The effects of bioturbation and herbivory by the polychaete Nereis diversicolor on loss of saltmarsh in south-east England

Authors

  • O. A. L. PARAMOR,

    1. School of Biological Sciences, Queen Mary and Westfield College, University of London, London E1 4NS, UK
    Search for more papers by this author
    • *

      Present address: Dove Marine Laboratory, School of Marine Science and Technology, University of Newcastle, Cullercoats, North Shields NE30 4PZ, UK.

  • R. G. HUGHES

    Corresponding author
    1. School of Biological Sciences, Queen Mary and Westfield College, University of London, London E1 4NS, UK
      R. G. Hughes, School of Biological Sciences, Queen Mary and Westfield College, University of London, London E1 4NS, UK (e-mail r.hughes@qmul.ac.uk).
    Search for more papers by this author

R. G. Hughes, School of Biological Sciences, Queen Mary and Westfield College, University of London, London E1 4NS, UK (e-mail r.hughes@qmul.ac.uk).

Summary

  • 1The saltmarshes of south-east England are eroding rapidly. Field and laboratory experiments were used to test the hypotheses that: (i) at the mudflat–saltmarsh boundary there are two alternative states, one dominated by pioneer zone vegetation that excludes burrowing infauna, and the other dominated by infaunal invertebrates that exclude vegetation; and (ii) the major cause of the loss of saltmarshes in south-east England is internal creek erosion, which is exacerbated by bioturbation and herbivory by the infaunal polychaete Nereis diversicolor.
  • 2In laboratory experiments Nereis ate the seeds and seedlings of Salicornia spp., and Salicornia deterred burrowing by Nereis. In field experiments, at Tollesbury in Essex, UK, exclusion of Nereis from the sediment surface increased the density of Salicornia, but only when a source of seeds was close by. In the Tollesbury saltmarsh the plants and Nereis had mutually exclusive distributions within a vertical zone of overlap. The recently vegetated area of the managed realignment site at Tollesbury contained no Nereis, but Nereis colonized areas where Salicornia had been removed. These observations and data support the first hypothesis.
  • 3Much of the loss of the Tollesbury saltmarsh is by lateral erosion of the internal creeks. Physical factors alone cannot be responsible for this erosion because experimental exclusion of Nereis led to sediment accretion. These results support the second hypothesis. Creek erosion may create a positive feedback where creek enlargement leads to faster tidal currents and further erosion until creeks have widened to their new equilibrium morphology.
  • 4Synthesis and applications. We conclude that the infauna are a major cause of reduction in area of saltmarsh vegetation and this has implications for the management of saltmarsh restoration. These results call into question the assumption that saltmarsh erosion in south-east England is due to sea level rise and coastal squeeze, and demands re-examination of the role of management realignment in the regeneration of saltmarshes. Reducing the rates of saltmarsh creek erosion, by exclusion of the infauna, and/or by reducing current velocities in the saltmarsh creeks, would reduce the need to replace eroded marshes by managed realignment, and would reduce future erosion of existing sea walls by wave action.

Introduction

Saltmarshes are important habitats for nature conservation and coastal flood defence. Approximately 80% of the saltmarsh area in the UK has been notified under one or more national or international conservation designations. Saltmarshes are listed in Annex I of the Council Directive 92/43/EEC on the Conservation of Natural Habitats and of Wild Fauna and Flora (usually called the EU Habitats Directive), and the UK government has declared a Habitat Action Plan for saltmarsh under the Convention on Biological Diversity (United Kingdom Biodiversity Group 1999). The estuaries of south-east England (mostly Essex but also Suffolk and North Kent) are of particular international importance, largely because of their use by resident, migratory and overwintering wildfowl and wading birds. The Essex Estuaries European Marine Site is part of the Natura 2000 network and has within it candidate Special Areas for Conservation, Special Protection Areas and Ramsar sites. Saltmarshes are fundamental to the functioning of estuarine and coastal ecosystems, not least as areas of high primary productivity, much of which may be exported as detritus to adjacent habitats, including the mudflats where many birds feed on the invertebrate infauna.

Most of the coast of south-east England is low-lying and protected from tidal flooding by sea walls (Dixon & Weight 1996; King & Lester 1995). Saltmarshes protect the sea walls by attenuating wave and tidal energy (Möller et al. 1999; Christiansen, Wiberg & Milligan 2000) and are of enormous economic value (French et al. 1999). Dixon & Weight (1996) calculated that a sea wall need only be 3 m high, at a cost of £400 per linear metre, if 80 m of fronting saltmarsh is present, as opposed to 12 m high, and a cost of £5000 per metre, if no marsh is present.

