Influence of environmental factors on the growth and interactions between salt marsh plants: effects of salinity, sediment and waterlogging

Authors


Jonathan Huckle (fax 01244 373379; e-mail j.huckle@chester.ac.uk).

Summary

1 Artificial environmental gradients were established in a series of pot experiments to investigate the effect of salinity, sediment type and waterlogging on the growth, and interactions between Spartina anglica and Puccinellia maritima. In each experiment, one environmental variable was manipulated and plants grown in pairwise combinations to examine the effect of the environmental factor on the intensity of intra- and interspecific interactions, quantified using the Relative Neighbour Effect (RNE) index.

2Puccinellia was found to exert an asymmetric, one-way competitive dominance above ground over Spartina in experiments where gradients of sediment type and waterlogging were established. The intensity of the competition was highest in conditions with the least abiotic stress and lower or non-existent where stress was increased.

3 The intensity of the above-ground competition was greatest in loam and least in sand sediments. Reduction in competitive intensity in sand was accompanied by an increase in below-ground Spartina biomass and it is suggested that the production of rhizomes is a potential mechanism by which this species can expand vegetatively into areas without competition.

4 Interspecific competition on Spartina from Puccinellia also varied in intensity in the waterlogging experiment, being more intense in non-immersed treatments, where abiotic stress was reduced.

5 The competitive dominance of Puccinellia and the competition avoidance mechanism shown by Spartina in these experiments help to explain the successional interactions between the species along environmental gradients in natural salt marsh communities.

Introduction

The zonation of plant communities on many salt marshes has been interpreted as reflecting stages in a successional sequence (Chapman 1974). However, attributing the zonation observed across a spatial elevational gradient to changes on a temporal scale is an oversimplification of a complex dynamic habitat (Gray 1992). It has occasionally been possible to infer historic succession from changes in zonation patterns associated with the sediment accretion that is enhanced by vegetation (de Leeuw et al. 1993), but most studies indicate that zonation patterns are relatively stable over periods of decades (Roozen & Westhoff 1985).

The composition and distribution of plant communities along the elevational gradient of a salt marsh is related to the ability of individual species to tolerate environmental conditions associated with tidal variation. Tidal inundation may vary in frequency and duration (Brereton 1971; Rozema et al. 1985; Scholten & Rozema 1990), both of which decrease as the elevation of the marsh increases (Hubbard 1969). A range of abiotic factors vary in association with tidal inundation and those which have been intensively studied include salinity (de Leeuw et al. 1991; Rozema & van Diggelen 1991), sediment structure (Othman 1980; Thompson et al. 1991), immersion duration (Ranwell et al. 1964), soil redox potential (Armstrong et al. 1985; Groenendijk et al. 1987), disturbance in the form of burial by debris (Brewer et al. 1998) and nutrient levels (Levine et al. 1998; van Wijnen & Bakker 1999). The zonation of plant communities has often been related to variations in one or more of these factors, but relatively few studies have teased apart the specific effects of individual abiotic factors on the interactions occurring between salt marsh species at different points along the elevation gradient. Whilst it has been recognized that the type of interspecific interaction between plants may vary along abiotic gradients (Bertness & Callaway 1995; Brooker & Callaghan 1998), relatively few empirical studies have investigated changes in the relative importance of positive and negative interactions along a single gradient (but see Callaway 1998).

It is believed that individual species distributions at lower marsh elevations are broadly determined by tidal factors whilst interspecific competition becomes increasingly important at higher levels (Gray 1992). The abilities of a species to be competitive and tolerant to adverse physical factors are considered to be mutually exclusive (Grime 1979), and in areas of abiotic stress selective forces related to survival are more important than forces related to ‘winning’ competitive interactions with other plants (Brooker & Callaghan 1998). In the salt marsh environment, an increase in elevation is coupled with an increase in the number of potential species able to tolerate the conditions and there is a corresponding increase in the presence and intensity of interspecific interactions, which may either take the form of interspecific competition (Gray 1985; Scholten et al. 1987; Bertness & Yeh 1994), or of facilitation (Bertness & Shumway 1993; Callaway 1994; Castellanos et al. 1994).

