Differential effects of microorganism–invertebrate interactions on benthic nitrogen cycling

Authors

  • William W. Gilbertson,

    Corresponding author
    1. Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen, UK
    • Oceanlab, Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeenshire, UK
    Search for more papers by this author
  • Martin Solan,

    1. Oceanlab, Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeenshire, UK
    2. Ocean and Earth Science, National Oceanography Centre, Southampton, University of Southampton, Southampton, UK
    Search for more papers by this author
  • James I. Prosser

    1. Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen, UK
    Search for more papers by this author

Correspondence: William W. Gilbertson, Oceanlab, Institute of Biological and Environmental Sciences, University of Aberdeen, Newburgh, Aberdeenshire AB41 6AA, UK. Tel.: +44 (0)1224 274432; fax: +44 122 427 2703; e-mail: w.gilbertson@abdn.ac.uk

Abstract

Infaunal invertebrate activity can fundamentally alter physicochemical conditions in sediments and influence nutrient cycling. However, despite clear links between invertebrate activity and microbially mediated processes such as nitrification, the mechanisms by which bioturbating macrofauna affect microbial communities have received little attention. This study provides strong evidence for differential stimulation of microbial nitrogen transformations by three functionally contrasting species of macrofauna (Hediste diversicolor, Corophium volutator, Hydrobia ulvae). Despite increased nitrification, abundance of ammonia-oxidising bacteria (AOB) and ammonia-oxidising archaea (AOA) at the sediment–water interface did not significantly change in the presence of macrofauna. However, species-specific differences in macrofaunal activity did influence ammonia oxidiser community structure, increasing AOB abundance relative to AOA in the presence of C. volutator or H. ulvae, but with no change in H. diversicolor and no-macrofauna treatments. Denaturing gradient gel electrophoresis profiles were similar between macrofaunal treatments, although one AOB band increased in relative intensity in the presence of C. volutator, decreased in the H. diversicolor treatment and was unchanged in the H. ulvae treatment. These data suggest that links between bioturbating macrofauna and nutrient cycling are not expressed through changes in the abundance of ammonia oxidisers in surface sediments, but are associated with changes in the AOA : AOB ratio depending on the invertebrate species.

Introduction

The productivity of marine systems tends to be limited by nitrogen availability (Galloway et al., 1996). Sediments are important sites for N cycling in coastal ecosystems (Laverock et al., 2011), and the active redistribution of sediment particles and porewater fluids by benthic invertebrate fauna plays a fundamental role in determining habitat quality (Godbold & Solan, 2009; Teal et al., 2010). Indeed, the active redistribution of particles (bioturbation) and fluids (bioirrigation) by infaunal invertebrates directly contributes to spatial and temporal heterogeneity of oxic and anoxic zones (Forster & Graf, 1992; Pischedda et al., 2008; Bertics & Ziebis, 2009; Teal et al., 2009), organic matter (OM) availability (Levin et al., 1997), pH (Stahl et al., 2006) and the distribution of metabolic electron acceptors (Fanjul et al., 2007), and therefore have considerable implications for microbial communities and nutrient cycling (Botto et al., 2005). Whilst the role of macroinvertebrate biodiversity in mediating nutrient cycling is well documented (e.g. Emmerson et al., 2001; Marinelli & Williams, 2003; Mermillod-Blondin et al., 2005; Ieno et al., 2006; Norling et al., 2007; Solan et al., 2008), the contribution from microbial ecologists to concepts of biodiversity–function remains in its infancy (Prosser et al., 2007). Moreover, the mechanistic interactions that lead to changes in the internal pool of dissolved and particulate nutrients are not fully understood (Hansen & Kristensen, 1997; Kristensen & Kostka, 2005), despite the accepted importance of bioturbation for levels of ecosystem productivity, biomass and nutrient cycling and availability over recent and geological timescales (Martin, 1996).

