Introduction

The growth of diatom biofilms has been shown to be important in mediating the properties of and processes in muddy sediments (Paterson, 1997; Black et al., 2002). The net effect of biofilm growth is usually considered to be stabilising, however, diatoms can destabilise sediments (de Jonge & van der Bergs, 1987; Sutherland, 1998) and natural biofilms exhibit a wide range of erosion thresholds (Defew et al., 2002). The stabilising effect of biota is usually attributed to the secretion of extracellular polymeric substances (EPS) (Dade et al., 1990; Decho, 1990; Tolhurst et al., 2002), however, recent work indicates that the relationship may be more complex, with the organisms themselves being important in structuring or stabilising the EPS and sediment (de Brouwer et al., 2002; Tolhurst et al., 2003; de Brouwer et al., 2005).

Unfortunately many studies of muddy sediments are confounded by temporal and spatial variability (Chapman & Tolhurst, 2004). There is limited information on the temporal scales at which biofilms mediate sediment properties and processes and how quickly this mediation could begin after a disturbance event. Such information is essential for experimental design and for including biosedimentological processes in models of sediment dynamics. This article presents data from a laboratory study examining the micro-scale changes in the properties of sediment associated with the growth and development of a biofilm over a period of 7 weeks. Whilst the data are primarily relevant to muddy estuarine sediments, they are also indicative of potential changes due to biofilm growth and development in sandy sediments and other sedimentary habitats such as lakes (Cyr & Morton, 2006).

Materials and methods

Cohesive sediment was collected from the Eden Estuary, Scotland (UK) (56°22′ N, 02°51′ W). Surface sediments were collected by removing the upper few cm of sediment with a shovel. The sediment was sieved in seawater through a 1 mm sieve to remove larger macrofauna and macroalgae to form one composite sample and then frozen at −70°C to kill organisms (Ford et al., 1999). The sediment was thawed, homogenised and divided amongst 35 petri dishes, then divided amongst four aquarium tanks with a flow through of natural seawater, so a biofilm could develop. The samples were constantly immersed and illuminated with a 300 W halogen light source (12 Light:12 Dark) which provided ∼650 μmol m−2 s−1.

On days 0, 1, 3, 5, 7, 13 and 45 the sediments in five randomly selected replicate dishes were tested for stability using the Cohesive Strength Meter (CSM). After measurement with the CSM three of the dishes were randomly selected and quench frozen in liquid nitrogen and stored at −20°C for later biogeochemical analysis.

Epipelic diatom activity is usually limited to the top 2 mm of intertidal sediments (Paterson et al., 1986; Hay et al., 1993; Consalvey et al., 2004). One reason for the poor progress in understanding biota—sediment interactions is a failure to measure on relevant scales (Wheatcroft & Butman, 1997; Flemming & Delfontaine, 2000; Kelly et al., 2001). If the mechanisms by which diatom biofilms influence sediment properties are to be understood, it is vital to investigate these properties at a scale relevant to diatom influence i.e. the sub-millimetre scale. A section of the frozen sediment samples undisturbed by the CSM measurement was cut using a diamond tipped lapidary saw into a block of approximately 1 cm2. The exact size of the block was measured, so that biogeochemical data could be expressed as a concentration where appropriate. These blocks were micro-sectioned on a freezing microtome in 200 μm sections down to 1000 μm, then 500 μm sections to 2000 μm. The wet weight was measured before biochemical analysis and the dry weight after analysis. For the graphs, data are expressed as the midpoint of the section, so the top 200 μm section is designated 100 μm the next 300 μm and so on. The sections were analysed for bulk density, water content, chlorophyll a and colloidal carbohydrate to determine the distribution of these variables with depth.

Previous work has shown that biogeochemical data from soft sediments expressed as contents can be confounded by spatial and/or temporal variability in grain size, sample density and dewatering (Flemming & Delafontaine, 2000; Perkins et al., 2003; Tolhurst et al., 2005). Thus colloidal carbohydrates and chlorophyll a were expressed as concentrations. Water was expressed as % content.

