Spartina alterniflora genotype influences facilitation and suppression of high marsh species colonizing an early successional salt marsh



    Corresponding author
    1. National Wetlands Research Center, US Geological Survey, 700 Cajundome Boulevard Lafayette, Louisiana 70506, USA;
    Search for more papers by this author

    1. National Wetlands Research Center, US Geological Survey, 700 Cajundome Boulevard Lafayette, Louisiana 70506, USA;
    2. University of Louisiana at Lafayette, Biology Department, PO Box 42451, Lafayette, La 70504–2451, USA
    Search for more papers by this author

    1. National Wetlands Research Center, US Geological Survey, 700 Cajundome Boulevard Lafayette, Louisiana 70506, USA;
    2. University of Louisiana at Lafayette, Biology Department, PO Box 42451, Lafayette, La 70504–2451, USA
    Search for more papers by this author

    1. Department of Ecology, Faculty of Biological Sciences, University of South Bohemia, Branisovska 31, Czech Republic
    Search for more papers by this author

    1. National Wetlands Research Center, US Geological Survey, 700 Cajundome Boulevard Lafayette, Louisiana 70506, USA;
    Search for more papers by this author

C. Edward Proffitt (tel. +1 337/266 8603, fax +1 337/266 8586; e-mail


  • 1Genetically based phenotypic and ecotypic variation in a dominant plant species can influence ecological functions and patterns of recruitment by other species in plant communities. However, the nature and degree of importance of genotypic differences is poorly understood in most systems.
  • 2The dominant salt marsh species, Spartina alterniflora, is known to induce facilitative and competitive effects in different plant species, and the outcomes of interactions can be affected by nutrients and flooding stress. Clonal genotypes, which maintained their different plant architecture phenotypes throughout 31 months of a field experiment, underwent considerable genet-specific senescence in their centres over the last 12 months.
  • 3Different clonal genotypes and different locations (robust edges vs. senescent centres) permitted significantly different levels of light penetration of the canopy (14.8–77.6%), thus establishing spatial heterogeneity for this important environmental factor.
  • 4S. alterniflora clonal genotype influenced the degree of suppression of the previously dominant Salicornia bigelovii as well as facilitation of recruitment and growth by other plant species. Aster subulatus and Atriplex patula performed better in Spartina clone centres, and experienced reduced growth in Salicornia-dominated areas.
  • 5Four other high marsh species (Borrichia frutescens, Aster tenuifolius, Iva frutescens and Limonium carolinianum) colonized only into Spartina clones but not into the Salicornia-dominated area.
  • 6These results suggest that differences in clone size, centre senescence, stem density, height, total stem length and biomass in different genotypes of a dominant marsh plant species can influence recruitment and growth of other plant species. The spatial pattern of habitat heterogeneity is, at least in part, dependent on the genotypic diversity, and possibly the genetic diversity, of such foundation species.
  • 7We hypothesize that as genotypic diversity increases in populations of a dominant plant species like S. alterniflora, the number and diversity of interactions with other species will increase as well.


The evaluation of species interactions is fundamental to an adequate understanding of plant community structure and ecosystem functioning. Species interactions have been shown to be important factors in determining spatial patterns of plant distribution and dominance (Grace & Wetzel 1981; Grace & Tilman 1990; Keddy 2001). A plant population, however, is comprised of an array of individual genets that can differ in their responses to environmental factors, one another and other species. Our understanding of the contribution of the different plant genotypes within a population to responses to other organisms, such as competitors, herbivores and disease organisms, as well as to disturbance and environmental gradients, lags behind basic studies of species–species interactions, in which it is often assumed that all individuals are homogeneous or that the law-of-large numbers will ‘average out’ individual variation. It is possible that the number and relative magnitudes of biotic interactions will increase as the genotypic and genetic diversity in dominant, or foundation, plant species increases. Booth & Grime (2003) reported that experimental calcareous grassland plots containing greater within-species genetic diversity lost species more slowly than did plots with low genetic diversity. This suggests that genetic diversity of dominant plant species can mediate important biological interactions.

