sediments.
Aquatic plants usually grow in reduced sediments. Continuous oxygen supply from photosynthesis or the water column is therefore crucial to satisfy the respiratory demand of the roots (Smits et al., 1990; Colmer, 2003). While gas diffusion is slow in water and water-filled tissues, it is rapid through the aerenchyma in leaves, stems and roots, which most aquatic plants possess (Colmer, 2003). Some species are even plastic and respond to more reduced sediment conditions by increasing tissue porosity, or by forming a diffusive barrier to radial oxygen loss (ROL) from root surfaces (Smits et al., 1990; Visser et al., 2000; Soukup et al., 2007).
At every point along the root pathway there is a competition between consumption in respiration, diffusion down the root and diffusion out to the sediment, often creating an oxygenated rhizosphere (Visser et al., 2000; Colmer, 2003). Maintaining an oxygenated rhizosphere is assumed to prevent uptake of potential phytotoxins by oxidizing them to harmless compounds before entering the plant tissue (Armstrong, 1971; Pedersen et al., 2004).
Species with extensive root systems that cannot form barriers to oxygen loss, such as the small oligotrophic species of rosette growth form, the isoetids (Smolders et al., 2002), are vulnerable to strongly reduced sediments or sediments of varying oxygen demand (Smits et al., 1990). Reduced sediments result in shortening of their roots, which leads to weak anchoring (Sand-Jensen et al., 2005a). The successful growth of isoetid plants is therefore strongly dependent on the sediment conditions and many of their former growth habitats have been lost by eutrophication (Smolders et al., 2002; Sand-Jensen et al., 2005a).
Several investigators have addressed the effect of Fe-plaques on uptake of nutrients and protection against phytotoxins from the sediment (Begg et al., 1994; Saleque & Kirk, 1995; Sand-Jensen et al., 2005a). To our knowledge, however, the effect of Fe-plaques on the oxygen dynamics in plants is unknown. We hypothesize that the crystalline nature of the Fe-plaques could create a barrier to oxygen loss from the roots and, thereby, elevate internal oxygen concentrations in the root lacunae and enable a larger proportion of the oxygen supply to reach the meristems of the root apices. Thus, Fe-plaques could provide some protection against reducing sediment conditions by permitting unimpeded aerobic respiration in the meristems characteristic of high oxygen demand. In the present investigation ROL and internal oxygen concentrations were, therefore, measured along the length of roots with and without Fe-plaques for the iseotid plant Lobelia dortmanna L. to address the two questions:
• Do Fe-plaques reduce the ROL from roots of L. dortmanna by establishing a diffusion barrier to ROL?
• If Fe-plaques reduce the ROL from the roots does this result in elevated oxygen concentrations at the root tip compared with roots without plaques?
Materials and Methods
Plants and growth conditions
Two sediment turfs each containing c. 20 plants were collected from a homogeneous Lobelia-population in Lake Värsjö, Sweden. Sediment turfs were circular (diameter 30 cm) and had a sediment depth of 15 cm that ensured intact root systems. In the laboratory, one turf was cultured submerged in an aquarium at 16°C exposed to a 14 h light (65 µmol photon m−2 s−1) and 10 h dark cycle. These plants were expected to maintain oxygen in the pore-water to c. 40 mm depth in the sediment in the light (Sand-Jensen et al., 2005b). Therefore, Fe should be present as insoluble Fe3+ in surface sediments and no Fe-plaques should form on the root surfaces. The other turf was cultured in continuous low light (7 µmol m−2 s−1), which was expected to reduce radial oxygen loss (ROL), lead to reduction of precipitated Fe3+ to soluble Fe2+ in the sediment and subsequent diffusion to and reprecipitation of Fe3+ on root surfaces. Turfs were cultured for 4–6 wk until relatively homogeneous reddish plaques had formed on plants kept at low light, while plants at high light remained without visual plaques. Because this setup was used selectively to produce either homogeneous plaques or no plaques on roots of approximately the same length and diameter, no replicates of the number of turfs were made since roots had to be sampled according to these criteria.
