The extent of bioturbation activity by oligotrophic macroinvertebrates in the Lake Marken, The Netherlands

(1)

The extent of bioturbation activity by oligotrophic

macroinvertebrates in the Lake Marken, The Netherlands

Bsc thesis – Final report

University of Amsterdam

Lake Marken indicated in the Netherlands Macroinvertebrates during the experiment at the sediment-water interface

Esther Bos 10543171

Course Bachelor project Biology Supervisor: M. van Riel, Alterra Examiner: H. van der Geest, UvA

Word count 8869

(2)

Abstract

Lake Marken is a large oligotrophic lake with a marine origin and enclosure by dykes has caused the accumulation of a thick anoxic silt layer. Bioturbation can play an important role to enhance nutrient bioavailability at the lake and ultimately enhance its’ biodiversity. This research focusses on identifying the bioturbation process at Lake Marken, with a focus on nutrient release through sediment oxygenation. Laboratory experiments were conducted to visualise pore development and excretion layer development. Additionally, a vertical profile of dissolved oxygen in a bioturbated microcosm was made. The research showed that macroinvertebrate make burrows up to 14cm depth, which are continuously made and vacated. Bioturbation increased dissolved oxygen concentrations up to 14mm depth. These results emphasize the need of further research on the effect on solute fluxes at Lake Marken.

(3)

Introduction

Aquatic ecosystems and wetlands are under huge threat worldwide: it has been estimated that 64 to 71 percent of wetlands have declined and both decline and degradation is continuing (van Eerden & van Rijn 2003, retrieved from Padro et al., 2012; Zedler & Kercher, 2005; Ramsar, 2015). Fortunately, more and more incentives are in place to restore aquatic ecosystems (Zedler & Kercher, 2005). Wetland restoration or creation is conducted frequently to mitigate aquatic ecosystem decline and degradation (Mitsch et al., 1998). However, changes in aquatic ecosystems’ quality is hard to track due to missing ecological data and approaches to restore or create wetlands and aquatic ecosystems have been controversial due to limited knowledge and limited available ecological data (Zedler & Kercher, 2005; Mitsch et al., 1998).

In the Netherlands Lake Marken (see figure 1) is an example of an artificial freshwater lake that will be enhanced in biodiversity by creating a wetland (Natuurmonumenten, n.d.). This 700 square km artificial freshwater lake has a marine origin, but has been enclosed from the sea since the construction of the Enclosure Dyke in 1932, which created Lake Ijssel (Padro et al., 2012; Natuurmonumenten, n.d.). Lake Marken has also been separated from the Lake Ijssel by another dyke since 1976 (Natuurmonumenten; Lammens et al., 2008).

Figure 1 Map of Lake Marken, The Netherlands (Bouwhuis et al., 2008, p240).

Lake Marken is characterized by the high content of marine silt particles (Lammens et al., 2008; Bakker 2012) and the dykes inhibit flushing out the sediment. This has caused accumulation of salt and clay and formed a thick layer of anoxic sediments (Ibidem.). Due to the shallowness of the lake, wind and waves resuspend particles that causes high turbidity which inhibits plant and algae growth (Natuurmonumenten, n.d.). This has caused the ecosystem to degrade drastically and fish and bird populations have declined for more than 75% (Ibidem.).

The Markerwadden project aims to enhance the Lake Marken aquatic ecosystem and its ecosystem services and ultimately create a ‘birds’ paradise’ (Natuurmonumenten, n.d.). The project tackles current problems of the lake, such as the steep land-water gradient caused by its artificial shores, the thick accumulated silt layer and the oligotrophic state of the lake (Lammens et al., 2008). By ecosystem engineering artificial islands in the lake are created from local sediment in order to restore plant growth and to change the trophic state of the lake from oligotrophic to eutrophic (Natuurmonumenten, n.d.). To reach the latter, improvement of nutrient bioavailability is needed (Lammens et al., 2008).

Nutrients at the Markerwadden Lake are present in the marine sediment clay, but are now covered by silt and are in an anoxic state (Natuurmonumenten, n.d.; Lammens et al., 2008; Pardo et al., 2012). In the anoxic sediment, solute transport takes place by molecular diffusion (Anschutz et al., 2011; Shull, 2001) which significantly limits nutrient’s bioavailability.

Bioturbation is the dispersal of sediment particles due to small scale burrowing and feeding activities of macroinvertebrates and can have consequences on large scales, including the process of landscape formation, by increasing solute transport at the sediment-water interface (SWI) (Meysman et al., 2006; Majdi et al., 2014; Mermillod-Blondin & Rosenberg, 2006; Anschutz et al., 2011; Pardo et al., 2012; Matisoff et al., 1985; Mermillod-Blondin et al., 2008; Mulsow et al., 1998; Pelegrí et al., 1995; Michaud et al., 2005; Mitsch et al., 1998; Nogaro et al., 2009). Bioturbation leads to biogeochemical changes in the sediment, among others, changes in sediment-bound nutrients and increased oxidation of the soil. In addition, bioturbation can alter texture, porosity and changes in the sediment surface can lead to changes in erodibility and resuspension sensitivity (Boudreau & Jorgensen, 2001, retrieved from Meysman et al., 2006).

(6)

grained system with higher flow velocities (Mermillod-Blondin & Rosenberg, 2006). In diffusion-dominated systems, such as Lake Marken, bioturbators acts as a ‘direct vector of water and sediment fluxes’ (Mermillod-Blondin & Rosenberg, 2006, p435) and significantly influences oxygen consumption in the sediment in similar systems in contrast with advection-dominated systems, where bioturbation plays a minor role.

However, the influence of bioturbation on the sediment depends not only on the present sediment type (particle selective feeding), present macroinvertebrate biomass and bio volume, but also on the presence of the respectively five bioturbation functional groups (Anschutz et al., 2011; Smith, 1992, retrieved from Boudreau, 1994, retrieved from Mulsow et al., 1998) but also on the bioturbation category of the present macrofauna (Majdi et al, 2014; Anschutz et al., 2011; Shull, 2001). Five categories have been identified (Kristensen et al., 2012, retrieved from Majdi et al., 2014). The categories of bioturbators differently altered redox potential (Eh) (Hunting et al., 2012). For example, conveyor-belt deposit feeders, which are among others, the focus of this research, lower overall Eh but increase subsurface Eh (Ibidem.). Michaud et al. (2005) conducted an experimental microcosm research on marine water systems and measured the effect of bioturbation on sediment oxygen uptake. The results confirmed that the functional group approach was useful to quantify influence of bioturbation on water-sediment oxygen fluxes.

The macroinvertebrates that are present in Lake Marken are, among others, Tubifex tubifex and Chironomid riparus (van Riel, 2016). T. tubifex belongs to the conveyor-belt deposit feeders (Fischer et al, 1980). This functional group selectively, according to particle size, ingests sediment at depth (Matisoff et al., 1999) and egest at the SWI (Fischer et al, 1980). They feed head down up to depths of 6 to 9 or 10 cm (Fischer et al, 1980; Matisoff et al, 1999), which causes downward sediment transport (Pardo et al., 2012). Downward burrowing velocities are measured at 0.33 to 0.49cm/d/100.000 individuals/m2 (Matisoff et al., 1999). Respiration is ensured by keeping the back of the body in the water column (Lagauzère et al., 2009). The pores of T. tubifex are not irrigated (Ibidem.). The C. riparus belongs to the surface-deposit feeders and have a low burrowing activity (Lagauzère et al., 2009). Faecal pellets are deposited at the SWI (Palmer 1968, retrieved from Lagauzère et al., 2009) and a dominant upward sediment particle transport is realized. The burrows through which this happens are actively irrigated (Ibidem.).

Being classified as conveyor-belt deposit feeders (T. tubifex) it is expected that these macroinvertebrates burrow the pores and maintain them. The same is expected for the surface-deposit feeders, who actively irrigate their pores (Lagauzère et al., 2009). However, Fischer et al. (1980) observed that vacated pores became filled with sediment. While experiments showed burrowing activities up to 10 cm of depth, it is argued that the depth to which bioturbation occurs depends on the energetic burrowing cost and DO concentration, where a lower DO facilitates deeper burrowing (Boudreau, 1994; Fischer and Beeton, 1975). Both species enhance solute fluxes at the SWI and enhance sediment oxygen uptake (Lagauzère et al., 2009). The effect on solute fluxes is correlated with the density of the organisms (Ibidem.).