This study attempts to separate the relative importance of three environmental factors (salinity, sediment type and waterlogging), all of which vary with marsh elevation and the corresponding degree of tidal inundation, on the growth and competitive interactions between two salt marsh species: Spartina anglica C.E. Hubbard and Puccinellia maritima (Huds.) Parl. (hereafter referred to as Spartina and Puccinellia, respectively). Salinity and waterlogging have been shown to affect both the distribution of species in a pioneer marsh dominated by Salicornia europaea and Puccinellia (Brereton 1971) and the competitive interactions between Puccinellia and two species of grass, Festuca rubra and Agrostis stolonifera, that are confined to high marsh elevations (Gray & Scott 1977). Spartina and Puccinellia are common perennial grasses at relatively low elevations in salt marshes throughout northern Europe. Spartina is characteristic of the pioneer zone and is often replaced by Puccinellia-dominated communities at adjacent mid-marsh elevations.

The presence of Spartina is a major problem facing managers of coastal habitats (Gray et al. 1991) as it colonizes estuarine mud-flats and accelerates the process of salt marsh development through increased sediment accretion. Throughout this century, expansion by natural mechanisms has been exacerbated by anthropogenic introductions of Spartina anglica and, in north-west America, of S. alterniflora (Daehler & Strong 1996) to the extent that Spartina spp. are present in salt marshes across the globe.

Understanding the relative importance of environmental factors and interspecific interactions in determining salt marsh community structure at different marsh elevations is essential if we are to clarify the ecological processes underlying the development of salt marshes. Environmental gradients that mirror the extent of natural variation were established to investigate the separate effects of three environmental factors and the interactions between two of the most important salt marsh species in the early stages of development of the salt marsh.

Methods

Preparation of experimental plant material

Spartina and Puccinellia plants were collected from the same area of the lower marsh zone of the Dee Estuary (National Grid Reference SD 270770) in April 1997. Spartina plants were unearthed carefully from the marsh sediment and transported from the marsh in water to prevent desiccation of root material. Puccinellia plants were collected by removing and transporting small areas of monospecific turf from the lower marsh. For each species, individual plants consisting of a single tiller with attached root material were separated and planted into individual pots containing a 50:50 sand/loam mix (standard horticultural sand and John Innes no. 2. potting compost). The plants were grown for 6 weeks in a polytunnel at Ness Botanic Gardens, Cheshire, UK (National Grid Reference SD 305756), and watered daily to provide a large pool of healthy individuals for use in the experiments. One week before transplantation, the experimental stock of plants was transferred outside to allow acclimatization to outdoor conditions.

Experimental design

In June 1997, three separate experiments were set up in which a single environmental variable (salinity, sediment or waterlogging) was applied at three levels while the other factors were kept constant.

Within each experiment a randomised block split-plot design was used with four replicate blocks. Each of the 12 main-plots (3 treatments × 4 replicates) contained subplot treatments (in pots of 10 L volume, 28 cm in diameter, 20 cm in height) representing a series of pairwise combinations of the two species (Table 1). For each species, a standard plant density of three individual plants or tillers per pot was used to establish subplot treatments of the species alone at standard density and in combination with itself and the other species. The plants for each block were selected at random from the experimental stock in an attempt to reduce the possible influence of clonal variation and plant history.

Table 1.  Planting densities for each species combination treatment grown in the salinity, sediment and waterlogging experiments
Number of plants of each species
Species combination codeSpartinaPuccinellia
  • P = Puccinellia, S = Spartina.

Control (0)
S3
P3
S + S6
P + P6
S + P33

The three salinity treatments were 30 g NaCl L−1, 15 g NaCl L−1 and freshwater (0.6 g NaCl L−1), and the solutions surrounding the pots were re-adjusted to these concentrations on a weekly basis. Sediment and waterlogging factors were standardized in this experiment by using a 50:50 sand/loam mixture and immersing all pots to a depth of 10 cm above the pot base, respectively.

Three sediment types were obtained by planting in 100% sand, 50:50 sand/loam mix, and 100% loam (details as mentioned above). Salinity was maintained at 15 g NaCl L−1 and all pots were immersed to a depth of 10 cm.

Plants were exposed to three levels of waterlogging by maintaining water levels in the main-plot at 10 cm below the pot base (No Immersion), 10 cm above the pot base (‘Half’-immersion) and at the level of the soil surface (‘Full’-immersion). Salinity was maintained at 15 g NaCl L−1 and a 50:50 sand/loam mix was used throughout. Pots that were not immersed were watered on a weekly basis using saline (15 g NaCl L−1).