Chlorophyll a concentration (as a proxy for biomass) was determined using High Performance Liquid Chromatography (HPLC) analysis (Wiltshire et al., 1998). The Dubois assay (Dubois et al., 1956 as modified by Taylor & Paterson, 1998) was used to measure the amount of water-extractable colloidal carbohydrates in each sectioned sample. The sectioned sediment samples were also used to ascertain water content and wet bulk densities were calculated using the following formula:

$$ {\text{Bulk}}\,{\text{density}} = \frac{{{\text{Wt}}_{{{\text{wet}}}} }} {V} $$
(1)

where Wtwet = wet mass in grams, V = volume in cm3.

A CSM was used to measure the critical erosion threshold of the sediments immediately prior to sampling for the biogeochemical analysis (Tolhurst et al., 1999). The CSM measures over spatial (cm) and temporal (minutes) scales impossible with traditional flume systems, this is vital for measuring changes without spatially or temporally confounding the data (Tolhurst et al., 2000, 2006). In this case, it enables the measurement of laboratory cultures on a scale impossible with any other device.

The CSM consists of a 30-mm diameter chamber that is pushed into the sediment and filled with local seawater (Tolhurst et al., 1999). The jet of water comes from a downward directed nozzle in the chamber, the velocity of the jet is increased systematically through each experiment. Bed erosion is inferred from the drop in the transmission of infrared light across the chamber caused by the suspension of sediment. A drop in transmission below 90% is taken as a critical drop (Tolhurst et al., 1999), and is approximately equal to erosion of 0.01 kg m−2 of cohesive sediment. The slope of the averaged erosion profiles was used to give a relative measure of the erosion rate, the suspension index (Si).

The microstructure of the sediment/biofilm as well as the occurrence of diatoms and the distribution and microstructure of the EPS in relation to the diatoms was investigated using Low Temperature Scanning Electron Microscopy (LTSEM; Paterson, 1995). Sediment samples were fractured and mounted on to specially constructed cryostubs under liquid nitrogen. The cryostubs were then placed on to an adapted stage of a JOEL 35CF SEM fitted with a LTSEM (Oxford Instruments CT 1500B). Samples were freeze dried under vacuum and sputter coated with gold before being examined at −170°C. Further details of the examination of frozen sediment/biofilms under LTSEM are given by Paterson (1995).

Data were analysed using the Gmav5 programme for analysis of variance and SNK tests for sequential pairwise comparison of ranked means. For the CSM erosion threshold there was one factor (time) with 7 levels, orthogonal and fixed, n = 5. The Si value was an average of the profiles, thus it was not possible to do ANOVA. For the biogeochemical properties two analyses were done, the first was on the average of all the sections i.e. the top 2 mm of the sediment. This analysis had one factor (time) with 7 levels, was orthogonal and fixed, n = 3. For the analysis of changes in the depth profile of the biogeochemical variables with time, time and core depth are both fixed factors, thus it was not valid to do an ANOVA on these data. Instead, the slope of the depth profile was calculated for each variable in the three replicate cores at each time and ANOVA was done on these data to determine if there were any changes in the depth profile with time. This analysis had one factor (time) with 7 levels, was orthogonal and fixed, n = 3. Differences were determined using SNK tests. Pearson’s correlation analysis was done for the whole 2 mm section and the 100 μm and 1750 μm sections to investigate relationships at the surface and deeper in the sediment.

Results

For the 2 mm section, erosion threshold, water content and colloidal carbohydrate concentration were all significantly positively correlated with time (Table 3a, Figs. 1, 2). Erosion threshold was significantly positively correlated with water content and colloidal carbohydrate concentration, which were positively correlated with each other. For the top 200 μm section, erosion threshold, water content, chlorophyll a and colloidal carbohydrate were all significantly positively correlated with time (Table 3b). Erosion threshold was significantly positively correlated with water content, chlorophyll a and colloidal carbohydrate. Water content and chlorophyll a were significantly positively correlated with colloidal carbohydrate. For the 1750 μm section, erosion threshold and water content were significantly positively correlated with time (Table 3c). Erosion threshold was significantly positively correlated with water content and bulk density was significantly positively correlated with chlorophyll a and colloidal carbohydrate.