Salt marsh communities provide a good model for assaying for genotypic differences because they are floristically simple and often dominated by one or a few herbaceous species (Pennings & Bertness 2001). On the Atlantic and Gulf of Mexico coasts of the United States, the low-mid tidal dominant grass species, Spartina alterniflora Loisel, has been found to have considerable genetic and genotypic diversity (Travis et al. 2002; Travis et al. 2004; Travis & Hester 2005), which affects growth and morphology (Proffitt et al. 2003) and ecological functions (Seliskar et al. 2002).

Salt marsh zonation results from a combination of species-specific differences in response to stress gradients and resource regimes, recovery following disturbance and competition among species (Bertness & Ellison 1987; Ellison 1987; Bertness 1991a,b; Levine et al. 1998a,b; Costa et al. 2003; Ewanchuk & Bertness 2004; Pennings et al. 2005). In the low intertidal, S. alterniflora is often the dominant species because of its high tolerance of frequently flooded, hypoxic conditions (Valiela & Teal 1974; Mendelssohn & Seneca 1980; Gleason & Zieman 1981; Mendelssohn et al. 1981). Although S. alterniflora can grow over a wide elevation range, its distribution and dominance at higher marsh elevations is limited by competition (Bertness 1991a,b; Crane et al. 2004). Competitive outcomes, however, can be affected by nutrient conditions (Covin & Zedler 1988; Levine et al. 1998a,b). Its various genotypes show different responses to elevation in terms of growth rates and reproductive patterns (Proffitt et al. 2003), and thus it is possible that genotypes, and their diversity, density, and distribution within the population, may also influence the outcomes of competitive and facilitative interactions.

Phenotypic variation in clone growth and morphology of S. alterniflora is, in part, genetically based (Seneca 1974; Gallagher et al. 1988; Seliskar et al. 2002; Proffitt et al. 2003). Variation in plant architecture can influence spatial heterogeneity within the marsh, which affects ecosystem function properties such as microbial respiration, edaphic chlorophyll and larval fish use (Seliskar et al. 2002), as well as interactions among plant species (Boyer & Zedler 1999). The interaction between genotype and microenvironmental conditions may lead to a variety of plastic growth and morphology responses that greatly increase the effects of this plant on salt marsh structure and function (Seliskar et al. 2002). Thus the wide-ranging S. alterniflora can be described as a foundation species, modifying the environment and providing resources for other species (Bruno & Bertness 2001).

Populations of S. alterniflora become established by both vegetative means (fragmentation and tidal transport followed by clonal growth) and seedling recruitment (Proffitt & Young 1999; Proffitt et al. 2003; Travis et al. 2004). Clonal and genetic diversity increases in young, rapidly expanding populations and later declines, possibly through interclone competition or genet loss through stochastic processes (Travis et al. 2002; Travis et al. 2004; Travis & Hester 2005). It is possible that the diversity of plant–environment interactions follows a similar pattern as interactions in an increasingly heterogeneous habitat may vary with differences in phenotype and clone morphology. Temporal as well as spatial heterogeneity would translate into variation in the ecological functions affected by S. alterniflora populations.

One form of habitat heterogeneity is produced by the variable extent to which S. alterniflora clonal patches show senescence in their centres (Stiller & Denton 1995; Proffitt et al. 2003). This aspect of clonal morphology can vary with plant genotype and environmental gradients such as elevation (Proffitt et al. 2003), but its effects on other species is poorly understood. Habitat heterogeneity may also be produced by among-clone variation in plant size or density.

We posed the following research questions. (i) Does the transplantation of S. alterniflora clones into an area of marsh dominated by the annual Salicornia bigelovii Torr. alter plant species composition and the rate of assemblage development in a young salt marsh? (ii) Do differences in S. alterniflora morphology (such as clone area, clone centre senescent area, stem density and height) among clonal genotypes produce additional variation in heterogeneity in soil and light variables? (iii) Does the strength or nature of facilitation or suppressive interactions vary with S. alterniflora genotype?


site background

The study site was a c. 200-ha mudflat located north of West Cove in the Sabine National Wildlife Refuge, south-western Louisiana, USA. The area was created in 1996 from dredged sediment with high clay and low organic matter contents (Proffitt & Young 1999; Edwards & Proffitt 2003). Pore and surface water salinities ranged from fresh to near seawater strength, depending on season and year (Proffitt & Young 1999), typically ranging from < 10 to > 20 g L−1. The site is bounded on two sides by levees but adjoins natural salt marshes dominated by S. alterniflora, Spartina patens (Ait.) Muhl. and Distichlis spicata (L.) Greene on the other sides.