Salinity and water levels were monitored weekly and adjusted back to initial conditions, either by adding water or saline so that salinity concentrations remained within ± 1 g NaCl L−1 of the baseline. The experimental pots were weeded for the first 10 weeks of the trials.

In November 1997, morphological measurements of Spartina were made in each pot before harvest. The number of vegetative and reproductive tillers, the height of the tallest flowering culm, and the length of the third leaf down from the culm were recorded since these variables have been shown to exhibit significant variation in previous studies on this species (Thompson 1991). Thereafter, each Spartina individual was cut at ground level, dried at 80 °C for 24 h and weighed: vegetative and reproductive components of above-ground biomass being weighed separately and the weights combined to give the total above-ground biomass per pot. The above-ground biomass of Puccinellia in each pot was also determined. The roots from each pot were removed by thorough washing and the below-ground biomass of each species in each pot determined.

Data analysis

The biomass and morphometric data for each species were analysed by analysis of variance (PROC GLM, SAS 1985), using a split-plot randomised block design. Biomass data for Spartina and Puccinellia were transformed (log10) prior to analysis. Morphological data for Spartina were transformed using either log10 (height of tallest culm, length of third leaf) or square root (number of tillers).

Both interspecific and intraspecific plant interactions were quantified using the Relative Neighbour Effect (RNE) index (Markham & Chanway 1996):

RNE  =  (P - N  -  P + N) / x

where P is a measure of plant performance in the presence (+N) and absence (–N) of neighbours, and x is the measure for the species with the greatest performance. This method is an improved version of the Relative Competitive Intensity (RCI) index, a method that has often been used to measure competition over a productivity gradient (Wilson & Keddy 1986; Belcher et al. 1995). The RNE index enables both competitive and facilitative interactions to be quantified without bias on a scale of − 1 to + 1. A value of 0 indicates that no interaction is occurring, with negative values indicating facilitation and positive values indicating competition. In this experiment, the different planting densities of the two species represent the presence or absence of neighbours: treatments with three plants per pot represent growth in the absence of neighbours, while those with six plants represent growth in the presence of either conspecific neighbours or plants of a different species.

Intra- and interspecific interactions were determined using both above and below-ground biomass as indicators of plant performance. For each experiment, the level at which the RNE index was calculated was dependent on which effects were significant in the analysis of variance of the biomass data for each species. Thus, where no environmental effects were detected, the mean RNE value for each type of interaction (interspecific and intraspecific) was calculated by pooling the values obtained from each of the 12 main-plots (4 blocks × 3 treatment levels). Where environmental effects were detected, however, the mean RNE for each type of interaction was calculated at each treatment level individually. Statistical analysis of the RNE values was conducted using arcsin square root transformed data. Relative neighbour effects were defined as competitive or facilitative when the mean values of the index differed from zero (no interaction) by more than the Least Significant Difference (LSD) calculated from the data.

Results

Effects of individual treatments on plant performance

Effects of salinity

There were no significant differences in the above or below-ground biomass of Spartina and Puccinellia when grown at the three salinity levels (Table 2a). For all three Spartina biomass variables, species combination effects were due to a significantly greater biomass where Spartina was grown at double the standard planting density (i.e. S + S treatment) when compared with pots where Spartina is grown on its own (S) or with Puccinellia (S + P) (data not shown). Similarly, Puccinellia biomass was only significantly greater in pots grown at double density (P + P) compared with the other treatments.

Table 2.  Summary of split-plot analyses of variance showing the significance of the effects of block, environmental treatment, and species combination on biomass variables of Spartina and Puccinellia (*P < 0.05, **P < 0.01, ***P < 0.001)
SpartinaPuccinellia
ExperimentEffectAbove-groundBelow-groundTotal biomassAbove-groundBelow-groundTotal biomass
(a) SalinityBlock      
Treatment      
Species***************
(b) SedimentTreatment × Species
Block
    **
Treatment*     
Species****    
(c) WaterloggingTreatment × Species
Block
 *
**
*
**
*  
Treatment*** *   
Species*** ***** **
Treatment × Species***   

Salinity showed no effects on the Spartina morphological characteristics but species combination effects were found in this experiment for the number of vegetative tillers (and consequently total tillers), and on the length of the third leaf (Table 3).