In sediments rich in OM, mineralised ammonia is the principal substrate for nitrifying microorganisms in the oxic zones (Aller, 1982; Tobias et al., 2003). Although several factors are known to regulate nitrogen transformation, including substrate availability, temperature, pH, redox and nutrient enrichment (DeLaune et al., 1981; Sloth et al., 1995; Baeseman et al., 2006; Bissett et al., 2009; Beman et al., 2011; Wankel et al., 2011), many of these variables are strongly influenced by macrofaunal activities. Indeed, increased ammonium (math formula) flux from the sediment to the overlying water column is frequently observed in the presence of burrowing invertebrates (Pelegri & Blackburn, 1994; Mermillod-Blondin et al., 2004; Jordan et al., 2009). However, rather than simply being a product of the relative amount of OM, math formula is known to reflect the active redistribution of labile material from the surface to deeper zones (Levin et al., 1997) and the combined effects of invertebrate excretion, bioirrigation and rates of microbial mineralisation (Papaspyrou et al., 2007). These regulatory factors can simultaneously affect aerobic and anaerobic processes within the nitrogen cycle and, in each case, may lead to differential net effects on ecosystem functioning because microorganism–invertebrate interactions are context-dependent (Rossi et al., 2008; Godbold et al., 2009; Hiddinck et al., 2009; Langenheder et al., 2010). For example, increased sediment oxygenation resulting from infaunal activities, particularly bioirrigation and sediment redistribution (e.g. Vopel et al., 2007), may inhibit anaerobic processes, such as denitrification, whilst simultaneously stimulating aerobic processes, such as nitrification. However, coupling of these processes through an oxic–anoxic interface may result in a net loss of bioavailable N (An & Joye, 2001), but the outcome will largely depend on infaunal community structure and how individual species respond to the dynamics and configuration of the habitat (Biles et al., 2002; Ieno et al., 2006; Bulling et al., 2010; Godbold et al., 2011).

Any changes in sediment conditions associated with invertebrate activities may be reflected in nitrifier abundance or community structure, including the ratio of ammonia-oxidising archaea (AOA) (belonging to the thaumarchaea, Spang et al., 2010) to ammonia-oxidising bacteria (AOB). Whilst there is debate over the relative contributions of AOA and AOB in various environments (reviewed in Prosser & Nicol, 2008), there is evidence that they respond differentially to environmental conditions, in particular math formula concentration (Martens-Habbena et al., 2009), such that microbial contributions to ammonia oxidation will be affected by microenvironmental conditions. These community shifts may be significant at the ecosystem level when linked with altered rates of nitrification, but there is currently no consensus on the relative contribution of AOA and AOB to nitrification in estuarine environments. Whilst some studies have found greater abundance of AOA (e.g. Caffrey et al., 2007; Abell et al., 2010), the reverse may also be true under alternative environmental conditions, in particular in the more saline regions of an estuary (Mosier & Francis, 2008; Santoro et al., 2008) and in model systems using higher levels of salinity, reflecting that of sea water (Wankel et al., 2011). Indeed, a difficulty in determining the relative importance of AOA and AOB is that their ratio will vary with both spatial and temporal changes in context, including seasonality (Magalhães et al., 2009).

Whilst it is clear that microbial activity and biogeochemical processes tend to be influenced by the presence of macrofauna, studies investigating the consequences for microbial community structure and function have typically used bulk parameters, such as total bacterial cell counts or microbial biomass. Such studies show that bacterial abundance tends to be positively stimulated by the presence of bioturbating invertebrates (Stoeck & Kroncke, 2001; Dufour et al., 2008; van Nugteren et al., 2009), most often expressed as increases in total cell abundance and/or proportional differences in the number of active bacteria (Mermillod-Blondin et al., 2004). It is important to emphasize that such changes in bacterial populations are dependent on invertebrate species identity and even vary between individual burrow structures within the population (Laverock et al., 2010). It is not clear, however, which faunal-induced mechanisms lead to alternative shifts in microbial assemblage structure. Kristensen & Kostka (2005) have proposed that, rather than simply extending the sediment–water interface, burrows provide microniches that are distinct from both subsurface and surface sediment in terms of physical and chemical stability and therefore support unique microbial communities; this concept is supported by molecular fingerprinting from ghost shrimp (Pestarella tyyrhena) burrows (Papaspyrou et al., 2005). Other authors, however, argue that microbial communities are affected by invertebrate behaviour (in particular bioirrigation) that influences geochemical properties and oxygen concentration of the burrow such that, where burrow conditions resemble that of surface sediment, microbial communities are more likely to converge (Bertics & Ziebis, 2009).

Irrespective of the mechanistic basis, the similarity (or otherwise) of microbial communities inhabiting biogenic structures and those inhabiting the sediment–water interface is known to be directly affected by infaunal activity. Yet, there has been little investigation of the influence of infaunal burrowing on specific microbial functional groups and, consequently, the delivery of ecosystem functioning (Solan & Wigham, 2005). Where this has been studied, it has been difficult to establish a generic understanding; even within a single invertebrate species, bacterial communities vary between individual burrows (Laverock et al., 2010). Hence, the aim of the present investigation was to examine how the abundance and community structure of AOA and AOB, and the corresponding links to nitrogen cycling, are affected by representative, but ecologically contrasting, infaunal invertebrates in a model benthic system. Specifically, it was hypothesized that the presence of invertebrate fauna would increase the abundance of ammonia oxidisers, but that the relative response of AOA and AOB would depend on species-specific differences in invertebrate behaviour.