After many centuries of development the saltmarshes of south-east England have been eroding for approximately the last 50 years (Burd 1992), and the current rate of loss, estimated at 40 ha year−1 (Coastal Geomorphology Partnership 2000), accounts for approximately two-thirds of the total saltmarsh losses in the UK. This is arguably the most important habitat conservation problem in the UK, not only because of the importance of the marshes and the high rates of loss, but because of the lack of understanding of the causes and appropriate solutions. The general belief is that the cause of this saltmarsh erosion is coastal squeeze. In south-east England the relative sea level is rising by about 1·5 mm year−1 due to isostatic sinking of the land (Shennan 1989). The natural response of marsh vegetation to sea level rise is upward inland migration, but this is prevented by the sea walls and consequently the saltmarsh habitat is lost by being squeezed between the rising sea level and the hard sea defences (Boorman 1992; Burd 1992; English Nature 1992; Turner 1995). However, Hughes (1999, 2001) could find no evidence to support the coastal squeeze hypothesis, and enquiries and searches of the relevant unpublished reports during this study also failed to uncover any evidence. Hughes (1999, 2001) and Hughes et al. (2000) developed alternative hypotheses for recent marsh erosion based upon bioturbation and herbivory by the infauna at the saltmarsh–mudflat interface. Experimental exclusion of the amphipod Corophium volutator (Pallas) led to the growth of microphytobenthos and the pioneer zone plant Salicornia europaea (L.) agg. (Gerdol & Hughes 1993; Smith, Hughes & Cox 1996). (After the first use of the full name of the common species, only the generic name will be used.) Exclusion of the polychaete Nereis diversicolor (Müller), the dominant infaunal invertebrate in these estuaries, led to growth of microphytobenthos, Enteromorpha spp. and Zostera noltii Hornem. (Smith, Hughes & Cox 1996; Hughes 1999; Hughes et al. 2000, respectively). On the basis of these experiments, Hughes et al. (2000) proposed that there are two alternative states at the saltmarsh–mudflat interface. One state is domination by vegetation that excludes burrowing invertebrates, through their tightly meshed root systems which bind and compact the sediment, and the other state is domination by invertebrates that exclude plants through bioturbation and herbivory (including granivory). An increase in abundance of Nereis over the past half-century could explain the accelerating reductions in vegetation over this period (Hughes 1999, 2001).

In addition to loss of vegetation that may be may be due to local factors, such as herbicide pollution from agricultural run-off (Mason et al. 2003), most saltmarsh erosion is at the seaward face and by expansion of internal creeks (Burd 1992). The present study is concerned primarily with creek erosion. Hughes & Paramor (2004) discuss the possible reasons for the erosion of saltmarshes at their seaward face.

The aims of this study were to test the alternative states hypothesis, and that the infauna cause loss of vegetation, by preventing colonization by pioneer zone plants and by increasing the erosion of saltmarsh creeks. The approach was twofold. First, the prediction from the alternative states hypothesis that the distributions of saltmarsh plants and infaunal invertebrates would be distinctly different within a vertical zone of overlap was tested. Secondly, field and laboratory experiments were conducted to assess the effects of excluding the infauna on sediment accretion and colonization by pioneer zone plants in the saltmarsh creeks, and to examine the effects of pioneer zone plants on the colonization of sediment by burrowing infauna.

Methods

field observations

The main study area was the saltmarsh at Tollesbury, Essex, UK (national grid reference TM 963 113; Fig. 1). This marsh, like most of those in south-east England, is a mature estuarine fringing marsh with a gently sloping plateau from approximately 4·4–5·4 m above chart datum (CD). (Hereafter all elevations given are heights above CD.) Mean high water neap tide level is at 3·8 m and mean high water spring tide level is at 5·1 m. The marsh plateau is dominated by Puccinellia maritima (Huds.) and Atriplex portulacoides L. and is dissected by sinuous complex creek systems (as defined by Pye 2000). The creeks are up to 1·5 m deep and often terminate in large basins close to the sea walls (Fig. 1). The observations and experiments were conducted in the inner marsh close to the sea walls, but one experiment was conducted in the adjacent managed realignment (set-back) site, from where some material was also collected for the field and laboratory experiments (Fig. 1).

Figure 1.

Aerial photograph of a section of the marsh at Tollesbury and the adjacent managed realignment site. a, the location of Fig. 3; b, the location of Figs 2 and 11; c, the experimental area in the realignment site. The old sea wall, mentioned in the text, runs from the bottom right up to the top left.

distributions of infauna and plants

Cores of sediment (7·5 cm diameter, 20 cm depth) were collected periodically, and at least every 4 months, from October 1997 to September 2000 to examine the spatial and temporal variations in the distributions of the infauna. On each occasion two cores were collected from each of five separate creeks, one from the bottom (from 3·6 to 4·0 m) and one from the bank (from 4·0 m to 4·4 m). These samples were taken only from the creeks as previous (R. G. Hughes, unpublished data) and concurrent data (see below) had shown that no infauna occurred in vegetated sediment. To describe the vertical distribution of the macrofauna in the marsh in October 1999, five cores of sediment were taken at every 25-cm height interval from the bottoms of the lowest creeks, at 3·4 m, to the marsh surface, at 5·4 m, including some saltpans at 5·15 m.