Table 3.  Summary of split-plot analyses of variance showing the significance of the effects of block, environmental treatment, and species combination on morphological variables of Spartina. (*P < 0.05, **P < 0.01, ***P < 0.001)
Spartina morphological variable
ExperimentEffectNo. of tillersNo. of vegetative tillersNo. of flowering tillersCulm HeightLength of 3rd leaf
(a) SalinityBlock     
Salinity     
Species******  *
(b) SedimentSalinity × Species
Block
*
    
Sediment     
Species********** 
(c) WaterloggingSediment × Species
Block
**
*
**
  
Waterlogging*** * 
Species********  
Waterlogging × Species     

Although biomass and number of tillers were greater in the Spartina (S + S) treatments, levels were less than double those in the single density treatment (S) indicating that the average number of vegetative tillers per plant was less than when grown at lower density (data not shown).

Effects of sediment

The sediment type had a significant effect on the above-ground component of Spartina biomass (Table 2b), with biomass lower in sand compared to either loam or the mixed substrate (Fig. 1a). When sediment treatments were combined, biomass in double density Spartina pots was significantly greater than the other species combinations for both above- and below-ground biomass. For Spartina above-ground biomass the value in combination with Puccinellia (S + P) was significantly lower than the other species combinations (Fig. 1b). Differences in Spartina below-ground biomass between species combinations, although not significant when using combined sediment treatments, were insufficient for an effect to be visible in total biomass (Table 2b).

Figure 1.

Plant biomass in the sediment experiment: data shown are the mean biomass per pot for (a) effect of sediment type on Spartina above-ground biomass, (b) effect of Spartina species combination on Spartina above-ground biomass, (c) effect of sediment type and Spartina species combination on Spartina below-ground biomass, and (d) above-ground and below-ground biomass of Puccinellia with pooled block and sediment treatments. Treatments with different code letters are significantly different (Tukey HSD test: P < 0.05).

Effects of sediment on below-ground biomass showed a different trend for the S + P combination compared to the S and S + S species combinations: below-ground biomass was greater in loam than sand for the S and S + S combinations, but was significantly less than sand for the S + P combination (Fig. 1c). Comparing the above- and below-ground biomass for the S + P species combination in the different sediment treatments, a reduction in above-ground biomass is accompanied by a reduction in below-ground biomass in a loam substrate, but was associated with a significant increase in below-ground biomass when sand is present in the substrate.

No significant effects of sediment type or species combination were detected in the analyses of variance for any Puccinellia biomass variable (Table 2b, Fig. 1d), although the Tukey’s HSD test indicated a significantly greater above-ground biomass in the loam and mixed sediments than in sand.

In the sediment experiment, vegetative tiller production was greater in the S + S treatments and reduced when Spartina was grown with Puccinellia (S + P) (species effect in Table 3b; Fig. 2a). There was a general trend of greater vegetative tiller production in loam when compared to sand, except in the S + P treatment, where the number of tillers was greater in sediments containing sand (Fig. 2a). Similarly, Spartina reproductive tiller production in monospecific treatments was greater in loam than sand, a pattern reversed in the S + P treatment (Fig. 2b), again indicating an interspecific interaction. The reduced height of the flowering culms in the S + P combination in loam compared to sand suggests that interference from Puccinellia is exerting its effect on the Spartina above-ground tillers (Fig. 2c).

Figure 2.

Morphological characteristics of Spartina in the sediment experiment: data shown are the mean values using pooled block data for (a) number of vegetative tillers per pot, (b) number of flowering tillers per pot, (c) height of tallest flowering culm, and (d) length of the third leaf on the tallest culm. ▪, loam; bsl00008, mixed sediment; □, sand. Treatments with different code letters are significantly different (Tukey HSD test: P < 0.05): uppercase letters denote differences between species combinations; lowercase letters denote differences between sediment types.

Effects of waterlogging

Effects of immersion and species combination were detected only for the above-ground biomass of Spartina (Table 2c), although interaction between immersion and species combination treatments was also detected for below-ground Spartina biomass.