Oxygen loss from roots
The ROL along roots was measured using the cylindrical platinum sleeve electrode technique described by Armstrong (1967) and Armstrong & Wright (1975). Root measurements were performed in a cylindrical glass chamber (height 8.0 cm; diameter 7.7 cm) filled with a stagnant deoxygenated solution containing 0.05% agar (by weight) and 5 mm KCl to ensure electrical conductivity. Before measurements, roots were gently rinsed out of the sediment, cut off close to the stem with a razor blade and cleaned of sediment particles. A single root was then transferred to the deoxygenated solution held in position by inserting the base of the root through a piece of expanded polystyrene mounted on the lid of the setup thereby exposing the root lacuna and upper 1 mm of the root to atmospheric air and submersing the rest in the deoxygenated solution. The root was left for at least 1 h to acclimatize to experimental conditions. A cylindrical platinum oxygen electrode (height 5 mm, internal diameter 2.5 mm) was then positioned around the most basal part of the roots and moved stepwise towards the tip after a stable reading had been recorded. Measurements were performed at a low temperature of 3.2 ± 0.3°C (mean ± SD) to minimize the influence of root respiration on oxygen loss (see later) and stabilize the readings. Low temperatures of 0–6°C are typically found for almost half of the year (Nov–Apr) at the growth site. Replicate measurements were taken on three roots without Fe-plaques and three roots with Fe-plaques with similar thickness. One of the three roots with Fe-plaques had no visual plaque on the upper 1 cm and measurements from this part of the root were therefore discarded from calculations of mean values for roots with Fe-plaques.
The measurements on single roots with the upper part exposed to atmospheric air permitted highly reproducible and comparable measurements among roots of either no or homogeneous Fe-coating. The oxygen level is also representative of the expected level in field populations of Lobelia during periods of high irradiance (Fig. 2 in Sand-Jensen et al., 2005b).
Oxygen concentration in root lacunae
Oxygen concentration in the large lacuna of the roots studied for ROL was measured by inserting a 25 µm Clark-type oxygen electrode (OX 25; Unisense, Arhus, Denmark) in the lacunae at 6–10 locations distributed evenly along the length of the roots. Measurements were made in another chamber holding a similar deoxygenated solution at virtually the same temperature (3.8 ± 0.4°C) as used for determination of ROL. During experimentation, roots were attached to a plastic mesh every 1 cm and with the basal part of the root and the lacuna exposed to atmospheric air. The microelectrode was moved and positioned with an accuracy of 1 µm using a micromanipulator. The electrode was calibrated at the temperature of experiments in atmospheric air and N2 gas before and after experiments and the calibration remained constant.
Root wall resistance
Diffusion coefficients were used as a measure of resistance in root walls. Respiration of roots was assumed not to affect the calculated values since the experiment was conducted at low temperatures with very low rates of respiration relative to diffusion. Diffusion coefficients of the root walls were calculated from measurements of ROL, lacunal oxygen concentration and root wall thickness. Regressions of root wall thickness (see root anatomy) and lacunal oxygen concentration as a function of distance from the root tip were used to interpolate values at exactly the points along the roots where ROL had been measured.
To calculate the diffusion coefficient of the root walls, first oxygen concentration at the root wall surface at temperature t (Ct,wl, mol O2 m−3) was calculated from the equation (Armstrong & Wright, 1975):
(O2f (mol m−2 s−1) is the oxygen flux from the root surface; Dt,w (10−5 m2 s−1) is the diffusion coefficient in water at temperature to; rr (m) is the radius of the root within the electrode; and re (m) is the inner radius of the electrode). Afterwards, diffusion coefficients for the root walls (Dt,r, 10−11 m2 s−1) were calculated from the equation:
(O2f is the ROL (mol m−2 s−1); rr-w (m) is the radius of the root minus the root wall thickness; rr (m) is the radius of the root; and is the difference between oxygen concentration (mol O2 cm−3) at the outer and inner root surface). Calculations were made for diffusion coefficients of oxygen at 3.5°C (mean of 3.2 and 3.8°C, see earlier). Oxygen concentration at the inner root surface (liquid phase) was assumed to hold the same oxygen saturation as measured in the gas-filled lacuna. Diffusion coefficients through the root wall was set equal to diffusion in water in accordance with the high water content of the wall (see below) and assumes no turtuosity.