Fig. 1
figure 1

CSM erosion thresholds (squares) show that stability increases with time as the laboratory diatom biofilm develops. Error bars are standard error, n = 5. The relative erosion rate, Si values (circles), decreases with time

Full size image
Fig. 2
figure 2

Changes in the biogeochemical properties of the sediment with time, averaged over the surface 2 mm, (A) water content, (B) bulk density, (C) chlorophyll a concentration, (D) colloidal carbohydrate. Error bars are standard error, n = 3

Full size image

Low Temperature Scanning Electron Microscopy images of the sediments revealed changes in sediment structure as the biofilm developed with time. Samples from day 0 and day 1 exhibited the typical open “card-house” structure (Paterson, 1995) with only the odd solitary diatom (Fig. 7A, B, C). However by day 3 large numbers of small pennate diatoms had colonised the sediment surface forming a biofilm (Fig. 8A, B). The numbers of diatoms and thickness of the biofilm increase throughout the experiment and by day 45 the biofilm had become multilayered, epipelic diatoms are found on the surface of the biofilm, with chain-forming species below (Fig. 8C), harpacticoid copepods were also present in the biofilm at this time (Fig. 8D).

There was a significant increase in the erosion threshold of the sediment over time (Fig. 1 and Table 1). Initially the increase in erosion threshold was linear, but the rate of increase slowed after 5 days. The SNK test showed that day 0 = day 1 < day 3 = day 5 = day 7 < day 13 < day 45. The increase in erosion threshold was accompanied by a decrease in the Si values with time, showing that the erosion rate decreases as the biofilm develops (Fig. 1).

Table 1 Results of GMAV analysis of variance on changes in biogeochemical variables with time, n = 5 for CSM and 3 for other variables
Full size table

The water content of the surface 2 mm increased significantly with time (Fig. 2A, Table 1), the SNK test showed that day 0 < day 5 = day 3 = day 1 = day 13 = day 7 < day 45. Analysis of the water content depth profiles showed that there was a significant change with a general increase over time (Table 2). The SNK test could not, however, determine where the differences were, sometimes water content decreased with depth and sometimes it increased and there was no consistency to this pattern (Fig. 3). Bulk densities in the surface 2 mm were highly variable (Fig. 2B, Table 1), there was a significant effect of time, but the SNK test could not determine where the differences were. The bulk density depth profiles showed no significant change with time (Fig. 4, Table 2). There was a significant change in chlorophyll a concentration in the top 2 mm with time (Fig. 2C, Table 1). Chlorophyll a concentration generally increased with time, the SNK test showed that day 3 = day 0 = day 1 = day 7 = day 45 = day 5 < day 13. The chlorophyll a depth profiles were highly variable and whilst there was a general trend for chlorophyll to increase, showed no significant change with time (Fig. 5, Table 2). There was a significant change in colloidal carbohydrate concentration in the top 2 mm with time (Table 1), it initially increased, followed by a decrease then a steady increase (Fig. 2D). However, the SNK test could not determine where the differences were. The carbohydrate concentration depth profiles showed a significant change with time (Fig. 6, Table 2), although with inconsistent differences between depths from day to day. The SNK test showed that day 45 = day 1 < day 7 = day 13 = day 0 = day 5 = day 3.

Fig. 3
figure 3

Depth profile, showing the changes in water content with depth over time, (A) day 0, (B) day 1, (C) day 3, (D) day 5, (E) day 7, (F) day 13 and (G) day 45. Error bars are standard error, n = 3

Full size image
Fig. 4
figure 4

Depth profile, showing the changes in bulk density with depth over time, (A) day 0, (B) day 1, (C) day 3, (D) day 5, (E) day 7, (F) day 13 and (G) day 45. Error bars are standard error, n = 3

Full size image
Table 2 Results of GMAV analysis of variance on changes in the slopes of the biogeochemical depth profile with time, n = 3
Full size table
Fig. 5
figure 5

Depth profile, showing the changes in chlorophyll a concentration with depth over time, (A) day 0, (B) day 1, (C) day 3, (D) day 5, (E) day 7, (F) day 13 and (G) day 45. Error bars are standard error, n = 3