Methods

Sediment and invertebrate fauna

Sediment (40 L; mean particle size, 50 μm; silt content, 60%; and organic carbon content, 4–6%; Bulling et al., 2008) and individuals of three functionally contrasting invertebrates (Hediste diversicolor, Corophium volutator and Hydrobia ulvae) were collected from the Ythan estuary, Aberdeenshire, Scotland (57°20.085′N, 02°0.206′W). Hediste diversicolor, a neried polychaete, constructs extensive semi-permanent galleries that intersect the redox potential discontinuity and may reach depths > 15–20 cm. The mud shrimp, C. volutator, constructs U-shaped burrows, < 5 cm deep, that often extend to an elevated tubiculous terminus that enhances passive irrigation in the presence of near-bed flow. The activities of the surficial grazing gastropod, H. ulvae, are generally restricted to the sediment–water interface. All three species form codominants of most European temperate tidal mud flats.

Sediment was sieved (500 μm mesh size) with sea water to remove macrofauna and left to settle for 24 h to retain the fine fraction. Sediment was homogenised after removal of the supernatant sea water. To enable measurement of total accumulated ammonium and nitrate removal in the absence of nitrification, approximately half of the sediment was treated with a nitrification inhibitor prior to mixing (0.2 g L−1 dicyandiamide, DCD), which has no known effects on the invertebrate community (Aalders & Bell, 2008). Other than DCD addition, both sediment batches were treated identically. A total of 24 (12 DCD-treated) mesocosms (transparent acrylic cores, 30 cm high, 10 cm internal diameter) were filled to a depth of 10 cm sediment overlain by 20 cm sea water (UV-sterilised, 10 μm filtered) and left for 24 h before replacing the sea water to remove excess nutrients released during mesocosm assembly. Replicate (n = 3) faunal treatments (no macrofauna, H. diversicolor, C. volutator, H. ulvae) were assembled in untreated and DCD-treated mesocosms. To ensure that any observed effects were related to invertebrate species identity and not to differences in total biomass, macrofaunal biomass was held constant at 2.00 ± 0.12 g mesocosm−1. Mesocosms were incubated in the dark to prevent formation of microphytobenthos at a temperature of 12 ± 0.5 °C.

Nutrient and molecular analysis

Sampling was carried out immediately after the introduction of macrofauna and at 3-day intervals throughout the experiment (15 days). Sea water samples (12 mL) were analysed using a flow injection autoanalyser to determine concentrations of math formula and math formula). Surface sediment samples (~ 3 g) for molecular analysis were immediately frozen at −80 °C. Mineralization was estimated as the sum of the ammonium increase and total nitrification (i.e. ammonium removed). Nitrification was calculated as the sum of increases in nitrate under normal nitrifying conditions and nitrate removal from the system when nitrification was inhibited. Denitrification was estimated as the amount of nitrate removed from the system.

Sediment samples were centrifuged at 3000 g for 10 min to remove pore water, and nucleic acids were extracted using a bead-beating protocol, as described by Griffiths et al. (2000). Bacterial and thaumarchaeal amoA gene abundance was determined by qPCR amplification from extracted DNA following Nicol et al. (2008) but with alterations to reaction mixes to achieve optimal amplification (bacterial amoA, 0.5 mM primer concentration and 8 ng of template DNA; archaeal amoA, 1 mM primer concentration and 6 ng of template DNA for archaea). Standards were a dilution series of either genomic DNA from Nitrosospira multiformis (AOB) with amplification efficiency 94% or the 54d9 soil fosmid clone fragment for AOA (Tourna et al., 2008) with an amplification efficiency of 94.6%. Melting curve analysis and agarose gel electrophoresis were carried out on all qPCR products to confirm specificity of amplification.

