M. Wolters, Community and Conservation Ecology Group, University of Groningen, PO Box 14, 9750 AA Haren, the Netherlands (e-mail email@example.com).
1Deliberate breaching of sea defences is frequently practised with the aim of restoring salt-marsh vegetation on previously embanked land. However, experience so far has shown that it may take several years before salt-marsh vegetation is fully established, and it is possible that limited diaspore dispersal plays a role in this. In order to ascertain whether salt-marsh development may be constrained by limited diaspore dispersal, we studied the dispersal of salt-marsh species by tidal water.
2From October 2001 to the end of March 2002 a total of 38 species, of which 18 were salt-marsh species, was trapped in a restoration site and adjacent marsh. Aster tripolium, Limonium vulgare, Puccinellia maritima, Salicornia spp., Spergularia media and Suaeda maritima were the most abundant salt-marsh species, with more than 3 diaspores m−2 trapped during the study period.
3For most species, the number of diaspores trapped was representative of their abundance in nearby vegetation. Hence, despite the potential for long-distance transport by tidal water, our results indicate a predominantly local dispersal of salt-marsh species.
4Synthesis and applications. For the restoration of salt-marsh vegetation after de-embankment, relatively rapid colonization may be expected from pioneer and low-marsh species, provided they are present in a nearby source area and the restoration site is at the appropriate altitude. The establishment of species absent from the adjacent marsh may be dependent on the presence of birds or humans as the main dispersal agents. Breaching of sea defences should preferably take place before or during September, in order to take advantage of the peak in dispersal of salt-marsh species in the first year after breaching.
In managed realignment, regular tidal inundation is assumed to function as the key agent for dispersing diaspores. Laboratory studies have shown that seeds and fruits of various salt-marsh species can float in seawater for time spans varying from several hours up to several months (Koutstaal, Markusse & De Munck 1987). Most seeds retain their viability during submergence in seawater (Koutstaal, Markusse & De Munck 1987) and rapidly germinate upon transference to fresh water conditions (Woodell 1985). Both characteristics are likely to facilitate dispersal by tidal water.
Studies of the composition of driftline material deposited after storms and high tides show that although numerous propagules and vegetative parts occur in driftline material, the composition mainly resembles the local vegetation, suggesting limited dispersal distances (Bakker, Dijkstra & Russchen 1985; Persicke, Gerlach & Heiber 1999; Wolters & Bakker 2002). On the other hand, Huiskes et al. (1995), using standing and floating nets to trap seeds at incoming and outgoing tides, showed a net export of floating propagules at intertidal flats compared with a net import at salt marsh, indicating a potential exchange between different salt marshes. However, it is not clear whether the seeds are mainly floating over the salt marsh with the tide and becoming assembled in the driftline or actually settling at the salt-marsh surface.
In this study we assessed the deposition of diaspores in a salt-marsh restoration site by using seed traps. Our main objective was to determine whether salt-marsh development is constrained by limited diaspore dispersal. The specific aims of this study were to (i) examine temporal patterns in the dispersal of salt-marsh species in order to provide information on the appropriate timing of breaches during managed realignment for dispersal in the first year; (ii) compare the composition and abundance of diaspores trapped at different locations in the restoration site and adjacent marsh in order to identify possible barriers to dispersal, such as altitude and distance to breach; (iii) relate the composition and abundance of the dispersed diaspores to the established vegetation of the restoration site and adjacent marsh in order to identify the main source of diaspores; and (iv) discuss the relative importance of diaspore dispersal for salt-marsh development after managed realignment.
Materials and methods
This study was performed at the Tollesbury managed realignment site and adjacent marsh in the Blackwater Estuary, south-east England (51°46′N, 0°51′E) (see Figure S1 in the Appendix). The 21-ha restoration site had originally been a salt marsh, but was reclaimed in the late 18th century, after which it was used for agriculture (Boorman et al. 1997). The old seawall surrounding the site was breached at one place in August 1995, leaving a 50-m wide opening that connects the site to Tollesbury Creek. Other construction work has involved the excavation of a channel to connect an existing drainage ditch to the breach and the construction of a new seawall landward of the old embankment to prevent flooding of the neighbouring arable fields. The altitude of the site ranges from 0·9 m to 3·0 m above Ordnance Datum (OD), with the major part of the site lying below 2·0 m OD (Garbutt et al. 2002). Mean high water neap (MHWN) and mean high water spring (MHWS) tide levels are 1·50 m and 2·60 m OD, respectively (Pye & French 1993).