At SWI and below SWI (in sediment) oxygen transport only takes place by molecular diffusion (Anschutz et al, 2011; Shull, 2001). Therefore, oxygen concentrations are expected to decline exponential as soon as the SWI is reached (Anschutz et al., 2011, see figure 1a). However, in sediment where bioturbators are present, pores will be irrigated (Mermillod-Blondin et al., 2008; Krantzberg, 1985) and solute fluxes are transported by diffusion and irrigation, which can be modelled as advection (Anschutz et al., 2011; Mermillod-Blondin et al., 2008; Krantzberg, 1985). Bioturbators enhance aeration of the soil, to what extent macroinvertebrates can do so, depends on their functional group (as defined earlier) (Michaud et al., 2005) and characteristics of the sediment. Previous research has shown that up to 16 minutes after a burrow was made enhanced dissolved oxygen concentrations in the sediment were measured, however this effect is not present 30 minutes after the burrows were made (Lagauzère et al., 2009). Figure 1 (right hand side graph) shows the influence of tubificid worms on dissolved oxygen concentrations in sediment and shows that the tubificid worms mingle the sediment in such a way that at the top depth (up to 7 mm) oxygen concentrations are overall increased.

(7)

can cause leaps in pore water oxygen concentrations: oxygen concentrations can even increase at certain depth under influence of the presence of a pore (figure 2).

Figure 2Pore water dissolved oxygen profiles from shallow water with (right graph) and without burrowing macrofauna (left graph). Source: whoi.edu, slide 16.

Investigating the bioturbation process, followed by quantifying bioturbation and the associated biogenic structure is of importance, since the enhancement of solute fluxes correlates with the length and diameter of biogenic structures (Michaud et al., 2005). To be able to quantify the altered solute transport affects the nutrient cycling in the freshwater environment is essential to understand the effects of the presence of macrofauna on the ecosystem (Matisoff et al., 1994, Mermillod-Blondin et al., 2003 retrieved from Michaud et al., 2005). Therefore, this research will focus on examining the bioturbation process, with a focus on the influence of macroinvertebrates on sediment oxidation. The following research question is formulated: To what extent do oligotrophic macroinvertebrates contribute to nutrient release from the sediment through oxidation at Lake Marken, The Netherlands?

Conducting microcosm laboratory experiments during which macroinvertebrates, originating from Lake Marken sediment are observed, addresses the research question.The research focusses on investigating pore development, excretion layer development, tracer layers’ movement and influence of bioturbation on sediment oxygen concentrations. From previous research, it is expected that most pores develop in the top layer of the sediment (3 cm) and that the pore network will expand into more complex and dense network (van Riel, 2016). In addition, a correlation between macroinvertebrate density and pore area coverage is expected. Furthermore, it is expected to find the excretion layer to gradually increase over time. Finally, it is expected to find a gradually increase in pore water oxygen concentrations up to 7 mm compared to non-bioturbated sediments. It is not expected to find leaps in oxygen concentration in the profiles, since Anschutz’ research (2011) found that tubificid worms do not have such an effect.

Method

In order to answer the research question, multiple laboratory experiments were conducted. First a short description of the experiments is given, followed by general information, in order to conclude with a detailed description of each experiment and the statistical analysis.

The first experiment is a respectively 4- and 18-day experiment in which microcosms are kept in a climate room. The microcosms represent two groups: a control group (n=5) and a macroinvertebrate treated group (macroinvertebrate group hereafter, n=5). Regularly, pictures of the microcosms are being made to visualize pore and excretion layer development. Originally, the pictures were also meant to visualize vertical particle movement, using tracers, but this gave no result. Another experiment was conducted to measure a vertical profile of sediment dissolved oxygen in a bioturbated beaker. In addition to this experiment, another experiment was conducted to obtain the vertical macroinvertebrate distribution profile.

All water and sediment used in the experiment originates from Lake Marken. The used sediment is retrieved from an earlier conducted experiment, the sediment originates from Lake Marken, near Dyke Houtrib. A part of this sediment was treated with tap water, which caused the macroinvertebrates population to decline significantly: no activity was visible and no macroinvertebrates were counted in subsamples. This sediment, in combination with dried out sediment from the same experiment, was used as ‘macroinvertebrate poor’ sediment to fill all the microcosms and the beaker up to the height at which the incubation layer was deposited. The control group was incubated with a layer of macroinvertebrate poor sediment, while the other group and the beaker were incubated with ‘macroinvertebrate rich’ sediment, retrieved from a different site at Lake Marken: Harbour Putten. No additional treatment was conducted to the sediment. The water was retrieved from Lake Marken, near Harbour Putten (Almere). During

(8)

invertebrates are visualised and quantified. After the experiments with the microcosms was conducted, the number of macroinvertebrates present in each microcosm was counted.

Microcosm experiment

Instead of using perspex transparent colourless plates to make the microcosms, glass vases (n=10, dimensions: 8x11x20 cm) were used as microcosms, because the perspex microcosms were not waterproof. All microcosms were filled with the macroinvertebrates poor sediment up to 10cm, intermittent by a 1 cm layer of tracer (sibelco sand, very fine) at 8 cm depth. On top of the macroinvertebrate poor sediment a 1 cm layer of ornamental sand was added (yellow). After 12 hours, the microcosms were incubated with an additional 2 cm layer of macroinvertebrate poor sediment and macroinvertebrate rich sediment for the control group and the macroinvertebrate group respectively (n=5 for both groups). Thereafter, all microcosms got a 1 cm layer of sibelco sand and blue ornamental sand, see figure 3. A 3 cm water layer was added and the microcosms were not aerated by an aeration stone, because pilot experiments showed that this disturbs the excretion layer that formed on top of the final tracer layer. It was assumed that molecular diffusion provided sufficient oxygen to the microcosms.

Figure 3 Experimental set-up microcosm experiment. The microcosms contain water and sediment origination from Lake Marken. The sediment under the yellow tracer is ‘macroinvertebrate poor’ sediment and the top layer sediment is respectively ‘macroinvertebrate poor’ or ‘rich’ for the two treatment groups

From both experiments, the photographs were analysed using GIMP, where the pores and, respectively, the excretion layer were manually identified and selected, because it could not be done by software, see figure 4. The area of the pores and the excretion layer were analysed using ImageJ (output in pixels).

Figure 4 left: photo of microcosm middle: pores selected (GIMP) right: excretion layer selected (GIMP). This procedure was conducted for every microcosm at every measurement time of the experiment (t=4 d & t=18 d).

(9)

Tracers and particle transport

Originally the tracer layers were constructed to visualize particle transport caused by bioturbation. The different colours would enable identification of the origin of the particle which would enable creating a mathematical model of the particle transport by macroinvertebrates. During the pilot experiment it became clear that the tracers were not actively distributed throughout the microcosms, however the tracer layers facilitate identification of the excretion layer.

Density determination

After conducting the microcosm experiment all sediment present in the microcosms was sieved (500 mu and 250 mu meter) and the macroinvertebrates were counted using a large plastic transparent container on a table with integrated light. All macroinvertebrates were put into containers with 70% ethanol in order to be identified to species level later, but this is outside the scope of this research. The results from this experiment were analysed to test whether the control and the macroinvertebrate group are significantly. The test revealed that no significant difference between the ‘two’ groups. From now on the data from the microcosm experiment considers the microcosms one group (see Results section).

Pore experiment

The pore experiment ran for four days (measure points at t=0, t= 3, t=4, t=96 (h)). Pictures were made from the front of the microcosms, covering the 14 cm of sediment depth of the microcosm. The measurement points were determined following a pilot experiment in which pore development was so rapid that it was estimated that the experiment only needed to run for one day to get results.

The photographs were analysed on pore area occupation (pixels and percentage of total area), instead of calculating the area of the sediment that is in direct contact with the pores (using diameter), because it was unknown whether the pore ‘diameter’ on the photograph was the real pore diameter. The pore areas were categorized per depth (category 0 = 4-14 cm depth, 1= 3-4cm depth, 2=2-3cm depth, 3= 1-2 cm depth, 4=0-1 cm depth). The range of the categories was chosen after conducting the experiment and was determined after observations of bioturbation intensity (density of pores). Originally a comparison between the control and the macroinvertebrate group would have been conducted, however since the microcosms can not be treated as two groups, a PCA was conducted.