Net oxygen loss from root systems and respiration
Net oxygen loss from entire root systems to hypoxic water was measured by tracking changes in oxygen concentration with an oxygen microelectrode in a Perspex cylinder (volume 75 ml) sealed off from the atmosphere. Plants were mounted through a hole in the lid of the chamber placing the roots inside and the leaves outside the chamber. A seal was then secured with silicone grease, ensuring separation between the air surrounding the leaves and the water around the roots. Leaves were subsequently cut off exposing the lacunae to atmospheric air. A magnetic stirrer mixed the root chamber. The oxygen concentration was lowered to 5–10% of atmospheric saturation before measurements by pumping deoxygenated water into the root chamber via a peristaltic pump and gas-tight tubes from a reservoir. In each series of measurements on a root system, three measuring periods were used interrupted by three flushing periods. Net oxygen loss from root systems was calculated by linear regression of the rate of oxygen rise in the chamber with time normalized to the surface area of the root system. An average of the three measurements was used in later calculations. Measurements were performed at 4.3 ± 0.3°C. Replicate measurements were taken on three root systems with no Fe-plaques and three root systems with Fe-plaques.
Respiration of roots was measured by placing cut-off root systems completely submerged and isolated from the atmosphere in the root chamber in aerated water in the dark at 4.2 ± 0.4°C. The rate of decline of oxygen concentration was then recorded. The respiration rate was expressed per unit time and dry weight (DW) and later per surface area of the root system by applying the measured surface area to DW ratio. Replicate measurements were made on two root systems without and two root systems with Fe-plaques.
Iron contents of roots and surface plaques
Iron content of the roots was determined by the phenanthrolin method with slight modifications (Eaton et al., 1995). Roots were dried at 105°C for 24 h and combusted at 550°C. Iron was then extracted from the ash in a 0.5 m HCl solution containing hydroxyl-ammonium-chloride. Samples were boiled to convert insoluble Fe3+ to soluble Fe2+. Samples were diluted to bring Fe2+ concentration within the optimum measuring range of the analysis (2–40 µmol Fe2+ l−1). Dilution of the individual samples was based on previous tests with similar root mass and iron content and ranged from no to 100 times dilution. Phenanthrolin was added in excess, and the sample was left for 15 min to form the iron-phenanthrolin colour complex. Absorbance was measured in a spectrophotometer at 510 nm and converted to iron content from similarly treated standards of known iron content. Iron concentration was normalized to root surface area and compared with the ash weight of entire roots.
Root surface area was determined by measuring the length and diameter of roots kept in water. Root length was measured to the nearest mm and the diameter was measured with an accuracy of 10 µm under a dissection microscope at ×92 magnification for every 5 mm along the roots. Root systems with Fe-plaques were not completely covered with plaques. Therefore, when measuring the length and thickness of these roots the surface areas with and without Fe-plaques were determined.
Root wall thickness and porosity was measured on 10 roots of similar lengths and thicknesses as the roots used for measurements of ROL from single roots. Numerous hand-made cross-sections along the length of each root were prepared and wall thickness and the percentage of the cross section area occupied by lacuna was measured in Olympus DP-soft CVI on digital images taken by an Olympus ColorView imaging camera mounted on a light microscope (Olympus BX 41, at ×40–100 magnification depending on root diameter). Only cross-sections not deformed during the cutting procedure were used. The quotient of root surface area to DW was calculated using DW and root surface area from three root systems with and three without Fe-plaques. The water content of roots without Fe-Plaques was 89 ± 1% of the dry weight (n = 15). Roots without plaques had 116 ± 13 and root with Fe-plaques had 91 ± 14 m2 root surface kg−1 root DW.
Data analysis was performed in Graph Pad Prism 4. Significance levels below 0.05 were considered significant. When data failed to pass the F-test for variance homogeneity, data were square-root transformed to meet test criteria. Data are presented as mean ± 1 SD.
The lacunae were well developed in the examined roots (Figs 1 and 2). The porosity was high in the upper part of the roots and fairly constant (c. 60% of the cross section area) until 20 mm from the root tips, while porosity fell rapidly towards the tip.