established vegetation of restoration site and adjacent marsh
Approximately 30% of the restoration site was vegetated in 2001, with the lower limit of the vegetation following the 1·5-m OD contour line (i.e. MHWN tide level). The dominant species is Salicornia spp. (L.) followed by Suaeda maritima (L.) and Puccinellia maritima (Huds.). Ten other salt-marsh species have been recorded in the site, but with very low abundance (Garbutt et al. 2002).
Adjacent to the restoration site are the mature Old Hall and Tollesbury marshes (see Figure S1 in the Appendix), which are characterized by numerous creeks that dissect the salt-marsh surface. The vegetation of the creek banks (average elevation 0·3 m + MHWN) is dominated by Aster tripolium (L.), Salicornia spp. and Suaeda maritima, whereas the salt-marsh surface (average elevation 1·06 m + MHWN) is dominated by Atriplex portulacoides (L.), Puccinellia and Limonium vulgare (Mill.) (Garbutt et al. 2002). Data on species abundance in the established vegetation of the restoration site and adjacent marsh in 2001 are derived from Garbutt et al. (2002). In their study, the presence of a species in the restoration site was recorded in three transects of 125-m long and 20-m wide, divided into 2500 cells of 1 m2. Species abundance at the adjacent marsh was recorded in 60 quadrats (43 salt-marsh surface, 17 creek bank) of 1 m2 divided into 100 subplots.
Four locations (R1–R4), representing different stages in the development of salt-marsh vegetation, were selected in the restoration site, and diaspore dispersal was assessed by placing five 45 × 45-cm Astroturf® mats, 1·5 m apart, at each of these four locations (see Figure S1 in the Appendix). R1 and R3 were approximately 400 m from the breach at an elevation of 0·95 m and 0·93 m + MHWN. The established vegetation of the two sites was similar, with Salicornia as the dominant species, frequent occurrences of Suaeda and some patches of Puccinellia, but the two sites were located at different angles from the breach and might therefore have experienced different rates of diaspore dispersal. R2 was 15 m seaward of R1 at 0·87 m + MHWN and the established vegetation consisted almost entirely of Salicornia at high densities. R4 was located 160 m south-west of the breach at an elevation of −0·03 m + MHWN and here Salicornia was the only species present in the established vegetation at low densities. In order to compare dispersal in the restoration site with a reference site, four sets of five Astroturf mats were placed on the adjacent marsh (M1–M4; see Figure S1 in the Appendix). M1 and M2 were located at the interior marsh, whereas M3 and M4 were close to the creek entering the restoration site. M2 and M4 were placed on the creek bank and M1 and M3 on the marsh surface, to represent the two main plant communities of the adjacent marsh.
The Astroturf mats were collected and replaced every month, starting 6 September 2001, with the final collection on 31 March 2002. In the first month (6 September–12 October) only four out of the eight locations (i.e. R1, R2, M1 and M2) were sampled and the results of this period have not been taken into account for statistical analysis. The material trapped during the sampling period was carefully rinsed from the mats onto a 0·2-mm mesh sieve, spread onto trays filled with sterilized soil and transferred to the greenhouse after a 4–6-week cold stratification period, according to the method described by Wolters et al. (2004). For every 10 trays, a control tray without a sample was placed to check for seeds being blown into the greenhouse. Emerging seedlings and rooted vegetative parts were identified and removed as soon as possible, or transferred to separate flower pots in case flowering was needed for identification.