Excretion layer experiment

The excretion layer experiment ran for 18 days (measure points at t=0, t=96, t=151, t=175, t=240, t=272, t=314, t=338, t=410 (h)). Pictures were made from both the front and the back of the microcosm. Differences were tested between the front excretion layer and the back excretion layer (t-test). In addition, originally a comparison between the control and the macroinvertebrate group would have been conducted, however since the microcosms can not be treated as two groups, a PCA was conducted which only includes data from the front excretion layer, due to missing data for the backside in the PCA.

Dissolved oxygen profile experiment

For this experiment a 1000mL glass beaker was filled with macroinvertebrate poor sediment up to 650mL and incubated with 100mL macroinvertebrate rich sediment. After an incubation time of 15 days in a climate chamber (t=10 C), 10 random measurements were conducted with a micro sensor multi meter. The micro sensor was calibrated using two beakers of water that had been bubbling with water and nitrogen respectively for 30 minutes. The beaker bubbled with nitrogen was set as an oxygen concentration of zero (0mu mol/l), the beaker with water was set as 100% oxygen saturation (330 mu mol/l). The micro sensor was connected to software on a pc to read the output (figure 5).

(10)

Vertical profile distribution macroinvertebrates

For this experiment four cores (d=6, l= 30cm) and an AEE core cutter were used to analyse the vertical distribution of the Lake Marken macroinvertebrates. The cores were filled with macroinvertebrate poor sediment up to 14.5 cm and were incubated with a 3 cm layer of macroinvertebrate rich sediment. On top a 10 cm water layer was added and the water was aerated with aeration tubes. After 15 days in a climate chamber (t=10 o_{C) the cores were sliced with the AEE core slicer (see figure 6). The slices were taken at 0.5cm interval}

at the upper 3 cm of the cores, followed by 1cm intervals up to 8 cm of depth and 2cm intervals up to 12 cm depth. The slice samples were sieved using a 500mu meter sieve and stored in ethanol (70%) to be counted later.

In contrast to the proposed approach, using the AEE core cutter, it was not necessary to freeze the cores to enable slicing.

Figure 6 The AEE core cutter. The AEE core cutter enables to make slices from 0.5cm thickness up to several cm (0.5 cm interval). The core is pushed down until it stands on the plastic transparent slices (bottom). At the top, the slice is captured using a metal blade and it was captured in a tray (top of photo).

Statistics

All statistics were conducted using R and RStudio. Some graphs were made using Microsoft Excel (2016). In order to be able to analyse the data from the microcosms, the macroinvertebrate density had to be analysed

The assumption of normality was violated and therefore the two groups were analysed using the Wilcoxon-Mann-Whitney test. Since no significant differences were found, the microcosms were considered and analysed as one group. A PCA was conducted to estimate the ordination of the data.

The data from the pore development experiment was analysed on significant differences between pore area in depth categories and over time. The assumptions of ANOVA were tested using Shapiro-Wilk (normality) and Levene’s test (homogeneity of variances). When the assumption of normality was violated a Kruskal-Wallis test was performed instead of a one-way ANOVA. When the assumption of homogeneity of variances was violated, a ANOVA with Welch-correction was conducted. When both assumptions were violated, a log transformation was conducted, followed by the appropriate test. When the ANOVA, Kruskal-Wallis and/or the ANOVA-Welch-correction found significant differences between groups, the following Post-hoc tests were performed: TukeyHSD, Dunn-test and Games-Howell respectively. In addition, regressions were made to see what predictors affect the pore area. In order to make the regression, overall average pore area from the different depths were calculated (see appendix IV)

(11)

Results

Density experiment

The data of the control group had no normal distribution (Shapiro-Wilk, w=0.70082 p-value=0.009768).

An insignificant density difference was found between the control group (mean = 128.8) and the macroinvertebrate group (mean = 118.2), see figure 7 (Wilcoxon-Mann-Whitney test, W=16, p-value = 0.5476), see Appendix I and II. The found density mean represents 14000 individuals/m2_.

Figure 7 Boxplot of macroinvertebrate count per treatment. The hinges are the 25th_{and 75}th_{percentile, the whiskers represent data up to Q1-1.5*} IQR and respectively Q3+1.5*IQR. No significant difference between the groups are found, both groups have one or more outliers.

In all the following experiments all the data will be considered as one group.

Pore experiment

From the data retrieved using GIMP and ImageJ, the following boxplot of mean pore area per depth over time was constructed. See Appendix III and IV for data on the pore area in pixels and in percentages.

(12)

Figure 8 Pore percentage per depth category over time. At all depth categories the pore area increased at t=96 compared to the other

measurements (t=0,3, and 4 (h)). It appears that there is no large difference in pore area between t=0, t=3 and t=4 at all depth categories. Most pore area is found at t=96 at depths of 1-2 and 2-3 cm. The boxplot hinges are the 25th_{and 75}th_{percentiles. The whiskers extend from the hinge to} the highest (respectively lowest) value within 1.5*IQR of the hinges. Outliers are plotted as dots.

From the obtained data, the following scatter plots (figure 10 up to figure 15) were made of pore area (percentage and pixels) per depth category over time. The plots reveal an overall increase in pore area over time, although the increase in not continuous (t=0 and t=3) (see statistical test below figure 15). The non-continuous increase in pore area points towards theory of vacated pores which are subsequently filled with sediment, due to compaction of the soil, collapsing pores or sediment particles that are dragged along

(13)

Figure 9 This graph visualizes the pore area (mean) development per depth category over time. At all depth the pore area increases, however, the most increase in pore area is shown by depth categories 1-2 and 2-3 cm.

Figure 10 Pore area development over time, depth category is 0-1cm depth. All microcosms show parallel increase, except for microcosm h which shows the largest increase in pore area at this depth category.

0 2 4 6 8 10 12 -20Po 0 20 40 60 80 100 120 re a re a o ccup at ion (% ) t (h)

Pore area (%) per depth category over time

depth 0-1 depth 1-2 depth2-3 depth 3-4 depth 4-14 700 50700 100700 150700 200700 250700 0 20 40 60 80 100 120 ar ea (p x) t (h)

Pore development (px) 0-1 cm depth

a10 b10 c10 d10 f10 g10 h10 i10 j10 Linear (b10) Linear (c10) Linear (d10) Linear (f10) Linear (g10) Linear (h10)

(14)

Figure 11 Pore area development over time, depth = 1-2 cm. This graph shows more variance in pore area increase than depth category 0-1cm. The increase is bigger and the different microcosms show no real clustering.

700 50700 100700 150700 200700 250700 0 20 40 60 80 100 120 Are a(p x) t (h)l

Pore development (px) 1-2 cm depth

a11 b11 c11 d11 e11 f11 g11 h11 i11 j11 Linear (a11) Linear (b11) Linear (c11) Linear (d11) Linear (d11) Linear (e11) Linear (f11) Linear (g11) Linear (h11)

(15)

Figure 12 Pore area development over time, depth = 2-3cm. An overall increase in pore area is observed. All microcosms show similar increase; however small differences are observed in pore area increase at t=96.

Figure 13 Pore area development over time, depth = 3-4cm. Most microcosms show a small increase in pore area over time. Larger increases are observed in microcosm e, f, and j.

700 50700 100700 150700 200700 250700 300700 0 20 40 60 80 100 120 Are a (p x) t (h)

Pore development (px) 2-3 cm depth

a12 b12 c12 d12 e12 f12 g12 h12 i12 j12 Linear (a12) Linear (b12) Linear (c12) 700 50700 100700 150700 200700 250700 0 20 40 60 80 100 120 Are a (p x) t (h)

Pore development (px) 3-4 cm depth

a13 b13 c13 d13 e13 f13 g13 h13 i13 j13 Linear (a13) Linear (b13) Linear (c13) Linear (d13)

(16)

Figure 14 Pore area development over time, depth = 4-14cm. All microcosms show an increase in pore area at t=96 compared to t=0. All microcosms show parallel increase of pore area, however microcosms h, b, and d appear to have bigger increases than the other microcosms.

Multiple one-way Welch-ANOVAs reveal significant changes in pore area over time for all depth categories over time (see table 1). The R-script can be found in Appendix V-a.

Table 1 Results of one-way ANOVA conducted for each depth category. Significant results are indicated with *.