Full size image
Fig. 6
figure 6

Depth profile, showing the changes in colloidal carbohydrate concentration with depth over time, (A) day 0, (B) day 1, (C) day 3, (D) day 5, (E) day 7, (F) day 13 and (G) day 45. Error bars are standard error, n = 3

Full size image

Discussion

The growth of a biofilm significantly altered the biogeochemical properties of the sediment. There was an increase in microphytobenthos with time, the LTSEM images and chlorophyll a data demonstrated that diatoms rapidly colonise sediment and form a biofilm. The formation of the biofilm was accompanied by changes in the other measured sediment properties over time and there was strong correlation between chlorophyll a, colloidal carbohydrate, water content erosion threshold and time in the top 200 μm. This agrees with previous work showing such correlations (see for example: Underwood and Smith, 1998; Paterson et al., 2000; de Brouwer et al., 2000; Bellinger et al., 2005), supporting the conclusion that changes in these properties can be caused by growth and development of a diatom biofilm. Visually the structure of the biofilm in the first few days is consistent with images from natural biofilms (see for example Stal & Défarge, 2005); however, the structure at the end of the experiment is less likely to be found naturally, because of disturbance by macrofauna and wave/tidal currents.

The depth profile showed that both bulk density and water content were highly variable with depth, showing no clear pattern. Blistering of the biofilm caused sections of biofilm to be lifted up, revealing fresh sediment that was then colonised by new biofilm. LTSEM images showed that as the biofilm grew, it became multilayered, with epipelic diatoms overlying chain forming species. This layering and mosaic of older and newer biofilm is the most likely cause of the variable bulk density and water content. Many field studies have shown that areas with a biofilm have higher water contents than areas without, but these data are often confounded by concomitant differences in sediment grain size. This study shows that significant differences in water content between areas with and without a biofilm can occur simply through the growth of the biofilm.

Depth profiles showed that chlorophyll a and colloidal carbohydrate concentration was largest at the surface and decreased with depth, this pattern generally tended to strengthen with time (Figs. 5, 6), although the large variability in the data meant that for chlorophyll a these changes were not significant. This supports previous work showing the same pattern when data are expressed as contents (Taylor and Paterson, 1998; de Brouwer & Stal 2001; Kelly et al., 2001; Herlory et al., 2004) and refutes the contention that such profiles are entirely a function of density changes with depth (Flemming & Delafontaine, 2000). Flemming and Delafontaine’s point remains valid however, if density does increase with depth the pattern of values decreasing with depth would be artificially enhanced if data were expressed as contents. The biogeochemical variables are most strongly correlated with each other in the top 200 μm section (Table 3). The LTSEM images show that this is also the section where the majority of the biofilm occurs; this is also true for natural biofilms (see for example Tolhurst, 2003; Herlory et al., 2004; Stal & Défarge, 2005). This suggests that in this case the changes that occur in these variables over the course of the experiment are due to the biofilm and its growth. Whilst correlations of water content and carbohydrate might be expected for water-soluble carbohydrates, there is no correlation between these variables in the 1750 μm section, indicating that the strong correlation at the surface is due to the presence of the biofilm and not a causative relationship between the amount of water and amount of colloidal carbohydrate.

Table 3 Pearson's correlation of sediment properties, bold values are significant at P < 0.05, n = 7
Full size table

The rather bizarre pattern in the colloidal carbohydrate in the surface 2 mm, with large concentrations of carbohydrate on day 0 increasing on day 1 and a massive drop on day 3 followed by a steady increase over the rest of the experiment were unexpected. It is unclear what caused this pattern. It is possible that the increase at the beginning of the experiment was due to bacterial breakdown of insoluble organic matter, such as organisms killed by the freezing, into water soluble colloidal carbohydrate, such as bacterial EPS. Unfortunately no samples were taken for bacterial analysis, so this remains conjecture. The decrease on day 3 would then be explained by conversion of the water-extractable fraction of carbohydrate into an insoluble form or by respiration. Subsequently, the colloidal carbohydrate exhibits a logarithmic increase, presumably reflecting the growth and development of the biofilm and production of water-extractable carbohydrate due to photosynthesis (Stal & Défarge, 2005).