Soils in the marsh have a high clay content (40–50%) and soil organic matter is generally very low (2.9–7.4%) (Proffitt & Young 1999; Edwards & Proffitt 2003). Bulk density in the created marsh (0.95 g cm−3) is significantly higher than in natural marshes, although it is expected to decrease linearly over time (Edwards & Proffitt 2003). At −132 to −125 mV at 2 and 10 cm depths, the soils of the restored site are generally reducing, although a little less so than in a nearby reference marsh (−159 to −221 mV, means of three mid-day July readings; R. L. Chiasson, unpublished Masters thesis).

Spartina alterniflora colonized all elevations of the new marsh within the first year, via rafting of patches of roots and rhizomes from nearby marshes, and, in subsequent years, through seed (Proffitt & Young 1999; Proffitt et al. 2003; Travis et al. 2004). Densities of clones decreased with distance from the border with the natural marsh (Proffitt et al. 2003). The annual Saliconria bigelovii initially colonized at very low densities through tidal transport of seed, but became dominant in virtually all non-grassed regions of the restored site that were not permanently flooded, following several annual cycles of reproduction and seedling recruitment (Edwards & Proffitt 2003; Proffitt et al. 2003).

experimental design

Spartina alterniflora sods were collected as described in Proffitt et al. (2003) from each of five donor clones selected at random from a number of spatially distinct clones that had colonized a limited area in the interior, higher elevation region of the created marsh. DNA analysis (amplified fragment length polymorphisms) established that the five donor clones (A, B, C, D and E) were actually different genotypes (Proffitt et al. 2003). The donor clones came from similar elevations (+36.8 (3.3 SE) cm) and were approximately the same age (2.5 years) and diameter (8.84 (0.65 SE) m; range 3.6 m). Soil conditions at the donor location were very uniform (clay sediments, very low organic matter, relatively low soil salinity because of the mixing of the sediment-water slurry during dredge-and-fill of this site; see Edwards & Proffitt 2003 for more detail). Because of this uniformity in clone age, size and soils we did not deem it necessary to include a procedural control to control for transplant shock. Instead, any transplant that died was replaced (see Proffitt et al. 2003). Each sod collected consisted of 10 interconnected ramets (stems plus roots connected by rhizomes), which were left intact for re-planting. Intact sods were planted into a low-to-medium density monoculture of S. bigelovii (field observations) in February 1999 in a five-row experimental array starting with row 1 at 29°54′17.3″ N, 93°22′31.0″ W, and with successive rows located 10 m further east and 12 planting positions arranged at 7-m intervals further south in each row. Eight additional planting positions were not used in the current analyses because plantings had grown together completely and individual clones could no longer be distinguished. Two sods (either a single clone or a pair of clones) were planted 1 m east and west of each position, or a single sod was planted. All possible paired combinations of the five genotypes were used.

elevations, water and light penetration

Elevations were recorded as NGVD (National Geodetic Vertical Datum) using a closed-loop traverse survey with a Spectraphysics 750 laser level. A well was installed in September 2000 within c. 2 m of the north end of the experimental plantings and in a natural reference marsh dominated by S. alterniflora located about 200 m west of the experimental plots. Water depth was recorded every 30 minutes via Global Water WL14 Water Level Loggers.

Photosynthetically active radiation was recorded above and below S. alterniflora canopies using a light wand (Decagon Devices, Inc., Pullman, WA, USA) between 1000 and 1400 hours in July 2001. Percentage light penetration of the canopy was calculated from these values. Surface soil samples (upper 2 cm) were collected and pore water extracted for salinity analyses. Percentage soil moisture was determined by weighing, drying and re-weighing.