Depth category ANOVA

0-1 cm F(3,36)= 3.186 p-value = 0.03525 *

1-2 cm F(3,36)= 15.983 p-value = 9.052e-07 *

2-3 cm F(3,36)= 4.7126 p-value = 0.007099 *

3-4 cm F(3,36)= 3.7559 p-value = 0.02036 *

4-14cm F(3,36)= 3.5193 p-value = 0.02036 *

Post-hoc tests (Games-Howell) reveal significant differences between measurements for all depth categories, except category 0-1cm, see table 2. All detailed results can be found in Appendix V-c. It can be concluded that measurements of pore area at t=96 (h) differ significantly from the measurements at t=0, t=3 and t=4 for each depth category, except for depth category 0-1cm.

Measurements at t=0, t=3 and t=4 do not significantly differ.

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 20 40 60 80 100 120 Are a (% ) t (h)

Pore area development (%) depth 4-14 cm

a4-14 b4-14 c4-14 d4-14 e4-14 f4-14 g4-14 h4-14 i4-14 j4-14 Linear (a4-14) Linear (b4-14) Linear (c4-14) Linear (d4-14) Linear (e4-14) Linear (f4-14)

(17)

Table 2 summary of Games-Howell test of significant differences between groups (t) at multiple depth categories.

Depth category Groups p-value

4-14cm 0~96 3~96 4~96 p-value = 0.004 p-value = <0.001 p-value =<0.001 3-4cm 0~96 3~96 4~96 p-value = 0.009 p-value = 0.004 p-value =0.005 2-3 0~96 3~96 4~96 p-value =<0.001 p-value =<0.001 p-value =<0.001 1-2 0~96 3~96 4~96 p-value =<0.001 p-value =<0.001 p-value =<0.001

In addition, significant differences between depth categories per measurement (t (h)) was examined (see appendix V-c). A one-way ANOVA and respectively a Kruskal-Wallis test reveal significant differences of pore area between groups at t=3 and t=96, see table 3.

Table 3 Results Kruskal-Wallis tests and ANOVA respectively. For t=3, t=4 and t=96 significant differences between groups (depth categories 0-1, 1-2, 2-3, 3-4 and 4-14 cm depth were found

T Kruskal-Wallis ANOVA

0 F(4,44)= 1.3077, p-value = 0.2586

3 F(4,44)= 5.7919, p-value = 0.02008 *

4 log F(4,44)= 2.3098, p-value= 0.07273

96 Kruskal-Wallis chi-squared = 31.175, df = 4, p-value = 2.82e-06 *

Post-hoc tests, Dunn’s test for Kruskal-Wallis and TukeyHSD for ANOVA, were conducted and revealed where the significant differences between pore area in depth categories were found per measurement (t (h)). The depth categories are represented by 0 up to 4, where 0 is 4-14cm depth, and 4 is 0-1cm depth. At t=3, the following depth categories differed: 4-0, 2-1, 4-2, 4-3 (TukeyHSD, see appendix V-d). At t=96 the following depth categories differed: 1-3, 1-4, 2-3, 2-4, 3-0, 4-0 (Dunn-test, see appendix V-d for more details).

The following two graphs show the pore area percentage of the microcosms per depth category plotted against the number of macroinvertebrates found in the microcosms. The figure (16) shows the earlier observed trend of more bioturbation at depths of 1-2 and 2-3cm. No clear correlation is visible between the number of macroinvertebrates and the observed pore area (figure 17).

(18)

Figure 15 Pore area percentage per depth category plotted against macroinvertebrate count. No clear pattern is visible between pore area percentage at a certain depth and the number of macroinvertebrates in the microcosm.

Figure 16 This graph shows the pore area percentage at t=96 for depth 1-2 cm. No clear pattern is visible between the number of macroinvertebrates and the pore area percentage.

In addition to the plots, an overall PCA was made to check for principal components. In this section, the focus will be on pore development. First, a PCA for depth category 4-14cm is presented (figure 18). No clustering in microcosms (a up to j) is present, but ‘d’ and ‘b’ seem to be outliers. The principal component (PC) is t=96 (h), represented by array17.05.08. More details can be found in table 5. 0 2 4 6 8 10 12 14 16 85 95 105 115 125 135 145 155 165 175 185 Po re a re a (% ) Number of macroinvertebrates

Pore area (%) per microcosm macroinvertebrate count

t=96

depth 0-1 cm depth 1-2 cm depth 2-3 cm depth 3-4 cm depth 4-14 cm 0 2 4 6 8 10 12 14 16 85 95 105 115 125 135 145 155 165 175 185 Po re a re a (% ) Number of macroinvertebrates

Pore area percentage per microcosm (t=96, depth = 1-2 cm).

b c d e f g h i j a

(19)

Figure 17 PCA of pore development at depth 4-14 cm. A to J inclusive represent the individual microcosms, 13.05.9 represent the date.time(h) of the measured value. A up to E are the macroinvertebrate group, F up to J are the control group. The PCA shows no clustering of those two groups. D and H are outliers. The most important PC is the measurement at May 17th_{, at 8 O’clock. The accompanying eigenvalues are presented in the table} below.

Table 4 Summary of PCA (figure 18) details

Total variation 354509385464.700

Axis 1 Axis 2 Axis 3 Axis 4

eigenvalues 0.6422 0.6422 0.1145 0.0410

Explained variation (cumulative)

64.22 84.45 95.90 100.00

Additionally, a PCA was made, combining all collected data (figure 19). The PCs are the individual microcosms, see table 6. Depth categories are indicated with pbii-jj; where ii is the depth category (10 = 3-4 cm, 11=2-3cm, 12=1-2cm, 13=0-1cm) and jj is the measurement time (1-4, 0-96h), and porei for depth category 4-14cm, where i represents the measurement time. Clustering of pbii-jj is present, but a time factor is visible for depth category 4-14cm (porei ). The clustering of the pb1i-i factors point out that there is no time dependence of the pore area in the top 4 cm of the sediment. Therefore, it is concluded that the pores are temporal and collapse after being vacated, which in contrast to the hypothesised expansion theory.

(20)

Figure 18 PCA taking all factors into account. Focus here on pore development: pore1 up to pore4 represent pore development at depth 4-14cm on

time 0,3,4,96. pb10-1 represent pore development at depth 3-4 cm (depth = 0-1 ~ pb13, depth 3-4~ pb10 -i represents time (0,3,4,96). There is a clustering of the pb10-1 up to pb13-4. All the individual microcosms are also clustered and are the PCs. There is variance for the exi (excretion layer) and pore1 up to pore4.

Table 5 Summary of PCA (figure 19) details

Total variation 13425325457699.158

Axis 1 Axis 2 Axis 3 Axis 4

eigenvalues 0.9117 0.0397 0.0215 0.0124

Explained variation (cumulative)

91.17 95.14 97.29 98.53

Multiple regressions were made to visualise the effect of predictors, see Appendix V-a, V-b, macroinvertebrate density and time on the overall pore area (%). A regression with time plus density being the predictors revealed a significant effect of time F(2,37) =105.9, p-value <0.001 on pore area, R2_{= 0.8513. This confirms figure 16, where no effect of density on pore area was visible. A regression}

with time * density as predictors for pore area revealed no significant interaction, F(3,36) =73.19, p-value=0.165. This shows that density does not moderate for the effect of time on pore area.

Excretion layer experiment

First, the hypothesis about the burrowing behaviour of the macroinvertebrates can be tested by examining the excretion layer. Figure 20 shows the excretion layer that has formed on top of the sediment. The compaction of the faecal pellets further away from the SWI confirm the theory that faecal pellets are egested at the top of the sediment. At the end of the experiment (t=18d), excretion layers at the front cross section have formed with a thickness ranging from 1mm to almost 1 cm (photo analysis) within each microcosm.

(21)

Figure 19 Picture, zoomed in on excretion layer which is deposited on top of the top tracer layer. Scale bar on the left is a mm-scale. Pores and excretion pellets are visible. Compaction at the bottom of the excretion layer supports the hypothesis of egestion at the SWI.

From the data retrieved by using ImageJ after manual selection in GIMP (see Appendix VI) was used to make the following scatter plot (figure 21). Large differences seem to be present in excretion layer area between the groups ‘front’ and ‘back’ of the microcosm. Some excretion layers continue to increase in a steady rate, while other excretion layers appear to be stabilizing at a certain excretion area. The differences are analysed in table 8.