The first question we addressed was whether there were differences in number of diaspores between the four locations within the restoration site and within the adjacent marsh. Prior to statistical testing, screening of the data revealed a lack of normality and homogeneity of variances for the different groups, which only partly improved after different data transformations. Therefore, a non-parametric Kruskal–Wallis test (SPSS 11·5) was used to test for differences in the ranking of the mean number of diaspores trapped between different locations. The test was run separately for each salt-marsh species, with a cumulative mean number of at least 3 diaspores m−2 at any location over the period 12 October–1 March. The Nemenyi multiple comparison test (i.e. the non-parametric analogue of the Tukey test; Zar 1996) was used to determine which locations significantly differed from each other. A Sørensen similarity index (Jongman, ter Braak & van Tongeren 1995) was calculated to compare similarity in species composition between the established vegetation and diaspores trapped in the restoration site as well as on the adjacent marsh surface and creek banks.
total number of diaspores trapped
From 6 September 2001 to 31 March 2002 a total of 38 species was trapped at the Astroturf mats, of which 18 were salt-marsh species (see the Appendix). Many of these species were present in very low numbers at any location and, in fact, only six salt-marsh species were found with more than 3 diaspores m−1 location−1. These species included Aster, Limonium, Puccinellia, Salicornia, Spergularia media (L.) and Suaeda.
temporal patterns in diaspore dispersal
Temporal differences in number of trapped diaspores of all salt-marsh species indicated that the main dispersal period took place between October and December, both at the restoration site and the adjacent marsh (Fig. 1), but this pattern was mainly because of the abundance of Salicornia, which accounted for more than 95% of the diaspores trapped (Fig. 2). Some differences between the species could be observed, with Limonium, Puccinellia and Spergularia starting to be dispersed earlier than Aster, Salicornia and Suaeda (Fig. 2). Between September and October and between early February and late March hardly any diaspores were trapped.
spatial patterns in diaspore dispersal
Over the period from 12 October to 31 March, only 5 Salicornia seeds m−2 and no other species were trapped in the restoration site at location R4 (i.e. closest to the breach at an elevation of −0·03 m + MHWN). No significant differences were detected between the other three locations within the restoration site for the six salt-marsh species, and it was decided to pool the locations (R1, R2 and R3) to facilitate comparison with the adjacent marsh. On the adjacent marsh, significant differences between locations were found for Aster and Salicornia (P < 0·05), on the basis of which two groups were identified. The first group comprised the marsh surface locations (M1 and M3) and the second group consisted of the creek bank locations (M2 and M4). A non-parametric t-test (Mann–Whitney, SPSS 11·5) was applied to test for differences in number of diaspores between restoration site and marsh surface or creek bank for each of the six salt-marsh species (Fig. 3). Diaspores of Aster were trapped in significantly (P < 0·05) higher numbers on the adjacent marsh creek banks compared with the restoration site. No differences were observed between the restoration site and the adjacent marsh surface. Limonium, Puccinellia, Spergularia and Suaeda were either absent or trapped in significantly lower numbers in the restoration site compared with the adjacent marsh surface and/or creek banks (Fig. 3). Salicornia was the only species with significantly more seeds trapped in the restoration site compared with the adjacent marsh surface and creek banks. This species was by far the most abundant, with almost 4000 seeds m−2 trapped in the restoration site and 1900 seeds m−2 in the adjacent marsh creek banks (Fig. 3).
similarity between composition of vegetation and trapped diaspores
Out of the 10 salt-marsh species trapped in the restoration site, six were recorded in the established vegetation of this site (Fig. 4), resulting in a similarity index of 0·52 (Table 1). Higher similarity in species composition (0·70) was found between diaspores and vegetation of the adjacent marsh surface, which had eight species in common. A much lower similarity (0·42 and 0·43) was calculated for the diaspores in the restoration site compared with the vegetation of the adjacent marsh surface and creek banks (Table 1). In particular, Atriplex and Triglochin maritima (L.) showed a weak relationship between their abundance in the vegetation and the number of diaspores trapped (Fig. 4). Interestingly, diaspores of Agrostis stolonifera (L.), Leontodon autumnalis (L.) and Sagina maritima (L.) were trapped in the restoration site whereas they were not recorded in the established vegetation of this site or the adjacent marsh. Plantago maritima (L.), which was also trapped in the restoration site but absent from the established vegetation, may have come from the adjacent marsh, where it was recorded in 2·1% of the plots.