Denaturing gradient gel electrophoresis (DGGE) fingerprinting analysis of AOB communities was performed on 16S rRNA genes, amplified from extracted DNA using a nested PCR approach (Freitag et al., 2006). Primary amplification used betaproteobacterial AOB-specific CTO189f and CTO654r PCR primers (Kowalchuk et al., 1997), and cycling conditions were 95 °C for 2 min; 15 cycles of 94 °C for 30 s, 55 °C for 30 s, 72 °C for 45 s; 20 cycles of 92 °C for 30 s, 55 °C for 30 s, 72 °C for 45 s and 72 °C for 5 min. PCR products were diluted 1 : 10 to prevent amplification of nontarget sequences and then reamplified using general bacterial primers P3 (357f-GC) and P2 (518r) (Muyzer et al., 1993). Cycling conditions were 95 °C for 5 min; 10 cycles of 94 °C for 30 s, 55 °C for 30 s, 72 °C for 40 s; 25 cycles of 92 °C for 30 s, 55 °C for 30 s, 72 °C for 40 s and 72 °C for 10 min. For DGGE analysis of thaumarchaeal ammonia oxidisers, amoA genes were amplified using primers CrenamoA23f and CrenamoA616r (Tourna et al., 2008). Cycling conditions for amplification were 95 °C for 5 min; 10 cycles of 94 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min; 25 cycles of 92 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min; 72 °C for 10 min. The DGGE system, denaturant gradients, electrophoresis conditions and staining procedures were as described by Nicol et al. (2008).

Data analysis

Linear regression models (anova) were used to investigate the effects of macrofaunal treatments on the response variables (nutrient concentration and amoA abundance). Model assumptions were validated using quantile–quantile plots (normality), residuals versus fitted values (homogeneity of variance) and Cook's distance plots (influence of outliers). Where necessary, data were square root- or log-transformed. All analyses were performed using the R statistical and programming environment (R Development Core Team 2010).

Results

Rates of mineralisation, nitrification and denitrification (derived from nutrient measurements) all increased in the presence of macrofauna, although the relative effect was species-specific in all cases. Despite these observed changes in process rates, the abundance of AOA and AOB did not change across invertebrate identity treatments, but there were changes in ammonia oxidiser community structure, with increases in the AOB : AOA ratio and subtle changes in AOB 16S rRNA gene DGGE profiles, particularly in the presence of C. volutator. In the broadest terms, the presence of H. diversicolor had the greatest influence on the ammonia-oxidising microbial community and nitrogen cycling, followed by C. volutator and H. ulvae. These differences reflect differences in the life-style traits and behaviour of the invertebrates as well as relative differences in the depth of the burrow structure.

Nutrient concentrations and process rates

Mean math formula concentration increased in all non-DCD-treated mesocosms containing invertebrate macrofauna, in each case reaching a maximum concentration at days 3 to 6 before slowly decreasing over the remaining incubation period (Fig. 1a). math formula concentration in mesocosms containing no macrofauna remained below the detection level (< 0.06 μM) in all but two samples (1.5 and 0.9 μM, both at day 15). By day 15, mean (± SD) math formula concentration in the presence of H. diversicolor was 26.1 ± 1.6 μM, whilst concentrations in the presence of C. volutator, H. ulvae and no macrofauna were much less at 3.1 ± 1.2, 1.3 ± 0.7 and 0.8 ± 0.7 μM, respectively (Fig. 1a). In the DCD-treated mesocosms, math formula concentration accumulated to 189 ± 7.0 μM in the presence of H. diversicolor, 157 ± 15.4 μM in the presence of C. volutator, 71.1 ± 2.4 μM in the presence of H. ulvae and 56.1 ± 2.7 μM in the absence of macrofauna (Fig. 1b).

Figure 1.

Changes in ammonium concentration during incubation of mesocosms (a) without and (b) with DCD and in the absence (no macrofauna) and presence of macrofauna (Hediste diversicolor, Corophium volutator or Hydrobia ulvae). Error bars show 95% confidence intervals (n = 3).

Mean (± SD) NOx-N concentrations steadily increased throughout the incubation period in all non-DCD-treated mesocosms, including those containing no macrofauna where the concentration increased from 8.0 ± 0.2 to 12.8 ± 0.5 μM over 15 days (dot-dash line, Fig. 2a). In the presence of macrofauna, mean NOx-N concentrations after 15 days were higher (H. diversicolor, 18.9 ± 1.6 μM; H. ulvae, 18.5 ± 0.1 μM), particularly in the case of C. volutator (23.4 ± 2.4 μM) (Fig. 2a). In contrast, mean NOx-N concentrations in DCD-treated mesocosms showed a rapid but saturating decline from the starting concentration (8.0 ± 0.2 μM) to 7.1 ± 0.4 μM (H. ulvae and no macrofauna), 5.2 ± 0.7 μM (C. volutator) and 1.0 ± 0.1 μM (H. diversicolor) after 15 days (Fig. 2b). These findings (increased math formula concentrations coupled with no increase or reduced mean NOx-N concentrations) provide reassurance that DCD inhibited nitrification.