All sediment cores were taken to the laboratory but proved difficult to sieve because of their high clay content. Instead the cores were carefully broken down by hand to extract the large deep-burrowing macrofauna, the presence of which was often revealed by their burrows. The sorted sediment was placed in flat trays and flooded with seawater, and any remaining invertebrates removed as they moved over, through or from the sediment. The sediment was disturbed periodically and allowed to settle again until no further invertebrates emerged. This technique allowed live Hydrobia ulvae (Pennant) to be distinguished from empty shells.

Internal erosion of the creeks was identified by vertical or near-vertical banks, which often undercut the marsh surface, rotational slumps and mud blocks that had toppled from the vegetated surface (Fig. 2; Pye 2000). In July 1999 the vertical distributions of plant species over five separate examples of banks (cliffs), rotational slumps and older partly eroded slumps were described by placing 0·25-m2 quadrats at 20-cm height intervals from 4·0 to 5·0 m. All the plants in each quadrat were identified and counted. A core of sediment was taken from the centre of each of these quadrats, as described above, but none contained any infauna.

Figure 2.

Photograph of a creek within the Tollesbury marsh (see Fig. 1 for its location) showing three erosion features, fallen blocks (upper centre), undercut bank (top right) and rotational slump (right foreground). Also shown is the location of an experimental mat (arrowed) with accreted sediment (discussed in the text and shown in Fig. 11).

laboratory experiments

Salicornia was the most abundant pioneer zone plant in the marsh and in the upper areas of the realignment site, where it was the only vascular species at the interface with the bare mudflat (Paramor 2002). The dominant infaunal invertebrate in the creeks and saltpans on the marsh surface was Nereis. Nereis has different feeding methods (reviewed by Scaps 2002) but surface deposit feeding seems to be predominant in these habitats. The polychaetes partly emerge from their burrows, evert their probosces and consume sediment, microphytobenthos (Smith, Hughes & Cox 1996), filamentous algae (Hughes 1999) and seeds (Hughes et al. 2000). Burrowing Nereis may also damage the roots of plants (Hughes et al. 2000).

To test the hypothesis that Salicornia deter burrowing by Nereis, five cores (7·5 cm diameter, 15 cm depth) of three types of sediment were collected from the realignment site. The cores were from (i) an area of dense Salicornia (mean density 506 m−2), (ii) an area of sparse Salicornia (mean density 16 m−2) at its lower distribution limit at the boundary with the mudflat, and (iii) an area of this mudflat below the Salicornia zone where Nereis was abundant in the soft marine sediment that had accreted since the area was flooded. The last samples were to act as controls. The sediment colonized by Salicornia (in i and ii) was ex-agricultural soil with only a thin (< 1 cm) covering layer of soft marine sediment. The sediment cores were kept in the plastic corers and covered with 1 cm of seawater. Three Nereis, 3–5 cm in length, were added separately to each core. The behaviour of each worm was recorded, particularly the form of any burrow constructed, as was the time taken to construct each burrow, determined as the time taken for the worm to disappear from view.

To test the hypothesis that Nereis reduces the success of Salicornia germination, defined as the appearance of cotyledons, and seedling survival, three Nereis (3–5 cm in length) were allowed to burrow into each of 15 cores (7·5 cm diameter, 15 cm depth) of marine sediment from the realignment site. Fifteen control cores, from which the worms were removed, were also established. Twenty Salicornia seeds approximately 2 months old were soaked in distilled water for 24 h to stimulate germination and placed on the surface of each core and left for 8 weeks at room temperature. The sediment was kept wet with distilled water to maintain the salinity at 33. The seedlings in each core were counted at the end of the experiment. A similar experiment was conducted to investigate the effects of Nereis on seedling survival. Twenty Salicornia seedlings of about 1 cm height were planted into each of 30 cores, 15 with three Nereis each and 15 without worms. The number of surviving seedlings in each core was counted after 8 weeks.

field experiments

The basis of these experiments was exclusion of surface deposit-feeding Nereis by using mats laid on the sediment surface. This technique was a development of that used by Hughes (1999), who showed that laying small mesh netting on the sediment prevented Nereis from feeding at the surface and facilitated sediment accretion and colonization by filamentous algae. This method achieves the long-term exclusion of Nereis from experimental areas and was preferable to sieving the sediment, or using insecticide, as in the study by de Deckere, Tolhurst & de Brouwer (2001), which could also allow recolonization in long-term experiments. The experiments initially used three types of mat: (i) Greenfix Mulchmat, type W-670 DF with jute backing (Phi Group Ltd, Cheltenham, UK); (ii) Greenfix Eromat, type 7 with jute backing (Phi Group Ltd); and (iii) 28-gauge net curtain, as used by Hughes (1999). The mats were pinned to the surface of the sediment and were rapidly covered by accreted sediment. Thereafter the mats would not have affected processes at the surface of the accreting sediment directly, other than by preventing access by burrowed infauna.