Pore water from the root zone (10 cm deep) of the experimental clones was sampled in April and July in clone centres and edges. Water was extracted from soil by centrifugation: salinity was measured with a refractometer and nutrient (NH4-N and PO4-P) concentrations analysed with a LACHAT system (Quickchem IV, Lachat Instruments, Milwaukee, Wisconsin, USA).

biotic variables and statistical analyses

Clone diameter, diameters of clone centres that were senescing, densities of live and dead stems, mean stem heights and total stem length were recorded in July and August 2001. Initial collections of above-ground plant material showed that total length of stems m−2 (X) could be used as a non-destructive measure to predict above-ground biomass (Y) in g m−2 of S. alterniflora: (Y = 0.075X − 360.128; R2 = 0.758, F = 0.27.628, P < 0.0005).

We calculated the areas of central senescence and robust growing edge for each clone and tested these two response variables for differences among clonal genotypes by mancova. Plant architecture variables (mean height, total length of stems m−2, live stem density, dead stem density, and biomass) were analysed by mancova with main effects of clonal genotype and within-clone location (edge vs. centre zones) and their interaction (Systat 2002). Pore water variables (soil moisture, salinity and nutrients) were analysed by mancova with clone genotype, within-clone location and time (April or July) as main factors. Elevation was the covariate in all mancovas. Differences in percentage light penetration by genotype and location were analysed by two-way ancova with elevation as a covariate. We used multivariate analyses on these variables because we felt that they might operate in concert rather than separately. However, we present both the multivariate and univariate analyses for all mancovas.

A list of all plant species recruiting into the planted Spartina clones or Salicornia-dominated area around the clones was made over the 2.5 years of the project. At the final data collection in July and August 2001, species richness (number of species m−2) was assessed along with the total listing of species colonizing the experimental site. Species richness and a variety of stem density and size variables of several abundant species in the site were assessed by separate two-way anovas for each species and response variable with Spartina clone genotype and location (clone centre, clone edge, outside clones) as main effects and with elevation as a covariate. Species response variables included live plant densities, total stem lengths per quadrat and mean heights of Salicornia bigelovii (the dominant at the start of the experiment) and the two most abundant species colonizing the experimental setting (Aster subulatus Michx. and Atriplex patula L.). To correct for non-normality and heterogeneity height and total stem length variables were log10(X + 1) transformed and stem density was (X + 0.5)0.5 transformed.


elevations and water depth at the experimental site

Elevations within the experimental array ranged from 25.6 to 32.0 cm NGVD, with a mean of 29.6 (1.09) cm. For comparison, elevations in a natural marsh located on the east side of the created marsh ranged from 22.6 to 41.2 cm NGVD (data reported in Proffitt et al. 2003 and adapted to NGVD). In the natural marsh Spartina alterniflora dominated, or codominated with S. patens and Distichlis spicata, from the lowest elevation up to 29.9 cm NGVD, while S. patens dominated on a ridge (41.2 cm NGVD) that had formed from storm deposition. Mean (1 SD) elevation in the S. alterniflora-dominated zone of the natural marsh was 26.9 (4.01) cm.

The mean water depth in 2001 was +2.9 cm (minimum and maximum, −32.2 to +26.6 cm) in the experimental area, with positive values indicating standing water. Standing water occurred frequently from March to mid-May, and again from mid-June to early July (major portions of the early and mid growing season). This region of the Gulf of Mexico is microtidal and the marshes are flooded primarily when winds come from the south. The water well in the natural reference marsh to the west of the created marsh indicated a mean water depth of +15.9 cm. However, water levels in the reference marsh were generally 2–5 cm below ground surface during portions of June and July in the peak of the growing season.

senescent centre, edge, and total clone areas

Prior to planting the S. alterniflora clones in 1999, the annual Salicornia bigelovii formed a dominant monoculture over the entire 4200 m2 of the experimental area. Salicornia dominance decreased as the Spartina clones expanded through vegetative growth (Fig. 1a). During the July and August 2001 final sampling, Spartina dominated 1920 m2 and the Salicornia 2280 m2 of the experimental site. By this time all planted clone-pairs had grown together and developed a region of senescence in the contact zone. They therefore had the outward appearance of a single circular or elliptical clone. All planting units at which only one clone had been planted showed central senescence.

Figure 1.

(a) The change in the relative amounts of experimental site area covered by S. alterniflora and Salicornia bigelovii over 42 months. (b) Total numbers of species recruiting inside and outside S. alterniflora clonal patches over this time period.