Figure 2.

Changes in nitrate + nitrite (NOx-N) concentration during incubation of mesocosms (a) without and (b) with DCD and in the absence (no macrofauna) and presence of macrofauna (Hediste diversicolor, Corophium volutator or Hydrobia ulvae). Error bars show 95% confidence intervals (n = 3).

The effect of invertebrate species identity on microbial process rates (estimated from changes in the concentrations of math formula and math formula under normal nitrifying versus inhibited conditions) is shown in Fig. 3. Linear regression of mineralisation (log-transformed), with macrofaunal species identity as the explanatory variable revealed that rates of mineralisation increased in all species treatments relative to no macrofauna and that mineralisation was species-specific (F = 155.7, d.f. = 8, < 0.001). The greatest mean (± SD) rate of mineralisation occurred in the presence of H. diversicolor (90.6 ± 6.6 μM day−1; coefficient = 1.4, t = 19.9, < 0.001) followed by C. volutator (72.5 ± 8.8 μM day−1; coefficient = 1.2, t = 16.7, < 0.001), H. ulvae (43.5 ± 2.1 μM day−1; coefficient = 0.7, t = 9.6, < 0.001) and no macrofauna (22.1 ± 1.8 μM day−1).

Figure 3.

Estimated microbial process rates based on ammonium and nitrate fluxes in mesocosms with and without nitrification inhibitor (DCD) and in the absence (no macrofauna) and presence of macrofauna (Hediste diversicolor, Corophium volutator or Hydrobia ulvae).

Nitrification rate was also dependent on invertebrate species identity (linear regression, log transformation, F = 99.9, d.f = 8, < 0.001), and mineralisation and nitrification rates were greater when macrofaunal species were present. Greatest nitrification rates occurred in the presence of C. volutator (69.4 ± 9.1 μM day−1; coefficient = 1.2, t = 15.4, < 0.001) or H. diversicolor (64.3 ± 5.5 μM day−1; coefficient = 1.1, t = 14.5, < 0.001) followed by H. ulvae (42.2 ± 2.4 μM day−1; coefficient = 0.7, t = 9.0, < 0.001) and no macrofauna (21.2 ± 1.7 μM day−1). However, in contrast to mineralisation, there was no difference (coefficient = 0.07, t = 0.96, = 0.37) in the rate of nitrification between treatments containing C. volutator (69.4 ± 9.1 μM day−1) and H. diversicolor (64.3 ± 5.5 μM day−1).

Although denitrification reflected invertebrate species identity (linear regression, F = 151.9, d.f. = 8, < 0.001), in contrast to the patterns observed for mineralisation and nitrification, there was no significant difference (coefficient = 0.2, t = 0.1, = 0.89) between treatments containing H. ulvae (3.3 ± 2.1 μM day−1) and those with no macrofauna (3.5 ± 0.18 μM day−1). However, denitrification rate was greater in the presence of macrofauna, increasing to 10.0 ± 2.0 μM day−1 (coefficient = 6.5, t = 5.4, < 0.001) in the presence of C. volutator and, further, to 25.6 ± 0.4 μmoles L−1 day−1 in the presence of H. diversicolor (coefficient = 22.1, t = 18.4, < 0.001).

Sediment molecular analysis

The abundance of thaumarchaeal amoA varied between 0.28 × 106 and 2.4 × 106 g−1 at day 0 (Fig. S1a, Supporting information) and 0.35 × 106 and 1.3 × 106 g−1 at day 15 (Fig. S1b), but there was considerable variation within replicates (up to one order of magnitude, when H. ulvae was present). Bacterial amoA abundance was consistently greater than thaumarchaeal amoA abundance, ranging from 1.4–10 × 106 at day 0 (Fig. S1c) and 2.4–7.0 × 106 g−1 at day 15 (Fig. S1d). The relative change in thaumarchaeal amoA abundance (from day 0 to day 15) was not significantly affected by invertebrate identity (linear regression, F = 0.19, d.f. = 7, > 0.05), although there was some indication that mean thaumarchaeal amoA abundance may increase in the presence of H. diversicolor or H. ulvae and decrease when C. volutator is present or when macrofauna are absent (Fig. 4). Similarly, the relative change in bacterial amoA abundance over 15 days did not reflect invertebrate species identity (F = 0.23, df = 7, > 0.05), but there was some weak, albeit insignificant, evidence that bacterial amoA may increase in the presence of macrofauna and decrease in the absence of macrofauna (Fig. 4).