In January 1998, five 1 × 2-m lengths of each mat type were placed in the creeks to extend from the bank down into the creek bottoms. The mats were arranged in five groups; each group contained one mat of each type and was placed in a separate creek. Samples were taken from the mats, and from similar adjacent control areas of sediment identified at the outset, every 3 months for a year. After 6, 9 and 12 months the depth of the sediment over the mats was measured by pushing a rule into the sediment until it met with the resistance of the mat. Any plants on the mats or in the control areas were counted.

To estimate the chlorophyll a content of the sediment and the invertebrate densities, a 5 × 5-cm area of each mat and the sediment above it was removed, immediately placed in the dark and frozen within 5 h of collection until use. After thawing the sediment was separated from the mat material and any invertebrates removed and identified. The remaining sediment was spread thinly inside glass Petri dishes and freeze-dried for 36 h in the dark and transferred to centrifuge tubes where a known quantity of 90% acetone was added. The samples were refrigerated at 5 °C in the dark for 12 h before being centrifuged at 3000 r.p.m. for 7 min. The absorbance of wavelengths at 630, 647 and 664 nm by the supernatant was measured in a spectrophotometer and the chlorophyll a content of the sediment calculated using the appropriate formula for samples dominated by diatoms or green filamentous algae (Jeffrey & Humphrey 1975). Cores of sediment (7·5 cm diameter, 20 cm depth) were collected in the adjacent control areas. The top 2 cm was separated from the rest of the core, any invertebrates removed and the chlorophyll a content calculated. The invertebrates in the remainder of each core were also collected, identified and counted.

The mats in the field experiments were not colonized by Salicornia, contrary to the prediction of the alternative states hypothesis, but it became apparent that other factors could explain the absence of plants, including large distances from a source of seeds, as dispersal of Salicornia was generally limited to less than 1 m from the parent plants (Paramor 2002). Two further experiments were designed to overcome this potential problem. In November 1998 five strips of mulchmat (0·5 × 2 m) were placed, as described above, but close to adult Salicornia plants. Subsequently the sediment over these mats on the sides of the creeks was colonized by Salicornia (see below) but not in the creek bottoms. This absence too could reflect a lack of seeds, as seeds shed by the plants nearby may not reach the bottom of the creeks but may be carried away by the currents. Consequently in October 1999 a third experiment was conducted in which five squares of mulchmat (1·5 × 1·5 m) were pinned to the sediment in the bottom of creeks. Box-cores of sediment (0·25 × 0·25 × 0·15 m deep) containing mature Salicornia plants from the adjacent realignment site were transplanted into holes of the same dimensions cut from the centre of each mat. Five similar box-cores were planted into control areas 2 m from each mat. In August 2000, the numbers of plants of the next generation on each mat and in each control area were counted, and the depth of sediment over the mats measured.

The holes left by the removal of the box-cores of Salicornia from the upper area of the realignment site were filled with sediment scraped from the surface (to contain no Nereis) lower in the realignment site. One year later one core (7·5-cm diameter, 20-cm depth) was taken from each of these 10 areas and examined for invertebrates, as described above.

Results

field observations

Obvious signs of creek erosion, new rotational slumps and fallen blocks of sediment were occasionally seen (Fig. 2). In addition, a series of photographs of a bank in the innermost basin in the marsh taken over a 7-year period (Fig. 3) demonstrated that gradual, and otherwise imperceptible, erosion of the banks occurred too. In this case the base of the bank had been cut back approximately 10 cm in 7 years, undercutting the vegetation above.

Figure 3.

Photographs of the bank of an internal basin (see Fig. 1 for its location) showing the degree of erosion over a 7-year period. The views are approximately 3·5 m wide.

distributions of infauna and plants

Nereis and Hydrobia were the most abundant infauna in the bottom of the creeks (Fig. 4). Small Macoma balthica were present up to 2000 and Nephtys caeca were present only in 2000. Unidentified diptera larvae and nermertean worms were found occasionally in the creek bottoms but were more abundant in the creek walls. Nereis and Hydrobia were rarer in the creek walls than in the bottoms of the creeks, even though the densities of both were often several hundred m−2. No temporal trends in the abundance of the invertebrates in the marsh were apparent.

Figure 4.

Temporal variations in the mean (+ SE) densities (no. m−2) of invertebrates in the creeks in the Tollesbury marsh.