Figure 20 This figure shows the development of the excretion layer, both from the front and the back of the microcosms. The trend lines are moving average lines. The graph shows that in some microcosms the excretion layer does not increase continuously over time, it even decreases from t=240 for microcosm j. From t = 300, data gets more scattered.

0 500000 1000000 1500000 2000000 2500000 3000000 0 50 100 150 200 250 300 350 400 450 Are a (p x) t (h).

Excretion layer development over time

a b c d e f g h i j a2 b2 c2 d2 e2 f2 g2 h2 i2 j2

(22)

Table 6 Paired t-test between a up to j and a2 up to j2. For all measures, the difference between the two groups is significant (indicated with *). T t-test, p-value 96 t = -4.4768, df = 9, p-value = 0.00154 * 151 t = -4.8444, df = 9, p-value = 0.0009154 * 175 t = -3.7306, df = 9, p-value = 0.004693 * 240 t = -5.2349, df = 9, p-value = 0.0005384 * 272 t = -4.6945, df = 9, p-value = 0.001129 * 314 t = -4.9488, df = 9, p-value = 0.0007926 * 338 t = -5.5089, df = 9, p-value = 0.0003758 * 410 t = -4.4768, df = 9, p-value = 0.00154 *

To be able to retrieve more information from the scatter plot, figure 22 shows the development of the excretion layer at the front of the microcosms. The excretion layers do not develop uniform; the thickness of the layer varies between and within microcosms. This can be explained by non-uniform distribution of excretion ‘hotspots’. Furthermore, the cross sections of the excretion layer do not show a continuous increase. However, they do show an overall gradual increase in excretion layer area over time, which supports the hypothesis. The non-continuous increase might be explained by mingling of the sediment by the macroinvertebrates, which move and deposit the particles elsewhere at the SWI.

Figure 21 This figure shows the development of the excretion layer, data are from the pictures at the front of the microcosm. The trend lines are moving average lines. The graph shows that in some microcosms the excretion layer does not increase continuously over time, it even decreases from t=240 for microcosm j. Furthermore, excretion layer area increases fastest during the first 96 hours.

The one-way ANOVA to check for significant differences in excretion layer area over time gave significant results F(1,78) =9.881 p-value=0.002362), see Appendix VII-a.

The significant predictor time on excretion layer is also visualised by the trend in the scatter plot (figure 22) and the ordination of the PCA (figure 23). It is visible that there is no clustering of the multiple ex i measurements. Because of the scatter of ex i it can be concluded that time if of importance for the area of the excretion layer. This supports the hypothesis of a gradual increase of the

0 200000 400000 600000 800000 1000000 1200000 0 50 100 150 200 250 300 350 400 450 Are a (p x) t (h).

Excretion layer development over time at the front of the microcosm

a b c d e f g h i j

(23)

Figure 22 This graph shows the result of the PCA over all data, excretion layer at the back of the microcosm is not included. Focussing on the excretion layer part (indicated by ‘ex i’, where i is the time (h) at which the picture is taken, ex1 represents t=0, ex9 represents t=410 (h).

In addition to the overall PCA, figure 24 shows a PCA which only included the excretion layer area (both groups ‘front’ and ‘back’). In accordance with the previous PCA, this PCA reveals that time is an important explanatory variable for the observed variance in measurements in excretion layer area. Furthermore, the PCA shows that the excretion layer at the front (i) and of the back (i2) of the microcosm are clustered, which indicates that these can be treated as different groups.

(24)

Figure 23 PCA 1 up to 9 represent time 1 represents t=0, 9 represents t=410. The excretion layers over time are not clustered. A up to j represent excretion layers at the front of the microcosm, a2 up to j2 represent excretion layers at the back of the microcosm.

Multiple regressions were made to control for the effect of factors on excretion layer development, see Appendix VII-a and VII-b. A regression with time + density as predictors for excretion layer area revealed significant effects of both time and density on excretion layer, F(2,77) =24.01, p-value <0.001 for both predictors, R2_{= 0.3841. A regression with predictors time*density revealed no significant}

interaction, F(3,76) =15.95, p-value=0.5990. This indicates that the effect of time is not moderated by the effect of density.

Dissolved oxygen profile experiment

The output of the all the measurements is displayed below (figure 25), see also appendix VIII. The dissolved oxygen is depleted at a maximum depth of 14mm and large ‘leaps’ in the DO profile are visible. They indicate the presence of pores and within pores, the DO concentration remains constant at increasing depth. See figure 26 and 27 for more details on individual DO profiles.

(25)

At the SWI there is a significant change in oxygen concentration. However, under influence of pores, the oxygen concentration does not change with depth as expected. At depth of -1 to -3 mm and -4 to -9 mm a pore causes a constant oxygen concentration over different depths. In this measurement series, oxygen is depleted at depth -10mm.

At depth of 2 to 3 mm oxygen concentrations decrease gradually with depth, while an increase in DO is observed up between the depth of 4 to 7mm. Oxygen concentrations remain stable between 8-9mm and 10-12mm depth. DO is depleted at a depth of 13mm.

Figure 25 Oxygen concentration measurement series 4. Depth at SWI is zero, above (in water column) depth is positive. Below the SWI, depth has a negative value.

(26)

A boxplot (figure 28) was made to visualize the differences in measured DO per depth category. Up to the depth of 13mm DO concentrations varied per depth category. DO was depleted in all measurements at depth of 14mm. The shallowest depth at which DO was depleted was 5mm (see also figure 26, series 6).

Figure 27 Oxygen concentrations variance per depth category. The boxplot hinges are the 25th_{and 75}th_{percentiles. The whiskers extend from the} hinge to the highest (respectively lowest) value within 1.5*IQR (inter quartile range) of the hinge. The remaining values are outliers and plotted as points (source ggplot2.org, n.d.).

DO concentration changes were found up to values of -656 and +780 mu mol/l/mm increase in depth, see appendix VIII, with a mean of 21.66434 mu mol/l decrease per increase in 1 mm depth.

Vertical macroinvertebrate profile

I had no time to count the macroinvertebrates collected in the samples.

Discussion

The marine origin of the silt does not seem to effect the bioturbation macroinvertebrates negatively; the bioturbation process is clearly present at Lake Marken sediment during the experiments.

Pore experiment

The observed pore structures and excretion layer during the experiments confirm the presence of macroinvertebrates that ingest sediment at depth and egest particles at the SWI. However, it was expected that conveyor-belt deposit feeders would build and expand their network, but this was not observed and is in contrast to the pore expansion theory. However, pore area increased, but non-continuously: it supports the theory of vacated pores which are subsequently filled with sediment, due to compaction of the soil, collapsing pores and sediment particles that are dragged along with the macroinvertebrates (Fischer et al, 1980). The macroinvertebrates appear to ‘dig as they eat’. From this result it was assumed that the continuous creation of pores aerates the sediment thoroughly and a gradual decrease of DO at depth was expected, independent of the direct temporal presence of pores.

Most bioturbation activity was found at depth between 1 and 3 cm, which might be explained by the presence of surface-deposit feeders, such as C. riparus. Although most pores were found at shallow depths, pores were present throughout the whole

(27)

Excretion layer

Large volumes of egestion particles were deposited at the SWI. The difference among the two groups, ‘front’ and ‘back’ of the microcosm, might be explained by (a combination of) factors that are not controlled for, such as light and other edge effects, caused by the microcosms. The microcosms were not covered to block out light, which would be more as the natural situation. Besides, the tracer layer might influence burrowing behaviour and a possible difference in tracer layer thickness might cause the observed difference.

From the regression it is concluded that both time and macroinvertebrate density are significant predictors. The effect of density on bioturbation needs further research to be able to quantify the bioturbation effect.

Tracer experiment

The failure of the visualization of particle transport (using traces layers) supports the theory that macroinvertebrates feed selectively on particle size. This can have implications for Lake Marken when the stratification and spatial distribution of the sediment is changed when the project Markerwadden is conducted. Silt present at Lake Marken is divided into four grain size categories; ranging from <2 to >18 mu m (Vlag, 1992). The grain size boundary or preference of feeding activities of macroinvertebrates is species dependent (Coates & Reynoldson, 1996) and can cause spatial heterogeneity in bioturbation intensity and its aeration effect at Lake Marken. When a strong grain size preference is observed among Lake Marken macroinvertebrates and spatial heterogeneity of grain size is large, it is expected that bioturbation intensity and effect on the sediment can cause spatial variance at Lake Marken.