The root wall thickness (y, mm) increased as a function of root radius (x, mm) following the regression line: y = 0.10x + 0.018 with a slope significantly different from zero (P < 0.0001, r2 = 0.77). This equation was used to estimate root wall thickness when diffusion coefficients of root walls were calculated.
Roots without Fe-plaques used for ROL measurements were 62 ± 1 mm long and slightly thicker than the 57 ± 2 mm long roots with Fe-plaques. The diameter (y, mm) as a function of distance from the root tip (x, mm) increased towards the basal part of the roots following the regression lines: y = 0.0078x + 0.31 (r2 = 0.93) and y = 0.0075x + 0.25 (r2 = 0.92) for roots without and with Fe-plaques, respectively.
Radial oxygen loss
Radial oxygen loss occurred over the full length of L. dortmanna roots. For roots without Fe-plaques, ROL normalised to root surface area increased towards the tips, while roots with Fe-plaques had lower and variable rates along the roots (Fig. 3a). Since roots were thicker towards the base, ROL per cm of root was highest in the basal regions and decreased towards the tips (Fig. 3b). ROL from single roots with Fe-plaques was significantly lower (about twofold) than from single roots without Fe-plaques (t-test, P < 0.01, Table 1). ROL from entire root systems was slightly though not significantly higher than the corresponding values in single root measurements (Table 1).
Table 1. Comparison of radial oxygen loss (ROL), iron content (Fe) and diffusion coefficients of root walls (Drw) measured on three single roots and three root systems of Lobelia dortmanna with (+) or without (–) Fe-plaques
Visual Fe-plaque present
All measurements are means ± 1 SD.
ROL (nmol O2 m−2 s−1)
1016 ± 260
787 ± 130
849 ± 36
453 ± 99
Fe (mmol Fe m−2 root surface)
0.05 ± 0.04
16 ± 3
0.09 ± 0.05
30 ± 3
Drw (m2 s−1 × 10−11)
68 ± 20
7.8 ± 2
Root respiration was low compared to ROL and the difference between respiration of roots with Fe-plaques (2.5 ± 1.3 µmol O2 kg−1 root DW s−1) and without Fe-plaques (1.8 ± 0.4) was insignificant. Respiration of roots expressed per unit of surface area was not significantly different between roots with (23 ± 12 nmol O2 m−2 root surface s−1) or without (29 ± 7 nmol O2 m−2 root surface s−1) Fe-plaques. The respiration rate was only 5.0% and 3.4% of the measured ROL from roots with and without Fe-plaques, respectively.
Oxygen concentrations in lacunae
Oxygen concentrations in root lacunae decreased linearly towards the tips when roots were submerged in a deoxygenated solution (Fig. 4). Roots with Fe-plaques had elevated oxygen concentrations in root lacunae compared to roots without Fe-plaques. Regression coefficients of oxygen concentrations (kPa) in the lacunae as a function of distance from the root tip (mm) revealed significantly different slopes of the regression lines with mean values of 0.077 ± 0.004 kPa mm−1 for roots without Fe-plaques and 0.050 ± 0.0097 kPa mm−1 for roots with Fe-plaques (t-test, P < 0.01).
Diffusion across the root wall
Diffusion coefficients across root walls were more than eight-fold lower (t-test, P < 0.0001) when Fe-plaques were present (7.8 ± 0.2 × 10−11 m2 s−1) than when plaques were absent (68 ± 20 × 10−11 m2 s−1, Table 1, Fig. 5). Diffusion coefficients of root walls along the length of roots showed variable patterns for roots without Fe-plaques while roots with Fe-plaques had uniformly low diffusion coefficients. One root with no visual plaque in the basal 1 cm of the root, but otherwise fully covered with a Fe-plaque further down the root, had similar diffusion coefficients in this basal part as those measured in roots without Fe-plaques further supporting the notion that Fe-plaques were responsible for the difference in diffusion coefficients (Fig. 5).
Iron, other minerals and oxygen loss
Roots with high iron content had smooth reddish plaques on root surfaces (Fig. 6). The plaques appeared to be located on, or in, the outer epidermis layer, and did not increase the diameter of the roots appreciably.