The samples taken to assess the vertical distribution of the fauna contained only Nereis, which was found only in the creeks, at elevations of 3·4 m (mean density 407 m−2 ± 306 SE), and in the saltpans on the upper surface of the marsh, at 5·15 m (mean density 452 m−2± 99 SE). No worms were found in any of the core samples that contained saltmarsh vegetation from the seven intervening elevations, nor at 5·4 m, including those from blocks of sediment slumped into the creeks.

The abundances of the five vascular plant species found in the vertical transects taken over the three erosion features in the creeks are shown in Fig. 5. Atriplex, Puccinellia and Aster tripolium L. dominated the marsh surface at the tops of the creeks above 4·4 m. Salicornia, Suaeda maritima (L.) and Atriplex were present on some of the creek banks (cliffs), at and below 4·4 m, but at these lower elevations Salicornia and Suaeda were more abundant on slumped blocks and eroded slumped blocks, particularly Salicornia which occurred at mean densities of more than 100 m−2.

Figure 5.

The variations in mean (+ SE) densities of vascular plants with elevation on the (a) walls (cliffs), (b) slumped sediment and (c) eroded slumped sediment within the creeks of the Tollesbury marsh.

laboratory experiments

Nereis formed significantly fewer burrows in sediment containing Salicornia at naturally high densities than in either the controls (t = 2·68, d.f. = 8, P= 0·04) or the boundary sediment with Salicornia at low densities (Fig. 6). These data are for ‘true’ vertical burrows. Some worms that did not burrow, particularly those in the sediment containing Salicornia at high densities, constructed horizontal tunnels in the surface sediment by collecting sediment particles on the mucus adhering to their bodies until camouflaged by a tube of sediment.

Figure 6.

The proportions of Nereis diversicolor that burrowed vertically into the three different sediment types.

Nereis had a significant negative effect on Salicornia seedling production from seeds (t = 4·34, d.f. = 28, P= 0·0002) and on seedling survival (t = 11·63, d.f. = 28, P ≤ 0·0001) (Fig. 7). Some worms were observed to grasp seedlings and pull them into their burrows.

Figure 7.

The effect of Nereis on the (a) production and (b) survival of Salicornia europaea agg. seedlings.

field experiments

In the first exclusion experiment the data from the three different types of mats were not different and they have been combined. By the end of the experiment, after about a year, a mean depth of approximately 2 cm of sediment covered the mats, following consistent accretion over the previous 6 months (Fig. 8). The mats remained at the same level as the surrounding sediment, indicating that no accretion had occurred there. The chlorophyll a content of the sediment above the mats was always greater than in the control areas (Fig. 9) but on only two occasions, on the creek walls, was this difference statistically significant. No vascular plants colonized the mats at the bottoms of the creeks, and only a small number of Salicornia was present on some mats on the creek walls.

Figure 8.

The mean (± SE) depths (cm) of sediment accreted over mats laid on the sediment on (a) creek walls and (b) in creek bottoms.

Figure 9.

The mean (+ SE) chlorophyll a content of surface sediment from control areas and sediment deposited over mats laid on the sediment on (a) creek walls and (b) in creek bottoms. *P ≤ 0·02.

The abundances of the four invertebrate taxa found in the first exclusion experiment are shown in Fig. 10. Overall the most abundant species were Nereis and Hydrobia, and these were more abundant in the bottom sediments than in the walls, as was Macoma in October and February. In contrast, diptera larvae were more abundant in the walls than in the creek bottoms. With the exception of the April samples (after only 3 months) on almost all other occasions Nereis, Hydrobia and Macoma were significantly more abundant in the control sediments than over the mats. The sediment over the mats in the creek bottoms was colonized by Nereis but these were all small worms, shorter than 2 cm, in contrast to the worms in the control samples, which were up to 10 cm in length. There was no difference in the abundance of diptera larvae in the control and mat sediments on the creek walls, but in October and February they were found in sediment over the mats in the creek bottoms but not in the control sediment.

Figure 10.

The mean densities (± SE) of invertebrates in the sediment accreted above the experimental mats and in control areas (a) on creek walls and (b) in creek bottoms.

In the second exclusion experiment, after 5 months a mean depth of 1·1 cm (± 0·6 SE) had deposited over the nets and mulchmats on the walls of the creeks (2 mm month−1) and 1·5 cm (± 0·23 SE) in the creek bottoms (3 mm month−1). Again the mats were at the same level as the surrounding control sediment, indicating that no accretion had occurred there. Figure 11 shows the sediment accreted after 8 months on one of these mats placed across a narrow creek where the tidal currents were relatively rapid. On the walls of the creeks the mean densities of Salicornia were 164 m−2 (± 149 SE) on the mulchmats, 2 m−2 (± 1 SE) on the nets, and 12 m−2 (± 8 SE) in the control areas. In the bottom of the creeks only a single Salicornia plant was found in the sediment above the mats, while none was found in the control areas. In this experiment, which was conducted over the winter and early spring, there was no significant difference in the abundance of Nereis over the mats (3882 m−2 ± 2000 SE) and in the control areas (1584 m−2± 559 SE). However, as in the first experiment, the Nereis worms in the control areas were up to 10 cm long but only worms less than two 2 cm long were above the mats. Hydrobia was abundant only in the sediment in the creek bottoms (2128 m−2 ± 1795 SE) but had low densities over the mats there (108 m−2 ± 61 SE).