Figure 4.

Boxplots (n = 3) showing changes (final abundance/starting abundance) in ammonia oxidising bacterial (AOB, unshaded boxes) and thaumarchaeal amoA gene (AOA, shaded boxes) abundance between days 0 and 15 (i.e. > 100% indicates increased abundance and < 100% indicates decreased abundance) and in the absence and presence of macrofauna (species indicated on the x-axis). In each case, the median is indicated at the mid-point, the upper and lower quartiles are indicated by the hinges, and lines represent the spread.

Although there was no strong evidence for an effect of macrofauna on ammonia oxidiser abundance, the AOB : AOA ratio changed with invertebrate species identity (linear regression, F = 5.67, d.f. = 8, < 0.05). The mean AOB : AOA ratio at day 15 was the lowest (and reflected day 0 ratios) in the presence of H. diversicolor (6.0 : 1) and in the absence of macrofauna (6.2 : 1), rising to 7.9 : 1 for H. ulvae and 9.4 : 1 for C. volutator (Fig. 5). These differences in ratio were significantly different to the initial (day 0) ratios for C. volutator (coefficient = 4.32, t = 5.4, < 0.001) and H. ulvae (coefficient = 3.03, t = 3.8, < 0.01), and there was a marginally significant change when macrofauna were absent (coefficient = 1.71, t = 2.2, = 0.048), but not for H. diversicolor (coefficient = 0.67, t = 0.8, = 0.46). By the end of the experiment, and relative to when macrofauna were absent, AOB : AOA ratio was significantly higher in the presence of C. volutator (coefficient = 3.23, t = 4.1, < 0.01) and H. ulvae (coefficient = 1.73, t = 2.2, < 0.05) but not in the case of H. diversicolor (coefficient = −0.17, t = −0.2, = 0.83).

Figure 5.

Boxplots (n = 3) showing the ratio of AOB to AOA abundance, measured by quantification of amoA genes. AOB abundance was up to 10-fold higher than AOA abundance after 15 days (shaded boxes). In each case, the median is indicated at the mid-point, the upper and lower quartiles are indicated by the hinges, and lines represent the spread.

DGGE profiles of amplified thaumarchaeal amoA and betaproteobacterial 16S rRNA genes were generally consistent throughout treatments, and profiles from replicate mesocosms were reproducible (Fig. 6). A high number of relatively weak bands were present in AOA communities across all treatments. Only one mesocosm containing H. diversicolor showed any clear change in relative band intensity (lane 2, Fig. 6a). The dominant AOB bands were consistent for all treatments and at day 0. However, a band within the Nitrosomonas cluster, present at low intensity at day 0, became more intense in all H. diversicolor mesocosms (lanes 1–3, Fig. 6b), but was absent in all C. volutator profiles (lanes 4–6, Fig. 6b). DGGE profiles in mesocosms containing H. ulvae or no macrofauna did not differ from day 0 profiles (Fig. 6b).

Figure 6.

Replicate (n = 3) DGGE profiles of (a) thaumarchaeal amoA genes and (b) betaproteobacterial 16S rRNA genes of ammonia oxidisers in sediment incubated for 15 days in the absence (no macrofauna) and presence of macrofauna (Hediste diversicolor, HD; Corophium volutator, CV; Hydrobia ulvae, HU). Profiles for day 0 are also shown for comparison. ‘M’ is a marker lane containing nine 16S rRNA gene PCR products derived from sequences representative of seven bacterial ammonia oxidiser clusters; in descending order, Nitrosomonas (Nsm) cluster 7, Nsm 6, Nsm 5, Nitrosospira (Nsp) 4, Nsp 1, Nsp 2, Nsp 3 (3 bands).