Spartina above-ground biomass was significantly reduced in pots where there was no immersion. Species combination effects were similar to the sediment experiment, with significantly more biomass in S + S pots than in S, with a further reduction in S + P pots (Fig. 3a). The significant ‘immersion × species combination’ interaction (Table 2c) was a consequence of the significant reduction in Spartina above-ground biomass caused by the presence of Puccinellia being even more pronounced in pots without immersion than in immersed pots. An analogous reduction in Spartina below-ground biomass explained the significant ‘immersion × species combination’ interaction for this variable (Fig. 3b). For Puccinellia, only species combination effects were observed (Table 2c), attributable to the above-ground biomass of P + P pots being significantly higher than those of P and S + P (Fig. 3c).

Figure 3.

Plant biomass in the waterlogging experiment: data shown are the mean biomass per pot for (a) effect of immersion treatment and Spartina species combination on Spartina above-ground biomass, (b) effect of immersion treatment and Spartina species combination on Spartina below-ground biomass, and (c) above-ground and below-ground Puccinellia biomass with pooled block and immersion treatments. Treatments with different code letters are significantly different (Tukey HSD test: P < 0.05): uppercase letters denote differences between species combination; lowercase letters denote differences between immersion treatments.

Vegetative tiller production was greater in waterlogged plots (full and half immersion) than in non-immersed plots for all Spartina species combinations. Vegetative tiller production was least in the S + P combination and greatest in the S + S combination for all immersion treatments (species effect in Table 3c, Fig. 4a). Reproductive tiller production was significantly greater in the S + S treatment (Fig. 4b), but the S and S + P treatments were similar. For both vegetative and reproductive tillers, the increase in numbers produced in the S + S combination is less than the increase in original planting density, indicating that tiller production is inhibited by intraspecific competition.

Figure 4.

Morphological characteristics of Spartina in the waterlogging experiment: data shown are the mean values using pooled block data for (a) number of vegetative tillers per pot, (b) number of flowering tillers per pot, (c) height of tallest flowering culm, and (d) length of the third leaf on the tallest culm. ▪, full immersion; bsl00008, half immersion; □, no immersion. Treatments with different code letters are significantly different (Tukey HSD test: P < 0.05): uppercase letters denote differences between species combinations; lowercase letters denote differences between sediment types.

Effects on plant interactions

Effects of salinity

Puccinellia had no effects on the growth of Spartina in the salinity experiment (mean RNE index close to zero), although an intraspecific competitive interaction was detected for the above-ground biomass (positive RNE value; Fig. 5a). Puccinellia above-ground biomass was also subject to intraspecific competition, and its below-ground biomass was affected by competition from Spartina (Fig. 5b).

Figure 5.

Plant interactions in the salinity experiment: data shown are the mean interspecific and intraspecific Relative Neighbour Effects (RNE) index using pooled block and salinity treatment data for (a) Spartina and (b) Puccinellia biomass. □, above-ground biomass; bsl00008, below-ground biomass. Error bars indicate ± 1 SE; * indicates that the RNE value is greater than the Least Significant Difference from zero at P < 0.05.

Effects of sediment

RNE values showed that Puccinellia had a competitive effect on Spartina above-ground biomass in all sediments, although not significant in sand, and there was an inverse relationship between the intensity of the interaction and the amount of sand in the sediment mixture (Fig. 6a). Below-ground, although there was a significant interspecific competitive interaction of Puccinellia on Spartina in the loam, both the mixed and the sand sediments showed significantly negative RNE values, indicating facilitative effects (Fig. 6b). These data support the findings of the anova of the biomass data, suggesting that, in sandy substrates, Spartina is able to avoid above-ground competition from Puccinellia by investing in production of below-ground biomass. The only significant interspecific effect of Spartina on Puccinellia was an apparent facilitation of below-ground biomass (Fig. 6d).

Figure 6.

Plant interactions in the sediment experiment: data shown are the mean interspecific and intraspecific Relative Neighbour Effects (RNE) index in each sediment treatment using pooled block data for (a) Spartina above-ground biomass, (b) Spartina below-ground biomass, (c) Puccinellia above-ground biomass, and (d) Puccinellia below-ground biomass. ▪, loam sediment; bsl00008, mixed sediment; □, sand sediment. Error bars indicate ± 1 SE; * indicates that the RNE value is greater than the Least Significant Difference from zero (***P < 0.001, **P < 0.01, *P < 0.05).

Above ground, intraspecific competition was observed in all sediments for Spartina (Fig. 6a), although it was only statistically significant in the loam sediment. Intraspecific competition was detected for Puccinellia above- and below-ground biomass components in both the loam and mixed sediments, but not in sand (Fig. 6c,d).