Figure 11.

Photograph of the sediment accreted over a 2 × 0·5-m strip of mulchmat (arrows) (see Fig. 1).

Five years later the site was revisited (but no data were collected) and Fig. 12 shows the long-term effect of one pair of mats placed close to the blind end of a creek where the water movements were very slow. The position of the net could be identified by the approximately 2 cm of sediment accreted on it. The upper part of the net had no Salicornia and only a thin layer of sediment, because deeper layers that accrete eventually slide off. In contrast, the effect of the mulchmat, to the right of the net, was clear. It had accreted 7–10 cm of sediment along its length and was the only location close by where Salicornia was present.

Figure 12.

Photograph of a net (arrowed) and adjacent mulchmat (centre) showing sediment accretion and Salicornia colonization of the mulchmat after approximately 5 years. (To the right is a newly excavated burrow of the shore crab Carcinus maenas.)

In the third exclusion experiment the mulchmats with blocks of mature Salicornia at their centre accumulated a mean depth of sediment of 2·7 cm (± 0·06 SE) in 7 months (3·8 mm month−1), while no appreciable accretion occurred in the control areas. The sediment above these mats was colonized by seedlings of Salicornia, at a mean density of 8 m−2, while no plants were found in the control areas.

The mean density of Nereis in the sediment placed in the holes created by the removal of the Salicornia cores in the realignment site was 1175 m−2 (± 400 SE).

Discussion

the alternative states hypothesis

In the laboratory experiments Nereis significantly decreased the production of Salicornia seedlings from seeds and the survival of seedlings. Olivier et al. (1995) recorded similar observations, and the deliberate selection of plant material by Nereis that was pulled into the burrow, where it was consumed over a long period of time. Bioturbation was also a probable cause of seed and seedling mortality as the feeding movements of Nereis across the sediment buried some seeds and disturbed remaining seedlings. Nereis burrow formation was significantly reduced by the presence of Salicornia at naturally high densities, while at lower densities of Salicornia in similar soil (from the boundary area) Nereis burrowing behaviour was not different from that in the control sediment that had naturally contained worms. Thus the laboratory experiments indicate that the mechanisms responsible for the mutually exclusive distributions of plants and Nereis observed in the field are consumption and disturbance of seeds and seedlings by Nereis, and deterrence of burrowing of Nereis by plants (Salicornia in this case) whose roots prevent burrowing, either directly or by consolidating and compacting the soil.

Raffaelli, Karakassis & Galloway (1991) recorded Nereis to have the highest vertical distribution of a range of infaunal invertebrates in the Ythan estuary (Scotland). At Tollesbury the presence of Nereis in dry saltpans close to mean high water spring tide level indicates that the worms can tolerate long periods of emersion. Feeding by emersed Nereis was observed during this study, and by Esselink & Zwarts (1989), and it may be this behaviour that allows Nereis to colonize higher sediments than other infaunal invertebrates whose feeding activities are limited to periods of immersion. Because Nereis has a high vertical distribution it overlaps that of the saltmarsh plants, but within this zone of overlap their distributions are exclusive. Nereis was found only in the unvegetated sediment in the saltpans of the upper marsh surface and in the creeks. In the creeks, plants were found only on sediment that had slumped from the marsh surface, which contained no infauna. Therefore, as the creeks are within the vertical potential niche of the pioneer zone plant species, and the vegetated areas of the marsh within the potential niche of Nereis, mutual exclusion, as predicted by the alternative states hypothesis, explains their separate distributions. This conclusion is supported by the results of the field experiments in the realignment site, where Nereis colonized the upper areas only when Salicornia was removed.