Spartina clones had overall mean (1 SE) areas of senescent centre, edge and total of 9.16 (0.63), 23.9 (1.00) and 33.1 (1.32) m2, respectively. Area of senescent centre and robust edge differed among clonal genotypes (Table 1). The centroid vector of centre and edge areas also varied with elevation in multivariate tests, although only centre had a significant univariate F-value (Table 1).

Table 1. Spartina area of central senescence and robust edge area mancova with clonal genotype as the main effect and elevation as a covariate. The two area measures were significantly positively correlated (r = 0.312, P < 0.001). H-T F is the Hotelling-Lawley Trace F multivariate test statistic. Error d.f. = 48
Sourced.f.UnivariateMultivariate H-T F
Clone genotype

The area of central senescence was positively correlated (P < 0.05) with clone diameter for clones B, C and D (range of r-values 0.672–0.788), but not for A and E. Only clones C and E had significant correlations between senescent area and elevation but in opposite directions (r = −0.806 clone C, r = 0.856 clone E). Clone C showed the least senescence, 20.0 (2.1)% of total area, while clones E and A, 34.8 (3.9) and 30.1 (1.8)%, respectively, experienced the greatest senescence as a percentage of total clone area.

When combined, Spartina size and density variables showed a multivariate significant difference with both clone and location (centre vs. edge) (Table 2). However, univariate tests indicated that only biomass and mean plant height varied with clone type, while all variables differed with location (Table 2). There were also significant clone–location univariate interactions for these two response variables, but there was not a significant multivariate difference. The elevation covariate was not significant.

Table 2.  Summary of mancova analyses for Spartina alterniflora growth parameters (number of live stems, number of dead stems, mean stem height, total stem length, and biomass) by clonal genotype (A, B, C, D or E) and within-clone location (centre vs. edge). Elevation is used as a covariate. Data were transformed as noted in the text prior to analyses. Degrees of freedom: clone (4), location (1), elevation (1), error (96). The multivariate test for each independent variable is H-T F, the Hotelling-Lawley Trace probability value. Univariate tests are also given as F and P values. Probabilities significant at the 0.05 level are shown in bold
Source Live stemDead stemBiomassMean heightTotal stem length
CloneF 1.180.85 2.84 2.950.95
 H-T F = 0.122P 0.3250.496 0.029 0.0240.438
 H-T F = 0.0005P 0.00050.065 0.0005 0.00050.0005
Clone × LocF 0.590.32 3.08 2.912.42
 H-T F = 0.198P 0.6700.862 0.020 0.0250.053
ElevationF 0.130.03 3.08 3.481.40
 H-T F = 0.537P 0.7250.864 0.082 0.0650.241

soil conditions in clone centres and edges

Percentage soil moisture, pore water salinity, NH4-N and PO4-P did not differ with clone (multivariate test: H-L Trace F16,346 = 0.518, P = 0.938, univariate tests not shown). The multivariate centroid of the soil parameters did not vary at the 0.05 alpha level between clone centre and edge (multivariate test: H-L Trace F4,88 = 2.23, P = 0.072), although percentage soil moisture and PO4-P did differ among locations in univariate tests (F1,91 = 4.43, P < 0.038 and F1,91 = 6.84, P < 0.010, respectively). All variables were significantly different in April and July (multivariate test: H-L Trace F4,88 = 18.17, P = 0.0005, univariate tests not shown). None of the two- or three-way interaction terms were significant and none of the variables covaried with elevation.

light penetration of spartina clone canopies

Spartina growth and size variables interacted to affect light penetration: Fig. 2 shows light penetration of the canopy decreasing as combinations of density and height increase. Multiple regression indicated that 83% of the variation in light penetration of the Spartina canopy could be explained by these two growth variables (Fig. 2).

Figure 2.

Light penetration is shown as a combined function of S. alterniflora height and density for clone centres (triangles) and edges (open squares). Symbol size is scaled to light penetration (larger symbols indicate greater light penetration of the canopy). Gaussian bivariate combinations of X and Y points centred on the centroid of the X, Y combinations form ellipses with the major axis determined by the sample standard deviations and the orientation being a function of the covariance of X and Y. The solid-line ellipse is for clone centres and the dashed ellipse is for clone edges. Multiple regression of percentage light penetration on height and density is: (R2 = 0.831, P < 0.0005): Lightpen = −0.852 (Ht) –0.174 (dens) + 109.805.