Vertical oxygen profile experiment

Results from the DO profile experiment support the hypothesis that bioturbation at Lake Marken significantly enhances the aeration state of the sediment up to profound depths (14mm), even deeper than expected (9mm). In contrast to the hypothesis and observations of pore development over time, which did not confirm the pore expansion theory, it was expected to find a gradual DO decrease over depth instead of the irregular profiles found by the experiment. Other research on bioturbation has shown DO profiles with leaps, however the species used for the experiment was not specified and I assumed it involved larges macroinvertebrates. The macroinvertebrates in Lake Marken do cause leaps of oxygen concentrations in the sediment. DO concentration changes were found up to values of -656 and +780 mu mol/l/mm increase in depth, see appendix IX, with a mean of 21.66434 mu mol/l decrease per increase in 1 mm depth. The result supports the hypothesis that pores aerate the sediment, however they do this locally and/or temporally. Since DO levels were not measured over time, it is unclear to what extent the aeration effect is temporal and/or local. Molecular diffusion in combination with rapid DO consumption by chemical and/or biological processes (Matsui & Tsuchiya, 2006) might explain the observed DO profiles.

Between the leaps, oxygen concentration remains stable at increasing depth. This supports the hypothesis that within pores solute fluxes transport can be seen as advection (seen as advection) in addition to molecular diffusion (Anschutz et al., 2011; Mermillod-Blondin et al., 2008; Krantzberg, 1985).

The depth to which oxygen is present at measurable depth is significantly less deep than the depth at which the pores are present (14cm). This is caused by chemical and biological consumption of oxygen (Matsui & Tsuchiya, 2006) and the dependence on molecular diffusion of oxygen into the pores. The leaps in oxygen measurement also indicate that the range of molecular or radial diffusion of oxygen from the pores to the surrounding sediment is smaller than the distance between the pores ((multiple) mm). Figure 26 shows that 1mm depth difference caused a DO decrease of 90 mu mol*l^-1 (142 to 52 mumol). More research is needed to be able to make better conclusions on the aeration effect of bioturbating macroinvertebrates at Lake Marken.

Recommendations Microcosms

The control group was not significantly different from the macroinvertebrate rich group. The macroinvertebrates should have been sieved from the sediment, been counted and a known, fixed number of macroinvertebrates should have been added to the macroinvertebrate rich microcosm group. In addition, the macroinvertebrate ‘poor’ sediment should have been treated with gamma rays to ensure total absence of macroinvertebrates (van der Geest & van Riel, 2016).

An edge effect might have been present in the microcosms, indicated by a higher pore density at the corners of the microcosms (not tested, only observed). Multiple factors might explain this observation: it is possible that the abiotic environment was slightly different at the corners, more light and possibly more oxygen, might have influenced the macroinvertebrate’s behaviour. The microcosms were not protected against light (light dark cycle in the climate chamber 12/12h) and this might have influences the natural burrowing behaviour of the macroinvertebrates.

The microcosms were not aerated by aeration stones, because this disturbed the formation of the excretion layer during the pilot experiment. However, oxygen concentrations were not monitored and if an oxygen shortage has occurred, it is likely that this

(28)

From t=0 at the microcosm experiment, pores were present up to the maximum depth of 14 cm. This can be explained by the presence of macroinvertebrates in the macroinvertebrate ‘poor’ sediment. The increase in pore area at this depth category (4-14cm) might be explained by both the upward migration of buried macroinvertebrates and the downward migration of bioturbating macroinvertebrates. Possibly, only one of the two explanations might account for the observed increase in pore area at depth (4-14cm), however this can not be concluded from this experiment and the presence of the macroinvertebrates in the ‘poor’ sediment might have influenced the results of the depth at which macroinvertebrates burrow in Lake Marken sediment.

Tracer profiles

Although the tracers were not actively transported, it could have provided interesting information: observing change in depth of the layers reveals where the egested material (at the top of the SWI) originated from. However, I had not enough time to do so. It would be interesting for further research.

Temperature climate chamber

Fluxes at the SWI are influenced by water temperature (Osté & van de Waard, 2012). Standard experiments are conducted at 15o_C;

therefore, it is recommended to conduct the bioturbation microcosm experiment also at 15o_{C. Below this threshold temperature}

affects SWI fluxes only slightly, above this temperature SWI fluxes are significantly enhanced (appendix IX) (Ibidem). Therefore, the effect of bioturbation might differ from season to season and its importance can easily be underestimated when the water temperature exceeds the 15o_{C boundary in June up to September (klimaatinfo, n.d.).}

The climate chamber in which the core was stored for 14 days before conducting the measurements was set on 10 o_{C. This}

differs from the temperature of the climate chamber of the microcosms (12o_{C) and the difference in temperature might influence the}

oxygenation of the sediment.

Manual selection pores and excretion layer

Selecting the pores and the excretion layer manually is less objective than selecting it by using software. In the short time span of the research, I was not able to find or construct a suitable method to select the pores and excretion layer using software. Selection by colour was not successful, because sediment particles had the same colour as both the pores and the excretion layer. I was not experienced enough in programming to be able to write a pattern and colour recognition code.

Oxygen profiles

Although the ten replicates of the DO profiles gave interesting results, it can not be quantified, since no baseline measurement, without bioturbation, has been conducted. It is assumed that in case of no bioturbation DO will be depleted at a depth of 2mm. However, this has to be confirmed in future research.

The micro sensor was calibrated (see method section), however from the first measurement (see Appendix I) the measurement values exceeded the calibrated 100% saturation value (315 mu mol*l-1). This might be explained by size of the bubbles during bubbling: the bubbles were quite large, therefore reducing the effect of gas exchange between the water and the bubbles. Besides the calibration issue, the DO measurements were only at one time. To determine the local and temporal effect of the pores, it would be useful to use oxygen sheets.

Vertical macroinvertebrate profile

After the slicing, sieving and storing the macroinvertebrates in 70% ethanol, the macroinvertebrates became very hard to identify due to the death; this changed their colour from red to white. Besides, a lot of organic matter was present at the sediment and it was hard to tell the difference between OM and a macroinvertebrate. Because counting the macroinvertebrates would be very time consuming, I decided that I had no time to conduct this experiment. However, it is suggested that the macroinvertebrates can be counted easily by putting them into seawater which will make the macroinvertebrates float. This is a promising method for further research.

Further research

In addition to the previously mentioned improvements of this research, further research on bioturbation of macroinvertebrates at Lake Marken should focus on both direct and indirect effects of bioturbation on Lake Marken. It is expected that solute fluxes at the SWI are altered by bioturbation, and these changes need to be both quantified and qualified. Likewise, the density dependent effect

(29)

Conclusion

This work identified the bioturbation process of oligotrophic macroinvertebrates within the marine silt of the freshwater lake Lake Marken. The organisms built their burrows up to 14 cm of depth, with the highest burrow intensity found at depths between 1 and 3 cm. The pore expansion theory was not confirmed. An excretion layer of faecal pellets formed, as was expected due to the presence of conveyor-belt deposit feeders functional group. Finally, sediment dissolved oxygen measurements showed enhancement of sediment oxygenation up to 14mm of depth. The results show great potential to enhance solute fluxes at SWI and thus enhance the bioavailability of nutrients at the lake to turn it into a eutrophic lake. In addition, particle selective feeding was confirmed, likewise a correlation between macroinvertebrate density and bioturbation intensity was found. Further research should focus on investigating the solute fluxes at the SWI and density dependence of bioturbation.

Acknowledgements

I would like to thank my daily supervisor, Marielle van Riel, and my examiner, Harm van der Geest, for their input and assistance during the whole period of conducting the bachelor thesis. In addition, I would like to thank Dorine Dekkers (Alterra) for assistance during the practical part of the thesis.

(30)

References

Anschutz, P., Ciutat, A., Lecroart, P., Gérino, M., & Boudou, A. (2012). Effects of tubificid worm bioturbation on freshwater sediment biogeochemistry. Aquatic geochemistry, 18(6), 475-497. Bakker,M. 2012. Effect of Bioturbation on the Erodibility of fine sediments in Lake Markermeer. Mcs Thesis. Bouwhuis, H., Noordhuis, R., Portielje, R., & van der Molen, D. (2008). Towards ecological goals for

the heavily modified lakes in the IJsselmeer area, The Netherlands.Hydrobiologia, 599(1), 239-247

Coates, K. A., Reynoldson, T. B., & Reynoldson, T. B. (Eds.). (1996). Aquatic Oligochaete Biology VI. Hydrobiologia, 334, 97-101 Hunting, E. R., Whatley, M. H., van der Geest, H. G., Mulder, C., Kraak, M. H. S., Breure, A. M., &

Admiraal, W. (2012). Invertebrate footprints on detritus processing, bacterial community structure, and spatiotemporal redox profiles. Freshwater Science, 31(3), 724-732.