Development of vegetation on some of the Nereis exclusion mats (Salicornia) also supports this conclusion, but colonization by vascular plants of these mats was not consistently successful, particularly in the bottoms of the creeks. Seemingly the sediment on the mats was suitable for seedlings, apart from the reduction in the effects of Nereis. It was at an appropriate elevation and, being slightly higher and drier than the surrounding control sediment, with higher densities of microphytobenthos, it would have been more stable and conducive to seedling establishment. (Wetness alone may not be important because in the laboratory Salicornia has been grown from completely immersed seeds; R. G. Hughes, personal observation.) Moreover, the drier sediment over the mats had relatively low densities of Hydrobia and the seeds and seedlings here may have been less disturbed by the gastropods than in the wetter control areas. Observations of the distribution and behaviour of Hydrobia showed it to be most abundant and more active on sediment that remained wet throughout the period of emersion (Paramor 2002). Drier sediment may be less suitable for Hydrobia as it may not be able to feed throughout the low tide period (Paramor 2002). There are several possible reasons for the lack of success in establishing seedlings on the experimental mats. Primarily the success of the mats in promoting sediment accretion, particularly in the creek bottoms, would have decreased the success of colonization by Salicornia, whose seeds cannot germinate successfully when buried by even shallow depths of sediment (Paramor 2002). The accumulated sediment also allowed colonization by small Nereis, whose activities could have reduced the differences between the treatments and controls. Plant colonization also depends on a source of seeds and some of the mats, particularly in the first experiment, were probably too far from parent plants. In these creeks Salicornia seedlings were generally limited to within about 1 m of plants of the previous generation.

erosion of saltmarsh creeks

Before considering any effects of the infauna on erosion it was necessary to establish that the creeks at Tollesbury were actually eroding, because Adam (1990) and Gabet (1998) considered that saltmarsh creeks have been remarkably stable over decades, despite the appearance of instability. The erosion of the bank shown in Fig. 3 indicates that this is not true at Tollesbury and, together with the periodic appearance of new rotational slumps and fallen blocks, confirms the conclusion of Burd (1992), based on a comparison of aerial photographs taken 15 years apart, that much of the loss of marsh vegetation is by creek erosion. Even in the innermost basin at the head of the creek system the bank has been cut back approximately 10 cm in 7 years, a rate that if maintained would mean that all the vegetation between two creeks 3 m apart would be lost in 100 years. At Tollesbury this would mean almost all of the vegetation, and the marsh would be expected to erode to leave residual mud mounds, like those that characterize the internal areas of many marshes in south-east England.

Notwithstanding the incomplete success of the exclusion mats to stimulate development of pioneer zone vegetation, one conclusion to be drawn from this study is that Nereis reduces the abundance of Salicornia in the saltmarsh creeks, especially on the walls, and this may prevent the subsequent successional development of marsh vegetation. This is one mechanism by which the polychaetes destabilize the creeks, as the plants would otherwise stabilize the sediments with their roots and might enhance sediment accretion by slowing current speeds. Nereis also exacerbates sediment erosion within the creeks more directly. In all the experiments the exclusion of large worms led to significant sediment accretion, by reducing bioturbation and enhancing sediment stability through the increase in abundance of microphytobenthos (measured as chlorophyll a) and filamentous algae. These results complement those of de Deckere, Tolhurst & de Brouwer (2001), who removed Nereis and oligochaetes, by spraying insecticide on the mud, and recorded a threefold increase in sediment stability after only 4 days. The results presented here support the conclusion of Widdows & Brinsley (2002) that the biota can affect erosion/deposition of sediments by several orders of magnitude. They recorded a 100-fold difference in the erodability of cohesive sediments under different biological conditions, increased stability in the presence of microphytobenthos and decreased stability in the presence of the bivalve Macoma. Hydrobia too is known to destabilize sediments, by bioturbation and grazing on the microflora (Andersen et al. 2002), but the effects of these gastropods on erosion of the creeks were not distinguished in these experiments.

Bioturbation and herbivory in saltmarsh creeks may occur more widely, as the presence of extensive creek systems, like those at Tollesbury, are typical of many marshes, for example in south-east England (R. G. Hughes, personal observation; Crooks & Pye 2000) and in the USA (Gabet 1998; Hartig et al. 2002). However, in neither of the last two studies were the effects of the infauna in destabilizing sediment considered as a cause of creek erosion. Settlemyre & Gardner (1977) considered that the reworking of the sediment by fiddler crabs (Uca) in the summer was responsible for export of sediment from creeks in marshes in south-east USA. At Tollesbury the shore crab Carcinus maenas (L.) burrows extensively in the creek banks (Fig. 12) but the effect of the burrows in reducing marsh stability, and the behaviour of the crabs in reducing plant abundance and increasing sediment erosion, are unknown.

There are explanations for recent increases in creek erosion based on only physical parameters. Allen (1997, 2000) developed a conceptual model that predicted that sea level rise could lead to creek erosion. This is because as more water would flood and ebb across the marsh surface, and hence through the creeks, the ‘hydraulic duty’ of the creeks would increase, and consequently so would their cross-sectional area. However, this model may not be applicable to most saltmarshes because their elevation tends to increase at the same rate as sea level rise (for a discussion see Hughes & Paramor 2004) and the tidal volume over the marsh surface would not change significantly. Pye (2000) considered that saltmarsh creek erosion in south-east England might be due to higher tidal current velocities caused by a recent increase in the tidal range within the southern North Sea. However, the promotion of sediment deposition in these experiments, including in creeks where the tidal currents are relatively rapid (e.g. Fig. 11), indicates that explanations based on physical processes alone are untenable. The conclusion here, that loss of saltmarsh vegetation through creek erosion cannot be attributed to physical processes in isolation, is one argument against saltmarsh loss in south-east England being a consequence of sea level rise and coastal squeeze. Hughes & Paramor (2004) develop this point, with others, and conclude that coastal squeeze is not responsible for saltmarsh loss in south-east England.