Iron content per surface area of roots without visual Fe-plaques was 350-fold lower compared to roots with plaques (Table 1). Iron content of root systems with Fe-plaques was lower than the iron content of single roots with Fe-plaques. This was because only 50–80% of the surface area on entire root systems was covered with plaques, probably due to heterogeneity of the sediments.
Roots without Fe-plaques had, on average, 8 ± 2% ash content of the DW while roots with Fe-plaques had 40 ± 8% ash content of the DW. Iron constituted only 0.6 ± 0.4% of the ash content of roots without Fe-plaques, while it constituted 32 ± 11% in roots with plaques.
The ROL from roots was closely related to the iron content of the roots, with rates declining significantly with the iron content of single roots and entire root systems (Pearsons, P < 0.01; Fig. 7). Measurements on single roots and entire root systems fell on the same line implying that detailed measurements on single roots reflect the behaviour and magnitudes of entire root systems.
Oxygen loss from roots
High ROL occurred over the full length of L. dortmanna roots. This pattern was expected because L. dortmanna releases up to 100% of the photosynthetically produced oxygen across root surfaces and relies on sediment carbon dioxide for photosynthesis, hence, resistance to gas diffusion in the root walls is low (Sand-Jensen et al., 1982). Furthermore, high porosity along the length of L. dortmanna roots facilitates rapid diffusion of oxygen from the basal part, which was exposed to atmospheric air in the experiment with single roots, towards the root tip securing high lacunal oxygen concentrations along the full length of the roots (Figs 2 and 4). Similar ROL patterns have been found for other isoetid species (Smits et al., 1990).
The ROL from root surfaces without Fe-plaques increased slightly from the basal regions towards the root tips. Since oxygen concentration decreased towards the tips, lower root wall resistance towards the tip must account for elevated ROL. This resistance could either be the result of thinner root walls or higher gas permeability of root walls towards the tip, but it is not possible to discriminate with certainty between the two possibilities because of the substantial variability of diffusion coefficients. However, the linear decrease in wall thickness towards the root tips offers a straightforward explanation for the increase of diffusion of oxygen across the root wall.
The ROL from entire root systems was comparable to rates from single roots implying that little resistance was offered by the stem, which was present in measurements on root systems. The measured rates should also be close to those expected in situ since oxygen concentrations in the lacunae in the leaves and basal parts of the roots are close to air saturation (Sand-Jensen et al., 2005b). The slightly higher ROL measured on entire root systems without Fe-plaques was expected since stirring of the water in the root chamber probably reduced the thickness of the diffusive boundary layer surrounding the roots, causing steeper diffusion gradients to develop over the root walls and, in turn, supporting higher ROL compared with the experiments with single roots held in a stagnant solution.
Reduced ROL from roots with Fe-plaques could in theory be caused by a greater diffusive barrier in roots with Fe-plaques or alternatively high oxygen demand in the Fe-plaques. Since respiration was very low compared with the ROL and no chemical oxygen demand was expected during the experiments, with no reduced iron and manganese being available, the decreased ROL must certainly be caused by a greater diffusive barrier. Furthermore, the elevated oxygen concentrations in lacunae of roots with Fe-plaques imply that a diffusive barrier to ROL was present, while the opposite would be the case if high respiration or chemical oxidation in the plaque reduced ROL (see later).
A diffusive barrier induced by the plants as a response to anoxia, has in all previous studies on different plant species been located in the basal part of the roots, while ROL still took place near the root tips (Colmer, 2003). In the present study on L. dortmanna, ROL occurred over the full length of roots with Fe-plaques showing that ROL was lowered by the Fe-plaques per se and not by an internal barrier formed within the roots of a special organic compound. The ROL was closely negatively correlated to the Fe-content of roots, which further implies that plaques were responsible for the reduction of ROL, while formation of an internal tissue barrier of say lignin or suberin would lead to lower ROL, and thereby less Fe2+ oxidation to Fe-plaques on root surfaces opposite to the pattern actually observed (Mendelssohn et al., 1995). Moreover, the fact that absence of Fe-plaques along part of the length of an otherwise Fe-coated root immediately led to the same ROL as from roots without any Fe-plaques (Fig. 3a) suggest that no internal tissue barrier was formed despite the root having been exposed to a reducing environment during its development causing Fe-plaques to be formed in the more distal parts.