Table 1. Sørensen similarity indices comparing established vegetation and diaspores trapped on Astroturf® in the restoration site (R) and adjacent marsh surface (Ms) and creek bank (Mc)
total number of diaspores trapped
In our study a total of 18 salt-marsh species was trapped on astro turf mats, although only six with more than 3 diaspores m−2. It could be argued that certain species might be more efficiently trapped than others because of morphological differences between their seeds. However, in an experiment specifically conducted to test the seed-retaining efficiency of the Astroturf mats, no differences were found for three morphologically very different species (Wolters et al. 2004). In comparison, however, Rand (2000) trapped a total of 346 diaspores m−2 from six species over a 3-month period using styrofoam plates covered with a resinous material. This was 60 times less than the number of diaspores trapped on the Astroturf mats in our study over the same period, although this was mainly because of the abundance of Salicornia. It should be noted that the number of diaspores retained by the Astroturf mats in the present study was the net result of deposition and removal of diaspores by tidal water over the monthly replacement series. The results therefore represent a minimum rather than total number of diaspores dispersed during the sampling period. Dispersal by wind and birds may also have contributed to the number of diaspores trapped in our study. Huiskes et al. (1995) observed that a considerable number of propagules transported along the bottom of the water column with the flood tide returned to the same station with the ebb tide. However, the same was not found for propagules floating on the surface of the water column, suggesting different dispersal modes between species with different floating characteristics. Interestingly, four of the six most abundant species trapped on the Astroturf mats turned out to be the same four species predominantly caught in standing nets by Huiskes et al. (1995), namely Aster, Puccinellia, Salicornia and Spergularia. These species are characterized by relatively short floatation times of a few hours to a few days (Koutstaal, Markusse & De Munck 1987).
temporal patterns in diaspore dispersal
One of the aims of this study was to examine temporal dispersal patterns in order to provide information on the appropriate timing of breaches for managed realignment. Our results show a peak in diaspore dispersal between September and December for Limonium, Spergularia and Puccinellia, and between October and December for Aster, Salicornia and Suaeda. These results agree with the study by Hutchings & Russell (1989), who observed that the latter three species as well as Limonium produced most of their seeds in September and October. However, Puccinellia was found to produce a majority of ripe seeds in July (Hutchings & Russell 1989) and to grow vegetative tillers at a faster rate from August to October (Gray & Scott 1977), suggesting that our study may have started too late to capture the dispersal peak for this species. Other species that may have been absent from our traps because of their dispersal peak occurring before the start of our study include Armeria maritima (Mill.), Festuca rubra (L.), Juncus gerardi (Lois.) and Plantago, which set seeds in July and August (Hutchings & Russell 1989). Except for Plantago, these species were absent from the established vegetation of the restoration site and either absent or present in low frequencies on the adjacent marsh (Fig. 4), hence a limited local source of diaspores could also be an important reason for the absence of these species from our seed traps. Nevertheless, in order to take advantage of the peak in dispersal of salt-marsh species in the first year after breaching, it is recommended that breaching of sea defences takes place in early September at the latest.
spatial patterns in diaspore dispersal
An interesting result obtained from this study is the exceptionally low number of diaspores captured on seed traps located in the restoration site at low elevation (−0·03 m + MHWN) and relatively close to the breach (R4). It was expected that seeds would arrive at this location but fail to establish because of abiotic constraints related to the low elevation (Wiehe 1935). However, except for 5 Salicornia seeds m−2, no diaspores were trapped during the entire study period. This could be because of the scarcity of a nearby source of diaspores (there were only a few established Salicornia plants surrounding the seed traps) in combination with a predominantly short-distance dispersal of salt-marsh species as implied by the other results of this study. Another possibility is that diaspores may have been dispersed to this location but failed to settle on the seed traps because of strong tidal currents. Support for the latter possibility comes from a study where seeds of three salt-marsh species were added to Astroturf mats placed at three different elevations. After one tidal inundation between 50% and 80% of the seeds had disappeared from the mats at the lowest elevation, compared with 20–50% of the seeds from the mats at the higher elevations (Wolters et al. 2004). In a seed dispersal experiment conducted in a flume channel designed with different fluvial features, Merritt & Wohl (2002) also found a smaller number of seeds deposited in areas of high velocity compared with areas of slow flow velocity. These results suggest that low elevation associated with high tidal flow velocities may not only constrain establishment but also arrival of salt-marsh plants.