In mature saltmarshes that have developed over several centuries the creeks should have achieved an equilibrium state largely determined by the tidal prism and the area of marsh that they drain (Williams, Orr & Garrity 2002). The creeks at Tollesbury, like those throughout south-east England, are eroding (Burd 1992) and are not in equilibrium despite the marsh being several centuries old. Land claim through embanking of the upper marsh, as has happened throughout south-east England and elsewhere, reduces the upstream volume of creeks (Pye 1992). As the tidal current velocity at any given point in a creek system is proportional to the rate of change in upstream volume (including over-marsh flooding) divided by the cross-sectional area at that point, embankment should lead to a reduction in current speeds and progress towards a new equilibrium state by sediment accretion in the remaining creek system. The old sea wall at Tollesbury (Fig. 1) is at least 150 years old and the present creek erosion in front of it was not expected. The field experiments indicate that the creeks accrete sediment as predicted only without the effects of Nereis. This provides further support for the conclusion of Widdows & Brinsley (2002) that the behaviour of marine cohesive sediments cannot be predicted by models that consider only the physical parameters of the environment.

There is some evidence that Nereis has increased in abundance over the past few decades (Hughes 1999), and the accelerated creek erosion identified here and linked to its presence may also be similarly recent in extent. Export of sediment from the creeks may entrain a positive feedback as the increase in floodable volume leads to increased current speeds, especially at the seaward end of long creek systems, and hence to further erosion. The equilibrium morphology of an unconstrained tidal creek is approximately triangular in plan and widest at the mouth, where the ratio of the cross-sectional area to the up-tide floodable volume is approximately constant throughout its length. The complex creeks at Tollesbury, and on many other marshes in south-east England and elsewhere, do not have this shape, having approximately parallel sides for much of their length. It may be expected that further erosion of these creeks will occur towards the equilibrium morphology by lateral expansion, particularly close to their mouths. Some marshes have already undergone such internal disintegration, including, for example, that at Maldon at the head of the Blackwater Estuary (Essex) (Fig. 13). The position of the outer edge of this marsh remains remarkably similar to that mapped by the Ordnance Survey in 1881 but the significant loss of vegetation has been from expansion of two internal creeks, particularly near their mouth (Burd 1992).

Figure 13.

Aerial photograph of the saltmarsh at Maldon, at the head of the Blackwater Estuary (Essex), showing that much of the interior of the marsh is bare mud due to the development of two main creeks.

implications for saltmarsh conservation

The results presented here support the alternative states hypothesis, and in confirming bioturbation and herbivory by Nereis as a key factor in the loss of pioneer zone vegetation and saltmarsh creek erosion, have identified a major cause of an important economic and habitat conservation problem. These conclusions have implications for the management of saltmarsh restoration. The saltmarsh habitat action plan (United Kingdom Biodiversity Group 1999) commits the UK to recreating and maintaining the total saltmarsh area to that of 1992. Hughes & Paramor (2004) concluded that the current policy of relying on managed realignment to increase saltmarsh area is not satisfactory, not least because the areas available are too small. If creek expansion does continue, as predicted above, then managed reductions in creek erosion would reduce the need to create new marshes in future realignment sites and would at least maintain the protection from wave action offered to existing sea walls.

This research has identified one potential means of reducing creek erosion. The ridge of sediment that accreted over the mat across a narrow creek (Fig. 11) caused the retention of a shallow but long pool of water behind it. This pool reduced the tidal volume of the creek and filled with sediment until the upstream floor of the creek had been raised to the same level as the ridge over the mat (Fig. 2). The consequences of this reduction in creek volume on flow velocities and sediment accretion elsewhere in the creek system are unknown but this observation indicates that quite small interventions may halt creek erosion and stimulate sediment accretion. For example, shallow partial barriers (e.g. of sandbags or sediment dredged from elsewhere) that would retain some water at low tide and reduce the tidal volume of the creeks might create a positive feedback of deposition, where further reductions in creek volume would further reduce current speeds. Further research is required on the interactions of creek morphology current speeds and sediment erosion/deposition.

Acknowledgements

We wish to record our gratitude to MAFF (now Defra) for funding this research, to several friends and colleagues who helped with the fieldwork, and to two referees for their helpful comments.

Ancillary