Fe-plaques and internal oxygen
Roots with Fe-plaques had higher oxygen concentrations in the lacunae than roots without plaques; this could result from either increased root porosity or a stronger radial diffusion barrier. Plants that increase tissue porosity in reduced sediments will have a higher ROL when oxygen supply to the lacunae is unrestricted as in the present investigation (Colmer, 2003). Fe-plaques were associated with low ROL and higher oxygen concentrations in the lacunae and, therefore, support the prediction of a barrier to ROL. The elevated oxygen concentration in the lacunae is an interesting finding confirming that Fe-plaques, by creating a diffusive barrier to ROL, conduct a larger proportion of the oxygen flux towards the root tip.
The plaques did not result in a very large increase in internal oxygen and as such support some earlier predictions (Armstrong, 1979) that show that barrier formation should not have an enormous effect on internal oxygen concentrations provided that roots are highly aerenchymatous and, hence, very low internal longitudinal diffusive resistance, as is the case for L. dortmanna. In our measurements at a very low temperature and with minimal root respiration and with full air saturation at the basal part of the root, internal oxygen concentrations only declined to c. 80% of air saturation at 10 mm from the tip of roots with no formation of Fe-plaques. The influence of Fe-plaques will be stronger in the dark and with reduced oxygen concentrations in the water because internal concentrations in the lacunae will then fall below air saturation. The influence will be enhanced at higher summer temperatures where respiratory oxygen consumption along the diffusion path will be about fourfold higher (applying a typical Q10-value of 2) and lead to a steeper decline of oxygen towards the root tip. We know that the roots do indeed reduce their length from c. 70–100 mm to only 15–20 mm at typical summer temperatures as the concentrations of labile organic matter in the sediments increase (Sand-Jensen et al., 2005a). The roots show clear signs of stress as long roots die if the reducing capacity of sediments is suddenly enhanced by inserting pellets of labile organic matter and the often short roots replacing them develop very thick coatings of Fe or Mn (Raun, 2008). The protective role of these metal-plaques can be twofold by improving oxygen supply to the root meristem and reducing the ingress of phytotoxins from the sediment.
Diffusion coefficients and exchange of oxygen
Diffusion coefficients of root walls without Fe-plaques showed different patterns along the roots and were associated with some error, because coefficients higher than in water were found in three cases (Fig. 5). They may arise from differences in morphology among the roots examined, which leads to overestimates of the root wall thickness when calculating diffusion coefficients. Direct measurements of wall thickness of roots used for experiments are therefore recommended. With high diffusion coefficients of only half the value in water, ROL could rapidly deplete the root lacuna in oxygen when the sediment oxygen demand is high and no oxygen source is available.
Root surfaces with Fe-plaques had nine times lower diffusion coefficients than roots without plaques showing that plaques can have a profound influence on gas exchange between roots and sediment (Table 1, Fig. 5). However, roots of aquatic plants in their natural habitat are surrounded by sediment, which also offers diffusive resistance. Therefore, if diffusion coefficients of sediments are as low as for root walls with Fe-plaques, plaques will have no effect on diffusion of gases from the roots to the sediment. To evaluate the resistance in roots with Fe-plaques compared with that of sediments, sediment diffusion coefficients were estimated using porosity data from typically sandy sediments inhabited by L. dortmanna (Pedersen et al., 1995), and a model of the influence of porosity on sediment diffusion coefficients (Ullman & Aller, 1982). The estimated diffusion coefficients of Lobelia sediments ranged between 51 and 89 × 10−11 m2 s−1, similar to the values obtained for the root wall without Fe-plaques, but 7–11 times higher than those of root walls with Fe-plaques. Elevated oxygen concentrations in the lacunae of roots with Fe-plaques are, therefore, also expected to occur under natural conditions in the field.
We thank Ole Pedersen for technical advice, Jens Borum for comments on the manuscript and Tim D. Colmer for lending us sleeve electrode equipment and for comments on the manuscript. We thank three anonymous reviewers for insightful comments, which improved the paper. This project was part of the CLEAR-project (lake restoration centre) funded by the Willum Kann Rasmussen Foundation.