Percentage light penetration of the Spartina canopy differed significantly between within-clone locations and among genotypes, and there was a significant interaction between these main effects (Table 3). However, the light penetration did not vary with elevation as would have been expected if there were a strong relationship between elevation and plant architecture. Clone edges allowed less than half the light penetration recorded in clone centres (Table 3) and there was nearly a 2× difference among genotypes in penetration into areas of robust growth, with clones A and B allowing substantially less light through than clones C, D and E (Table 3).

Table 3.  Two-way ancova of the percentage light penetrating the S. alterniflora canopy in the experimental units. Independent variables are clone genotype (A-E) and location (centre or edge). Elevation was a covariate in the analysis. d.f. = degrees of freedom. Means and standard error (SE) values are shown below the analyses. Values significant at the 0.05 level are shown in bold
Clone 4 3.74 0.007
Location 154.96 0.0005
Clone × Location 4 3.07 0.020
Elevation 1 1.63 0.205
Mean (1 SE)
A42.3 (4.28)14.7 (4.28)28.5 (3.04)
B45.1 (4.14)15.5 (4.14)30.3 (2.93)
C39.8 (5.00)27.5 (5.00)33.6 (3.55)
D40.3 (4.78)29.4 (4.78)34.8 (3.38)
E77.6 (8.31)28.8 (8.31)53.2 (5.90)
Combined49.0 (2.47)23.2 (2.47) 

interactions between spartina and high marsh plant species

In addition to the pre-study dominant, Salicornia bigelovii, and the experimentally planted Spartina alterniflora, nine plant species colonized the experimental area naturally between 1999 and 2001. Most of these colonized between the 2000 and 2001 growing seasons (Fig. 1b). The annuals Aster subulatus and Atriplex patula were the most abundant colonizers (Table 4), with the woody groundsel or salt bush Baccharis halimifolia L. the only relatively common perennial species. Although the total areas dominated by Spartina and Salicornia were similar in 2001, nearly twice the number of species colonized into Spartina patches (Table 4). This suggests that recruitment was not random and that differences in numbers of species are a function of well-established species-area relationships and species interactions. Equal sampling areas showed that 94.8% of the Spartina clones were colonized by other plant species, whereas only 34% of sampled sites just outside the clones were colonized. Also, twice as many total species colonized the senescent centres of Spartina clones compared with clone edges, with centres much more frequently colonized by the three most abundant species (Aster, Atriplex and Baccharis), than edges (Table 4).

Table 4.  The percentage occurrence of colonizing plant species in Spartina alterniflora clones or in Salicornia bigelovii-dominated areas immediately outside the clones
 Within Spartina clonesIn Salicornia areas
Aster subulatus94.862.158.6
Atriplex patula34.525.922.4
Baccharis halimifolia24.1 1.7 5.2
Borrichia frutescens 5.2 1.7 0
Aster tenuifolius 8.6 0 0
Pluchea camphorata 5.2 0 1.7
Iva frutescens 1.7 0 0
Limonium carolinianum 1.7 0 0
Distichlis spicata 0 0 1.7

The mean numbers of new colonizing species m−2 (i.e. excluding Spartina and Salicornia) was significantly greater in clone centres (1.9 (0.13) species) than in either clone edges (0.92 (0.13) species) or outside clones (0.84 (0.13) species), but did not vary by clonal genotype or elevation (two-way anovaR2 = 0.287, location F2,158 = 23.56, P < 0.0005; clone F4,158 = 0.85, P = 0.493; location–clone interaction F8,158 = 1.03, P = 0.419; elevation coavriate F1,158 = 0.90, P = 0.343). Although the largest difference in number of species m−2 was small (c. 1 species m−2), the species concerned differed and the cumulative result over the entire experimental area was therefore that there was a twofold difference between Spartina and non-Spartina sites (Fig. 1b).

Spartina suppressed the annual dominant species, Salicornia bigelovii, but served to facilitate two of the more abundant annual colonizers, Aster subulatus and Atriplex patula (Tables 5–7). Only Atriplex varied significantly along the elevation gradient, with greater densities and taller plants at higher elevations (Table 7).