Discussion

It has long been demonstrated that the biogeochemistry of coastal sediment communities is strongly influenced by the presence and activities of benthic invertebrates (Gray, 1974; Rhoads, 1974), and that microbial nitrogen transformations (mineralisation, nitrification, denitrification) are intimately coupled with the presence, activity and biodiversity of bioturbating invertebrates (reviewed in Covich et al., 2004; Laverock et al., 2011). Indeed, the relationship between increased levels of infaunal biodiversity and nitrogen cycling is invariably positive (Hansen & Kristensen, 1997; Emmerson et al., 2001; Mermillod-Blondin et al., 2005; Ieno et al., 2006; Bulling et al., 2008; Jordan et al., 2009; Bulling et al., 2010) and can make substantial contributions to ecosystem properties at larger scales (Teal et al., 2010; Godbold et al., 2011). The findings of the present study are consistent with these conclusions but place greater emphasis on the presence of strong invertebrate species identity effects which, in turn, directly affect the microbial mediation of nitrogen transformations. The predominate mechanistic explanation has been that bioturbating infaunal invertebrates positively affect the abundance and activity levels of microbial populations through the provision of new habitat following particle displacement (Hylleberg, 1975) and/or the regulation of substrate supply during bouts of bioirrigation (Yingst & Rhoads, 1980; Sander & Kalff, 1993; Traunspurger et al., 1997). The approach we adopted in this study, however, has allowed us to ascertain that increased levels of microbial transformation are not expressed through changes in the abundance of ammonia oxidisers in surface sediment, but there were shifts in the AOA : AOB ratio. Inhibition of nitrification confirmed that the associated changes in nutrient concentrations reflected genuine rate changes, rather than simply the release of stored nutrients.

A naïve interpretation of our observations is that the effects of invertebrate activity on microbially mediated ecosystem processes are simply a function of the extent and intensity of bioturbation. However, the complex metabolic requirements of microbial pathways mean that invertebrate–microbial relationships must also relate to the form of burrowing, and other activities such as feeding, irrigation, biogenic structure construction and mucus secretions (e.g. Hansen et al., 1996; Bird et al., 2000; Kristensen, 2000; Wu et al., 2003; Papaspyrou et al., 2005). Indeed, differences in burrow morphology and in species behaviour (here, mostly differences in the propensity to bioirrigate; Pelegri & Blackburn, 1994; Hedman et al., 2011) can explain the differential effects of invertebrate species on nitrification and denitrification. For example, H. diversicolor constructs discrete, large burrows that penetrate deeply (~ 10 cm) into the suboxic sediment zones, whereas the activities of C. volutator generate fast turnover of the uppermost ~ 2–5 cm of the sediment profile. As such, an obligate aerobic process, such as nitrification, is favoured by both irrigation and increased surface area in H. diversicolor burrows and in highly oxygenated areas with C. volutator activity. In contrast, denitrification is stimulated at the oxic–suboxic interface, which is more extensive in N. diversicolor burrows than in those of C. volutator. This supports the hypothesis that invertebrate burrows are not simply extensions of the sediment–water interface, but in fact support distinct communities that vary according to the form of burrowing (Kristensen & Kostka, 2005). This explanation does not, however, explain why surficial grazing species, such H. ulvae, also stimulate nitrification and mineralisation. Instead, it is likely that H. ulvae affect microbial communities through active grazing (compensatory response, e.g. Langenheder & Jurgens, 2001; Godbold et al., 2009) and/or mucus secretion, as mucus and waste production can result in significant changes in bacterial community structure (Amon & Herndl, 1991; Dufour et al., 2008). Different combinations of bioturbating invertebrates will therefore support mixed microbial populations, but some combinations will favour certain microbial subsets over others depending on how the type, extent and variety of bioturbation affect substrate availability and other environmental conditions (Langenheder et al., 2010; Peter et al., 2011). Collectively, these findings suggest that whilst overviews of bacterial abundance in relation to fauna may be useful, investigation of specific microbial functional groups are fundamental in establishing a generic mechanistic understanding of how the invertebrate–microbial couple affects nutrient cycling.