Krantzberg, G. (1985). The influence of bioturbation on physical, chemical and biological parameters in aquatic environments: a review. Environmental Pollution Series A, Ecological and Biological, 39(2), 99- 122

Klimaatinfo. (n.d.). http://www.klimaatinfo.nl/nederland/markermeer.htm, viewed on 23-06-2016.

Lammens, E., van Luijn, F., Wessels, Y., Bouwhuis, H., Noordhuis, R., Portielje, R., & van der Molen, D. (2008).

Towards ecological goals for the heavily modified lakes in the IJsselmeer area, The Netherlands.Hydrobiologia, 599(1), 239-247.

Matisoff, G., Fisher, J. B., & Matis, S. (1985). Effects of benthic macroinvertebrates on the exchange of solutes between sediments and freshwater. Hydrobiologia, 122(1), 19-33.

Majdi, N., Bardon, L., & Gilbert, F. (2014). Quantification of sediment reworking by the Asiatic clam Corbiculaa fluminea Müller, 1774. Hydrobiologia, 732, 85-92.

Mermillod-Blondin, F., & Rosenberg, R. (2006). Ecosystem engineering: the impact of bioturbation on biogeochemical processes in marine and freshwater benthic habitats. Aquatic Sciences, 68(4), 434-442.

Mermillod-Blondin, F., Lemoine, D., BOISSON, J. C., Malet, E., & Montuelle, B. (2008). Relative influences of submersed macrophytes and bioturbating fauna on biogeochemical processes and microbial activities in freshwater sediments. Freshwater

biology, 53(10), 1969-1982.

Meysman, F. J., Middelburg, J. J., & Heip, C. H. (2006). Bioturbation: a fresh look at Darwin's last idea. Trends in Ecology &

Evolution, 21(12), 688-695

Michaud, E., Desrosiers, G., Mermillod-Blondin, F., Sundby, B., & Stora, G. (2005). The functional group approach to

bioturbation: the effects of biodiffusers and gallery-diffusers of the Macoma balthica community on sediment oxygen uptake. Journal of Experimental Marine Biology and Ecology, 326(1), 77-88.

Mitsch, W. J., Wu, X., Nairn, R. W., Weihe, P. E., Wang, N., Deal, R., & Boucher, C. E. (1998). Creating and restoring wetlands. BioScience, 48(12), 1019-1030.

Mulsow, S., Boudreau, B. P., & Smith, J. A. (1998). Bioturbation and porosity gradients. Limnology and Oceanography, 43(1), 1-9. Natuurmonumenten.(n.d.) https://www.natuurmonumenten.nl/marker-wadden/english viewed on 05-2016.

Nogaro, G., Mermillod-Blondin, F., Valett, M. H., François-Carcaillet, F., Gaudet, J. P., Lafont, M., & Gibert, J. (2009). Ecosystem engineering at the sediment–water interface: bioturbation and consumer-substrate interaction.

Oecologia, 161(1), 125-138.

Osté, L.A. en H. van de Weerd, 2012. Kennisregels in de Bodemdiagnose BaggerNut. Deltares-rapport 1201327-0014.

de Lucas Pardo, M. A., Bakker, M., van Kessel, T., Cozzoli, F., & Winterwerp, J. C. (2013). Erodibility of soft freshwater sediments in Markermeer: the role of bioturbation by meiobenthic fauna. Ocean

Dynamics,63(9-10), 1137-1150.

Pelegrí, S.P., & Blackburn, H. (1995). Effects of Tubifex on N-mineralization in freshwater sediments N isotopes. Aquatic microbial ecology, 9:285-294.

Ramsar COP12 DOC.23

Shull, D.H. (2001). Transition-matrix model of bioturbation and radionuclide diagenesis. Limnol. Oecanogr., 46(6), 905-916.

Timm, T.; Erséus, C. (2015). Limnodrilus hoffmeisteri Claparède, 1862. Accessed through: World Register of Marine Species at http://www.marinespecies.org/aphia.php?p=taxdetails&id=137556 on 2016-04-04

van de Bund, W. J., Goedkoop, W., & Johnson, R. K. (1994). Effects of deposit-feeder activity on bacterial production and abundance in profundal lake sediment. Journal of the North American Benthological

(31)

Appendix

I Results count macroinvertebrates in microcosms ... 32

II R-script counts macroinvertebrates microcosms ... 32

III

CSV file pore area (pixels) ... 33

IV

CSV file pore area (percentages). ... 34

V-a R script Pores ... 34

V-b Regression plots ... 37

V-c

Games-Howell post hoc ... 39

V-d Overview statistical tests ... 40

VI

CSV file excretion layer area (px) ... 43

VII-a

R script Excretion layer experiment ... 43

VII-b

Regression plots ... 46

VIII DO profiles ... 47

IX

Graph SWI flux ~ temperature ... 54

(32)

I Results count macroinvertebrates in microcosms

Table 7 number of macroinvertebrates counted in each microcosm. Treatment 1 = macroinvertebrate rich, treatment 0= control.

counts treatment a 90 1 b 157 1 c 107 1 d 112 1 e 125 1 f 119 0 g 108 0 h 118 0 i 120 0 j 179 0

II R-script counts macroinvertebrates microcosms

setwd("C:/Users/Esther/Documents/BP macroinvertebrates") counts= read.csv("tellingen microcosms alterra2.csv") #mean tapply(counts$counts,counts$treatment, mean ) ## control 128.8 ## macro 118.2 #s tapply(counts$counts,counts$treatment, sd ) ## control 28.47 ## macro 25.05

## verschil is 3, dus mean_control is 2 sd verwijderd van mean_macroinvertebrates ## verklaring door 1 outlier...

#variance

tapply(counts$counts,counts$treatment, var ) ## control 810.7

## macro 627.7

#F test to compare two variances, wel weinig data dus wellicht zegt dit niks

var.test(counts$counts[counts$treatment == 'control'], counts$counts[counts$treatment == 'macro']) ## F = 1.2915, num df = 4, denom df = 4, p-value = 0.8102 p>a, niet rejected

## 95% confidence interval: 0.134472 12.404640 ## ratio of variances: 1.291541

(33)

# conduct wilxon test Wilcoxon-Mann-Whitney

wilcox.test(counts$counts ~ counts$treatment, alternative="two.sided")

##W = 16, p-value = 0.5476 alternative hypothesis: true location shift is not equal to 0 # H0 (equal means) not rejected

#boxplot

b_counts = boxplot(counts$counts ~ counts$treatment, main="Counted macroinvertebrates for each treatment", xlab="Treatment", ylab="Number of macroinvertebrates", col=(c("gold","darkgreen")))

III CSV file pore area (pixels)

Table 8 Pore area per depth category (0-1, 1-2, 2-3, 3-4, 4-14cm) per microcosm (a up to j) over time (h) in pixels.