Effects of waterlogging

As in the sediment experiment, Puccinellia exerted an interspecific competitive effect on the above-ground biomass of Spartina. Intensity of competition was greater in main plots with no immersion when compared to half or fully immersed main plots (Fig. 7a). Relative neighbour effects of Puccinellia on Spartina below-ground were varied with no interaction with full immersion, facilitation at half immersion and competition at no immersion (Fig. 7b). Less intense above-ground competition in the half-immersion treatment presumably allows a facilitative increase in below-ground biomass, whereas the intense above-ground competition in the no-immersion treatment may ensure that the competitive effect is also expressed below-ground as well. No significant interspecific interactions were detected on the above- or below-ground biomass of Puccinellia (Fig. 7c,d). Intraspecific competition was detected for both above- and below-ground biomass for both Spartina and Puccinellia (Fig. 7a–d).

Figure 7.

Plant interactions in the waterlogging experiment: data shown are the mean interspecific and intraspecific Relative Neighbour Effects (RNE) index in each immersion treatment using pooled block data for (a) Spartina above-ground biomass, (b) Spartina below-ground biomass, (c) Puccinellia above-ground biomass, and (d) Puccinellia below-ground biomass. ▪, full immersion; bsl00008, half immersion; □, no immersion. Error bars indicate ± 1 SE; *, RNE value is greater than the Least Significant Difference from zero (***P < 0.001, **P < 0.01, *P < 0.05).

Discussion

Our experiments show that environmental factors have significant effects on the growth of the two salt marsh species and that interactions occur between them at both an intra-and interspecific level. Of the three environmental factors studied, sediment and waterlogging affected biomass production of Spartina and all three factors affected the competitive interactions between the species. In all experiments where plant interactions were detected, the interspecific competition was found to be more intense than intraspecific competition.

Effects of salinity, sediment and waterlogging on plant response

When grown in monospecific combinations, Spartina biomass was reduced when grown in sand, a result that agrees with previous population studies on this species (Thompson et al. 1991). However, our study showed a differential response of Spartina’s biomass components to the type of sediment: Spartina’s below-ground biomass was reduced in sand, whilst its above-ground biomass was increased in loam. We suggest that this slight variation in the response of the two components of the biomass may arise from influences of different components of the sediment, the coarse-grained sandy sediments inhibiting root development and loam having a nutrient-related effect leading to an increase in above-ground biomass.

Above-ground biomass of Spartina was reduced by both intraspecific competition and interspecific competition from Puccinellia and the intensity of these interactions was greatest in loam sediments. Physical stress from environmental factors (such as coarse sediment composition and low nutrient availability) are less intense in the loam sediments and the results support the assertion that plant interactions will become more important as stress is reduced. Further insight into the interspecific interaction between Spartina and Puccinellia was obtained from the effects of Puccinellia on Spartina below-ground biomass, where a competitive effect was detected in the loam sediment and a facilitative effect detected in the mixed and sand sediments. We interpret these results as a sediment-dependent response to the presence of Puccinellia. In loam, alleviation of treatment-imposed stress is combined with more intense interspecific competition from Puccinellia, thus inhibiting both above- and below-ground Spartina biomass. However, in sandy sediments, competition is less intense and Spartina appears to respond to above-ground interference from Puccinellia by the production of below-ground structures, a response not seen when exposed to intraspecific competition. This suggestion is supported by the morphological data, which showed that the numbers of Spartina vegetative tillers parallel the increase in Spartina below-ground biomass. We suggest that, by investing in underground rhizomes that give rise to innumerable small vegetative tillers, Spartina may be able to avoid above-ground competition by spreading vegetatively into areas more favourable for growth.

In the waterlogging experiment, lack of immersion reduced the above-ground biomass of Spartina grown in monospecific combinations. This contrasts with investigations into the effects of waterlogging in field conditions which have shown reduced growth of Spartina species in waterlogged sediments (Groenendijk et al. 1987; Mendelssohn & McKee 1988), attributable to long-term exposure to a reducing environment and the associated accumulation of sulphide and oxygen deficient roots. The duration of this experiment may be too short to produce conditions that affect growth. This would explain the lack of waterlogging effects on biomass in Puccinellia, a species that is known to survive only short periods of exposure to sulphide in the latter part of the growing season (Havill et al. 1985).