relationship between trapped diaspores and established vegetation
For most species, the number of diaspores trapped was representative of their abundance in the vegetation. Exceptions were Atriplex and Triglochin, of which less than 3 diaspores m−2 were trapped during the entire sampling period, although they occurred with a frequency of 50% and 18%, respectively, in the established vegetation of the adjacent upper marsh. The fruits of these two species have a floatation time exceeding 4 months (Koutstaal, Markusse & De Munck 1987), implying a high potential for long-distance dispersal. Chapman (1950) and Davy (1991) also report a large potential for dispersal by tidal water for Atriplex and Triglochin, respectively, and both species were trapped in relatively high numbers in the study by Huiskes et al. (1995). One explanation for the absence of diaspores of these species from astro turf mats in our study may be herbivory of the plants. Atriplex for example, forms a food supply for shorelarks Eremophila alpestris (Dierschke 2002) and hares Lepus europaeus (van der Wal et al. 2000) and both Atriplex and Triglochin are eaten by snow bunting Plectrophenax nivalis (Dierschke 2002) and brent geese Branta bernicla bernicla (Summers et al. 1993; van der Wal et al. 2000). Of the latter species, between 400 and 900 individuals year−2 have been reported feeding at the Tollesbury site between 1996 and 1998 (Atkinson et al. 2001).
Another interesting observation is the trapping of Agrostis stolonifera, Leontodon autumnalis and Sagina maritima in the restoration site when none of these species was present in the established vegetation of this site or the adjacent marsh. These species may have been transported over longer distances by tidal water, or they may have been blown onto the Astroturf mats from nearby seawalls, where they were observed to grow. All three species were classified as having a high potential for long-distance dispersal (i.e. > 100 m) in the IRIS database (Tamis et al. 2004) and they were described by Westhoff (1947) as anemochorous with potential dispersal distances of more than 1 km.
The much higher numbers of diaspores of Salicornia trapped in the restoration site compared with the adjacent marsh surface, while the percentage frequency of this species in the vegetation of the two sites was similar, could be explained by the fact that Salicornia occurred in much higher densities (i.e. higher percentage cover but not frequency) in the restoration site compared with the adjacent marsh surface. The absence of Puccinellia from seed traps in the restoration site is likely to be related to its low abundance in the vegetation of this site. This species was also absent from the creek bank vegetation, although it was trapped on Astroturfs, but this could be explained by the fact that diaspores from the marsh surface may have landed on the Astroturf mats at the creek banks (which were directly below the marsh surface) but failed to establish because of abiotic constraints related to the low elevation (i.e. being 80 cm lower than the marsh surface).
Overall, our results indicate a predominantly local dispersal of salt-marsh species, despite the potential for long-distance transport and wide distribution by tidal water.
implications for future managed realignment
Our results have shown that the peak dispersal of salt-marsh species occurs between September and early December, which is consistent with the data reported by Huiskes et al. (1995). It is important, therefore, that managed realignment should take place before this period in order to exploit the dispersal season. A relatively rapid establishment of pioneer and low-marsh species may be expected, as adult plants of these species are frequently inundated and their seeds stand more chance of being dispersed by seawater than those of high-marsh species. This is clearly the case for Salicornia, of which many seeds were trapped in our study and which has successfully established and maintained its abundance in the vegetation within 3 years of a breach (Garbutt et al. 2002). It is also probable that the seeds are dispersed via stepping stones, i.e. being initially deposited a short distance from the parent plant from whence they are transported further with subsequent tides. Spring tides and storms may be especially important in this respect, as discussed by Koutstaal, Markusse & De Munck (1987). As a result of the predominant short-distance dispersal of salt-marsh species, it is advantageous to have a well-developed salt-marsh in front of the restoration site, as this will function as the source area. Usage of the restoration site by birds and the implementation of a livestock grazing regime in connection with the fronting marsh may further enhance seed dispersal into the site, especially of species less adapted to hydrochory. In order to obtain rapid results and/or species absent from the source area, broadcasting seeds or driftline material collected from high diversity marshes may be necessary (Hölzel & Otte 2003). However, as Morris et al. (2004) point out, positive sediment budgets will be a prerequisite, not only for salt marshes to keep up with rising sea levels, but also to increase the elevation of many realignment sites that have run into a sediment deficit during their period of embankment and are often below levels at which salt-marsh vegetation can develop.
We would like to thank Sijbren Otto for his help in the field, three anonymous referees for the useful comments to the manuscript, and we are grateful for the financial support of the Schure-Beijerinck-Popping fund to M. Wolters.