a3-4 a2-3 a1-2 a0-1 a4-14 b3-4 b2-3 b1-2 b0-1 b4-14

0.01 26990.96 43636.52 34401.56 6998.44 301861 Jan-96 43667.8 53982.84 23243.8 338165 3 8848 35200 4665 2975 246399 150-18 47955 2030-1 2876 305451 4 17056 36381 20562 Mar-22 342483 22955 57999 28676 17261 276598 96 54864 197567 165486 2252-3 565253 55021 240546 234722 32092 688153 c3-4 c2-3 c1-2 c0-1 c4-14 d3-4 d2-3 d1-2 d0-1 d4-14 0.01 17768.88 24387.36 55465.88 5680.08 30-1074 33228.56 3900-1.52 42487.44 2-3252.52 502-353 3 9346 Feb-46 Jan-87 724 198197 26277 31865 22765 3548 223-452 4 8253 16842 37450 7638 229942 0-1531 32945 241-21 542-3 3-4552-3 96 38359 194329 166907 9059 338655 93346 237240 200578 42825 756200

e2-3 e1-2 e0-1 e4-14 f3-4 f2-3 f1-2 f0-1 f4-14

0.01 43543.6 26872.28 48692.84 290189 47663.36 25451.8 5391.2 4616.56 216251 3 44636 18750 Feb-61 167717 22973 26088 18753 Mar-61 144284 4 5933-4 37527 25894 22573-4 302-34 18262 16979 52465 182070 96 166595 234988 232374 495201 81-234 170-104 255522 3-48462 484845 g3-4 g2-3 g1-2 g0-1 g4-14 h3-4 h2-3 h1-2 h0-1 h4-14 0.01 223-41.16 223-41.16 24384.6 0 287463 16316.2 30441.88 35306.84 26282.56 220755 3 26549 27253 22862 6873 1530-19 Feb-07 16095 27360 7564 20652-3 4 18787 21551 23997 0-1244 219920 6289 9455 19500 8618 295346 96 47529 3-43951 3-47378 25241 428924 153515 1-20996 156270 39278 795361

i3-4 i2-3 i1-2 i0-1 i4-14 j3-4 j2-3 j1-2 j0-1 j4-14

0.01 1731-2.56 16316.2 16316.2 16316.2 368543 25233.76 29173.2 35177.1-2 26788.56 290215

3 Feb-17 21-235 34971 6461 288064 24243 192-33 25080 16738 309377

4 17428 3-43-45 33336 7527 272-341 3493-4 40-106 28835 16469 162694

(34)

IV CSV file pore area (percentages)

Table 9 Pore area per depth category (0-1, 1-2, 2-3, 3-4, 4-14cm) per microcosm (a up to j) over time (h) in percentage (pixels pores/ pixels microcosm section*100%).

Table 10 Pore area percentage of whole microcosm per microcosms over time (h). used for regression analysis.

V-a R script Pores

setwd("C:/Users/Esther/Documents/BP macroinvertebrates/pores r")

a3-4 a2-3 a1-2 a0-1 a4-14 b3-4 b2-3 b1-2 b0-1 b4-14

0.01 1.481717 2.395505 1.888535 0.384192 1.51-2189046950340.69697 2.397222 2.963485 1.27603 1.694056 3 0.485727 1.932367 0.2560935440-170220.163317962233-4061.234349 0.833 2.632576 1.2-352-3857707510.157883 1.530173 4 0.93632 1.9972 1.128788 0.588603 1.715687 1.260156 3.183959 1.57422 0.947574 1.385633 96 3.011858 10.84579 9.084651 1.235782 2.831666 3.020477 0-1.20520421607381-2.88548528765921.76174790-1921833.44734 c3-4 c2-3 c1-2 c0-1 c4-14 d3-4 d2-3 d1-2 d0-1 d4-14 0.01 0.975455 1.338788 3.044899 0.32-38181818181821.568362-378439521.824141 2.141717 2.332424 0.617727 2.53 3 0.50-1065436978480.62-38796662274920.67451690821-2560.03974527887570-170.9928786183-404411.442523 1.749286 1.249726 0.194774 1.3 4 0.453063240-1-467190.924572 2.055885 0.41930171-27799741.151907 0.742809 1.808575 1.324166 0.297046552480-1350.528563 96 2.1057861-220904710.66804 9.162659 0.49733 1.69653 5.1-243960-15265713.02372 11.01109 2.350955 3.78822

e2-3 e1-2 e0-1 e4-14 f3-4 f2-3 f1-2 f0-1 f4-14

0.01 2.390404 1.475202 2.673081 1.453718 2.616566 1.397222 0.29596 0.253434 1.08332 3 2.450373 1.02931488803-4540.629172 0.840187 1.262-344049187531.432148 1.02948 0.5578063240-1-46720.722799 4 3.255929 2.0602 1.421497584543-461.0-10706483405151.653162 1.002525 0.932093 2.88016 0.91 96 9.145531 12.90009 12.75659 2.480736 4.459486 9.40404 14.02734 5.954216073781-292.428857 g3-4 g2-3 g1-2 g0-1 g4-14 h3-4 h2-3 h1-2 h0-1 h4-14 0.01 1.201283 1.201283 1.338636 0 1.440061 0.895707 1.672362 1.938232 1.442828 1.3 3 1.457455 1.4963 1.255051 0.377306 0.7671581-239740430.626208 0.883564 1.501976 0.415239 1.034528 4 1.030146 1.183081 1.317358 0.727053 1.3-41702-364460860.345246 0.519049 1.070487 0.473341 1.479552 96 2.60919 5.706577 5.894708 1.385649978041-282.148718 8.427480-135090036.6422924902-38588.578722 2.156236 3.984404

i3-4 i2-3 i1-2 i0-1 i4-14 j3-4 j2-3 j1-2 j0-1 j4-14

0.01 0.950404 0.895707 0.895707 0.895707 1.846236 1.385253 1.601515 1.932323 1.470606 1.453847 3 0.648715 1.165733422-36821.919796 0.354688 1.443072 1.3308629776022-31.049242 1.376824 0.918863 1.549841 4 0.95674 0.5547323-436451471.83004 0.40-12081686429511.358295541-256281.916447 2.267567 1.582949 0.904095 0.815024 96 3.737264 2-3.52486824769438.05325 0.8288866930171-282.774051 4.674627 23.12599 23.56 5.565986 1.890005 a b c d e f g h i j 0.01 1.519417 1.733876 1.526185 2.301144 1.661102 1.099742 1.295844 1.353509 1.578563 1.494869 3 1.22712 1.438072 0.840571 1.259736 1.040069 0.867946 0.875549 0.983733 1.309166 1.425594 4 1.557699 1.487303 1.24857 0.746645 1.363612 1.111996 1.262476 1.228832 1.237548 1.058664 96 3.749624 3.325379 3.14781 5.218801 5.08645 4.391171 2.888177 4.379568 3.665824 5.416189

(35)

#depthcat 4 = i13, 3= i13,2=i11, 1=i10, 0=ia-14 data_b_o.m$depthcat = c(1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4, 1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4, 2,2,2,2,3,3,3,3,4,4,4,4,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4, 1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4, 1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0) #data_b_0.m to csv file

databo0.01 = subset(data_b_o.m, data_b_o.m$X == 0.01) databo3 = subset(data_b_o.m, data_b_o.m$X == 3) databo4 = subset(data_b_o.m, data_b_o.m$X == 4) databo96 = subset(data_b_o.m, data_b_o.m$X == 96)

databocat1 = subset(data_b_o.m, data_b_o.m$depthcat == 1) databocat2= subset(data_b_o.m, data_b_o.m$depthcat == 2) databocat3= subset(data_b_o.m, data_b_o.m$depthcat == 3) databocat4= subset(data_b_o.m, data_b_o.m$depthcat == 4) databocat0= subset(data_b_o.m, data_b_o.m$depthcat == 0) #normality

by(databocat0$value, databocat0$X, shapiro.test) by(databocat1$value, databocat1$X, shapiro.test) by(databocat2$value, databocat2$X, shapiro.test) by(databocat3$value, databocat3$X, shapiro.test) by(databocat4$value, databocat4$X, shapiro.test)

##ANOVA differences between means of certain depth categories over time #variance databocat0$X= as.numeric(databocat0$X) databocat1$X= as.numeric(databocat1$X) databocat2$X= as.numeric(databocat2$X) databocat3$X= as.numeric(databocat3$X) databocat4$X= as.numeric(databocat4$X) with(databocat0, leveneTest(y=value, group =X)) with(databocat1, leveneTest(y=value, group =X)) with(databocat2, leveneTest(y=value, group =X)) with(databocat3, leveneTest(y=value, group =X)) with(databocat4, leveneTest(y=value, group =X))

#ANOVA with Welch-correction: oneway.test(Count ~ Group, data=df, na.action=na.omit, var.equal=FALSE) attach(databocat0)

oneway.test(value~X, var.equal = FALSE) attach(databocat1)

oneway.test(value~X, var.equal = FALSE) #0 F(3,13)= 19.471 p-value = 5.292e-05 * #1 F(3,17)= 7.8725 p-value = 0.001647 * #2 F(3,35)= 18.937 p-value = 5.088e-08 * #3 F(3,38)= 18.714 p-value = 3.059e-08 *

The extent of bioturbation activity by oligotrophic macroinvertebrates in the Lake Marken, The Netherlands