Both intra- and interspecific competition affected the growth of above-ground Spartina in the waterlogging experiment. Intraspecific competition did not vary in intensity between waterlogging treatments, but above-ground interspecific competition was greater in non-waterlogged treatments compared to those experiencing full or half immersion. Spartina below-ground biomass was only affected when the stress of non-waterlogged soils was combined with the presence of Puccinellia. As in the sediment experiment, no effect of immersion was detected on the growth of Puccinellia. However, effects of waterlogged soils may have been obscured in the case of Puccinellia above-ground biomass. In full- and half-immersion treatments, increased stolon extension (not measured) may have resulted in above-ground biomass of Puccinellia being higher than would be expected in field conditions: stolons of Puccinellia were able to float on the water surface of immersed treatments and root into the water, thus spreading further than where the water level was beneath the base of the pots.

No effect of salinity on the growth of Spartina or Puccinellia was detected for either above- or below-ground components of biomass, despite the experimental treatments being set at comparable levels to those found to have significant effects on growth in other studies (Rozema & van Diggelen 1991) and reflecting the range of salinity experienced in field conditions. In this experiment, main-plots were individually established with species combination treatments immersed in a relatively shallow reservoir of saline water (10 cm). As a consequence, the species combination treatments may have been more exposed to the influences of rainfall, which would have had a diluting effect on the pots from above and counteracted the influences of saline water infiltrating into the soil from the base of the individual pots than if they had been fully immersed. This experiment may thus be more representative of high salt marsh elevations where rainfall has a greater influence than tidal water in determining soil salinity levels (de Leeuw et al. 1991). Intraspecific competition was, however, detected for Spartina and Puccinellia above-ground biomass, but not for below-ground biomass, suggesting an interference effect from conspecifics for each species.

Overall, both species generally showed intraspecific competition both above- and below-ground. Interspecific competition was asymmetric with Puccinellia competitively reducing above-ground production of Spartina to a greater degree than did intraspecific competition, but with Spartina having little or no effect of on Puccinellia biomass.

Comparison of experimental results with salt marsh conditions

Spartina colonizes bare mud-flats at lower marsh elevations in the Dee Estuary. Coarse-grained sediments and frequent tidal inundation impose environmental stress in these areas, and although Spartina is able to tolerate such conditions when experimentally imposed, it performs better in more nutrient-rich, loamy sediments, similar to those higher up the marsh. This finding corresponds well with previous studies where Spartina plants in pioneer zones were shorter than those from further up the marsh (Thompson 1990). This difference in growth of Spartina cannot be explained by phenotypic plasticity caused by age-related somatic variation of clones (Thompson et al. 1991; Thompson et al. 1993), as all plants in our study were collected from clones from the same area of pioneer marsh. However, the reduced biomass and tiller production of Spartina plants grown in coarse-grained sediment may reflect the greater degree of stress in these conditions.

The impact of the presence of Puccinellia on Spartina biomass, and the way that it changes across the sediment gradient is particularly interesting. In these experiments we have shown competition, as measured by the Relative Neighbour Effect index, to be most intense in loam sediments, resulting in Spartina biomass being inhibited both above- and below-ground. In coarse-grained sandy sediments, however, above-ground competition from Puccinellia is less intense. At this position along the environmental gradient, it appears that the stresses imposed by the environmental conditions play a more important role than the interactions between species. This finding supports the idea of an inverse relationship between stress tolerance and competitive ability (Grime 1979). Whilst above-ground competition becomes less intense in sandy sediments, at the below-ground level, Spartina biomass actually increases, suggesting a facilitative interaction. An investment in rhizomes could be interpreted as a response to nutrient-poor conditions and an attempt to migrate to more favourable patches; however, such a response would also be expected to occur in monospecific combinations. We suggest that this finding may be explained by the hypothesis that Spartina is able to respond to less intense levels of above-ground interspecific competition by investing in below-ground biomass.

In the waterlogging experiment, a similar trade-off occurs between the competitive ability of Puccinellia and the stress tolerance of Spartina. This experiment shows growth of Spartina to be highest in immersed treatments and Puccinellia to have a greater competitive effect in non-immersed treatments, suggesting that the stress imposed on Puccinellia by immersion has a debilitating effect on its competitive ability.