Despite the differences in microbial transformation rates, no significant change in archaeal or bacterial amoA gene abundance was observed. There was, however, a weak but insignificant trend for increased abundance of AOB amoA genes under H. diversicolor and C. volutator. Nevertheless, we acknowledge that it is possible that, if changes occurred, they may not have been detected owing to high levels of spatial heterogeneity in microbial populations at the sediment surface. Alternatively, it is possible that an increase in the proportion of the community that is active resulted in increased process rates, without any corresponding chances in ammonia oxidiser community structure (e.g. Nicol et al., 2008). Higher bacterial abundance has been recorded previously in samples taken from the burrow walls themselves (Papaspyrou et al., 2005; Dufour et al., 2008; Laverock et al., 2010), but it is still unclear whether prokaryotic abundance is affected in the sediment as a whole. Whilst a number of studies have reported higher abundance of AOA in both marine and terrestrial environments, in this case bacterial amoA genes outnumber AOA amoA genes, with AOB : AOA ratios ranging from ~ 5 : 1 at day 0 to ~ 9 : 1 in the presence of C. volutator. Studies of relative abundances of thaumarchaeal and bacterial amoA in estuarine ecosystems have resulted in differing conclusions; for example, Caffrey et al. (2007) found higher thaumarchaeal abundance in number of estuaries, whilst Mosier & Francis (2008) showed that bacterial amoA outnumbered thaumarchaeal amoA, except those with lowest salinity. As full strength sea water was used in the mesocosms, the dominance of AOB in this study concurs with previous evidence linking AOB dominance to areas of higher salinity in estuarine systems (Mosier & Francis, 2008; Santoro et al., 2008; Wankel et al., 2011). The factors that determine the relative abundance of the two groups are likely to be complex, as they appear to share a common substrate and ecological niche. Whilst the abundance of bacterial and thaumarchaeal amoA genes was variable even at day 0 and did not significantly change with invertebrate activity, the AOB : AOA ratio was less variable at day 0 (~ 5 : 1), but did change with invertebrate activity. Although bacterial amoA dominated in all treatments, there was a significant decrease in the relative abundance of thaumarchaea in C. volutator mesocosms. It is possible that the bacteria benefit from the higher ammonium concentrations generated by intensive reworking of the sediment–water interface, burrow construction and the associated irrigation activity, whereas the low ammonia conditions generated in the absence of macrofauna and species that have little impact on the sediment profile, such as H. ulvae, benefit other groups, specifically the thaumarchaea. It is unclear whether this niche separation exists universally, but all three cultivated representatives of AOA thrive at low ammonia concentration (Schleper, 2010) and growth of AOA in some soils has been seen to increase under low nitrogen loads (Di et al., 2010).

DGGE profiles were surprisingly consistent across treatments and replicates, given the variability in total amoA abundance. This may reflect the resolution of DGGE, and higher resolution techniques may have identified treatment effects on community composition, but DGGE and other fingerprinting techniques have successfully identified changes in thaumarchaeal and bacterial ammonia oxidiser community composition in other studies (see, for example, Prosser & Nicol, 2008). Bertics & Ziebis (2009) suggest that microbial diversity in bioturbated sediment is primarily driven by oxygenation. As all samples were taken from surface sediment, oxygen conditions are likely to be similar across treatments and may therefore explain the observed similarity in community profiles. However, there is evidence that burrowing does affect the surface sediment to some extent; the relative intensity of a weak Nitrosomonas band increased after 15 days in the presence of H. diversicolor, but was absent in the presence of C. volutator. Elevated nutrient and OM levels (including those from mucus secretions) under H. diversicolor burrowing may be driving the change; Magalhães et al. (2007), in a study of betaproteobacterial AOB composition in estuarine sediments, concluded that high nutrient, OM and chlorophyll levels favoured Nitrosomonas, whilst low nutrient levels favoured Nitrosospira. However, other mechanisms must also operate, as the presence of C. volutator results in increased nutrient levels, yet selected against Nitrosomonas. Previous analysis of sediment from the Ythan estuary indicates Nitrosospira-type sequences at marine influenced locations, whilst Nitrosomonas dominated freshwater communities (Freitag et al., 2006). The site used here is brackish but has a strong marine influence.

An important aspect of our findings is that microbial community structure and the capacity for such communities to contribute to important ecosystem processes are clearly influenced by species-specific life-history characteristics of macrofaunal invertebrates and, by extension, other biotic and physical processes that affect the sediment profile at a range of temporal and spatial scales (Sapp et al., 2010). Whilst much is known about how individual drivers of change may affect ecosystem functioning in microbial (e.g. biodiversity loss, Bell et al., 2005; temperature, Gudasz et al., 2010; environmental complexity, Langenheder et al., 2010) and invertebrate communities (e.g. biodiversity loss, Solan et al., 2004; temperature, Sanz-Lazaro et al., 2011; environmental configuration, Godbold et al., 2011; ocean acidification, Widdicombe & Needham, 2007), a critical knowledge gap is how the network of relationships between multiple abiotic and multitrophic biotic factors vary in strength over space and time and link to observed levels of ecosystem functioning. Whilst this study provides information on the broader-scale surface sediment microbial communities, there are clearly further mechanistic links to be investigated that were outside the scope of this study. For example, localised sampling of specific burrow structures will contribute to our understanding of how bioturbation affects microbial functional groups at different scales in sediments. Understanding how and when bacterial–archaeal–invertebrate interactions underpin nitrogen and other elemental transformations in a changing environment will be critical if we are to predict future levels of ecosystem sustainability and overall functioning in marine systems.

Acknowledgements

W. Gilbertson is funded by a University of Aberdeen 6th Century studentship. We thank Fiona Murray for assistance with specimen collection and Jenna Mcwilliam for assistance with water sample analysis.

Ancillary