In abiotic muddy sediments there is an inverse relationship between stability and water content, if sediment stability were governed purely by the abiotic properties there would be a positive correlation with the bulk density and negative correlation with water content. This was not the case. In this study the biofilm had a primarily stabilising effect on the sediments, although oxygen blistering did lead to localised surface destabilisation similar to that found by Sutherland et al. (1998). The increase in erosion threshold with time was not due to compaction, because there was no significant increase in bulk density with time. The positive correlation with colloidal carbohydrates and chlorophyll a supports various studies indicating that the stabilising effect of biofilms is due to secretion of EPS (Holland et al., 1974; Tolhurst et al., 2002) and armouring of the sediment surface by the diatom cells themselves (Tolhurst et al., 2003). However, it has been shown that the structuring of the EPS by the organisms is also important in stabilising the sediment (de Brouwer et al., 2002, 2005). The positive correlation with water content shows that this stabilisation is larger than any destabilisation due to increased water content supporting previous findings where large water contents have been associated with large erosion thresholds (e.g. Underwood & Paterson, 1993; Tolhurst et al., 2000) (Fig. 7, 8).

Fig. 7
figure 7

LTSEM images showing changes in the sediment microstructure and biofilm growth with time in the laboratory. (A) Cross sectional fracture face of the sediment on day 1 showing structure with depth, arrow indicates the surface of the sediment, the sediment exhibits the typical open ‘card-house’ structure of cohesive sediments, scale bar = 100 μm, (B) the surface of the sediment on day 1 is bare, scale bar = 100 μm, (C) a close up image shows the surface to consist primarily of mineral grains with only the occasional small diatom, scale bar = 10 μm, (D) a fracture face of the sediment on day 3 shows that some structural changes in the arrangement of sediment particles has occurred (compare to 7a), scale bar = 100 μm

Full size image
Fig. 8
figure 8

(A) by day 3, the sediment surface is covered in a loose organic layer, scale bar = 100 μm, (B) a close up image reveals this to consist of mineral grains and many small diatoms, scale bar = 10 μm, (C) by day 45 the biofilm (bottom) has peeled away from the sediment to reveal a network of filaments just below the biofilm(top), scale bar = 100 μm, (D) a close up image reveals these filaments to be chain forming species, as well as the presence of Harpacticoid copepods (centre emerging from the biofilm), scale bar = 100 μm

Full size image

On day 0, as the sediments were submerged in water for the CSM measurements, the surface tension of the water was seen to remove sediment flocs. This did not occur on day 1, indicating some stabilisation of the sediment surface had occurred. By day 3 a biofilm was visible to the naked eye and by day 5, this had begun to blister as oxygen bubbles formed in the biofilm. On day 6, the biofilm was considerably blistered by oxygen bubbles and was being pulled away from the sediment by the small water currents in the tank from the flow through of seawater. On day 7, whilst removing the dishes from the culture tank pieces of biofilm, and associated sediment, were washed away from the sediment surface, despite great care being taken not to disturb the sediment. After day 7, the blistering appeared to have reached its peak and no new bubbles formed. By day 13, despite the sediment surface being highly blistered, none of the biofilm floated away. The blistering and lifting away of the biofilm is likely to increase spatial variability in the biogeochemical properties of the sediment.

This work demonstrates that newly exposed sediments can be rapidly colonised by microphytobenthos supporting an earlier in situ study by Consalvey et al. (2003). It also shows that the growth of a biofilm can significantly alter the properties of the sediment it is growing on, supporting field measurements that have shown significant differences in the properties of sediments with a biofilm to those without one (Tolhurst et al., 2000, 2006; de Brouwer et al., 2005). This data can be used during experimental design, for example in the timing of measurements after disturbance or biocide treatments. Changes in biogeochemical properties can be expected after just one day, especially in the absence of fauna. The data also show directional responses in biogeochemical variables due to biofilm growth and development, this information can be used to make predictions about the expected magnitude of changes within experiments and aid in the interpretation of data, for example in understanding interactions between diatoms and fauna.