These factors may be strong forces in determining salt marsh structure and development. At lower elevations, Spartina may be able to evade competition by rhizome extension when above-ground interference hinders tiller development. As elevation increases, however, and environmental conditions become analogous to those experienced in the loam and non-waterlogged experimental treatments, the intensity of competition increases and results in competitive exclusion by Puccinellia.

Previous studies of the competitive interaction between these species have suggested that Puccinellia out-competes Spartina because it develops earlier in the year and reduces the competitive ability of Spartina by preventing light from reaching developing shoots (Scholten & Rozema 1990). The relatively late canopy development of Spartina is due to it being a C4 species, with very low photosynthetic efficiency at low temperatures resulting in the ability to grow and reproduce being limited to months where the average temperature rises above 9–10 °C (Long 1983). This experiment, where both species were planted simultaneously, suggests that the effects of earlier development of Puccinellia may be compounded by the direct physical interference of above-ground Puccinellia stolons on Spartina tillers in summer and autumn.

One interesting feature is the lack of environmental effects on the growth of Puccinellia, indicating that it grows equally well in all the experimental conditions and that other factors may be responsible for limiting growth in field conditions. One possible explanation is that Puccinellia development in the field is affected by physical stress imposed by the action of repetitive tidal inundation. In this experiment, no attempt was made to create tidal periodicity or the physical impact of tides. It can be postulated that in field conditions Spartina, being rhizomatous and relatively robust, would be more suited to withstand these physical stresses than the stoloniferous Puccinellia. Thus Spartina may facilitate the establishment of Puccinellia through provision of sheltered environments, in combination with enhanced accretion of silt sediments. Such facilitation has been shown to exist in salt marshes colonized by Spartina maritima (Castellanos et al. 1994) with subsequent suppression of S. maritima tillers by a more competitive invader.

Our study has shown how the interaction of abiotic and biotic factors, particularly competition, is fundamental to the generation of plant zonation patterns in salt marshes. In recent field-based studies in a North American salt marsh, it was found that competitive hierarchies could be reversed by nutrient-enhancement, leading to the domination of species usually displaced to physically stressful tidal elevations (Levine et al. 1998). In a long-term field study of a European coastal barrier salt marsh, it was found that nutrient enhancement led to an acceleration in successional change (van Wijnen & Bakker 1999). Our study, using an experimental pot-based approach, has demonstrated that the competitive interaction between Spartina and Puccinellia is asymmetric, with Puccinellia having a competitive effect on Spartina, but not vice versa. Furthermore, this competitive effect is intensified in less physically stressful environments, such as those found in fine-grained, nutrient-rich sediments exposed to less tidal inundation, allowing the more competitive species to become dominant. We suggest that this supports the findings of van Wijnen & Bakker (1999), and that the competitive hierarchy established may arise as a consequence of the different life strategies of the two species. Spartina is a rhizomatous perennial possessing a C4 physiology, whilst Puccinellia is a stoloniferous C3 perennial. In Northern Europe, Spartina is late to emerge, the temperature dependence of the C4 pathway causing its growing season to be delayed (Long 1983; Gray et al. 1991). Puccinellia, however, is able to start growing relatively early in the year. In this way, our findings support the idea that the competitive dominance exhibited by Puccinellia over Spartina is related to early emergence and driven by competition for light (Bertness 1991; Goldberg & Miller 1990). At latitudes lower than our study site, where the growing season of Spartina is unlikely to be delayed, the observed hierarchy may be diminished or even reversed.

The findings of these experiments have important implications for conservation and the prediction of salt marsh community development, particularly as these may interact with changes in climate. We have proposed a mechanism by which competitive displacement of species may occur, in that the displacement of Spartina in physically stressed environments is an active process. Where Spartina is affected by above-ground competition, increased production of below-ground vegetative rhizomes may allow vegetative spread of Spartina into areas with less competition. Further work is needed to elucidate the validity of this mechanism in field conditions.

Acknowledgements

This work was funded by a Departmental Research Bursary from the Biology Department, University College Chester, and was supported by a grant from the Royal Society for the Protection of Birds. We thank Wolfgang Bopp for his technical support at Ness Botanic Gardens, and Jean Almond, Katie Nield, Gary Smith and Martin Crockett for their assistance in the field. Alan Gray and two anonymous referees are thanked for their comments on an earlier version of this paper.

Received 27 May 1999revisionaccepted 4 January 2000

Ancillary