Table 5.  Two-way ancova for Salicornia bigelovii size and density variables. Independent variables are Spartina clone genotype (A-E) and location (centre, edge or outside). Elevation was a covariate in the analysis. d.f. = degrees of freedom. Means and standard error (SE) values are shown below the analyses. Values significant at the 0.05 level are shown in bold
A. Log10 (mean height). R2 = 0.595
Clone4 6.030.0005
Clone × Location8 6.480.0005
Elevation1 0.840.361
Means (1 SE)
A0.621 (0.095) 1.479 (0.095)1.274 (0.095)
B0.999 (0.092) 1.441 (0.092)1.271 (0.092)
C0.002 (0.106) 1.455 (0.106)1.271 (0.106)
D0.009 (0.179) 1.579 (0.179)1.058 (0.179)
E0.012 (0.184) 1.358 (0.184)1.257 (0.184)
B. Log10 (total stem length). R2 = 0.330
Clone4 2.440.049
Clone × Location8 2.220.028
Elevation1 0.0850.771
Means (1 SE)
A0.625 (0.222) 1.874 (0.222)1.496 (0.222)
B1.041 (0.215) 1.810 (0.215)1.540 (0.215)
C0.979 (0.259) 1.215 (0.259)1.095 (0.259)
D0.001 (0.248) 1.977 (0.248)1.635 (0.248)
E0.009 (0.431) 0.419 (0.431)1.585 (0.431)
C. (No. stems m−2)−0.5. R2 = 0.233
Clone4 0.350.847
Clone × Location8 2.680.009
Elevation1 2.300.131
Means (1 SE)
A1.089 (0.603) 3.443 (0.603)2.314 (0.603)
B2.202 (0.583) 2.182 (0.583)2.823 (0.583)
C2.196 (0.705) 2.906 (0.705)3.412 (0.705)
D0.017 (0.674) 2.256 (0.674)4.926 (0.674)
E0.130 (1.170) 1.708 (1.170)4.593 (1.170)
Table 6.  Two-way ancova for Aster subulatus size and density variables. Independent variables are Spartina clone genotype (A-E) and location (centre, edge or outside). Elevation was a covariate in the analysis. d.f. = degrees of freedom. Means and standard error (SE) values are shown below the analyses. Values significant at the 0.05 level are shown in bold
A. Log10 (mean height). R2 = 0.368
Clone4 2.420.051
Clone × Location8 0.580.798
Elevation1 0.190.664
Means (1 SE)
A1.898 (0.166) 1.219 (0.172)0.741 (0.172)
B1.729 (0.166) 1.150 (0.166)0.997 (0.166)
C1.457 (0.204) 0.607 (0.204)0.626 (0.204)
D1.903 (0.194) 0.872 (0.194)0.665 (0.194)
E1.987 (0.372) 1.391 (0.372)0.899 (0.372)
B. Log10 (total stem length). R2 = 0.384
Clone4 2.550.042
Clone × Location8 0.3460.947
Elevation1 0.1170.773
Means (1 SE)
A6.290 (0.730) 2.720 (0.756)1.078 (0.756)
B6.011 (0.730) 2.440 (0.730)2.438 (0.730)
C4.447 (0.894) 0.140 (0.894)0.563 (0.894)
D5.652 (0.852) 1.278 (0.852)0.691 (0.852)
E5.802 (1.633) 2.621 (1.633)2.068 (1.633)
C. (No. stems m−2)−0.5. R2 = 0.174
Clone4 2.360.053
Location2 2.540.083
Clone × Location8 1.220.294
Elevation1 0.4660.496
Means (1 SE)
A0.95 (0.183) 1.15 (0.189)0.62 (0.189)
B1.52 (0.183) 0.94 (0.183)0.83 (0.183)
C1.20 (0.224) 0.39 (0.224)0.52 (0.224)
D0.78 (0.213) 0.57 (0.213)0.59 (0.213)
E0.78 (0.409) 0.87 (0.409)0.92 (0.409)
Table 7.  Two-way ancova for Atriplex patula size and density variables. Independent variables are Spartina clone genotype (A-E) and location (centre, edge or outside). Elevation was a covariate in the analysis. d.f. = degrees of freedom. Means and standard error (SE) values are shown below the analyses. Values significant at the 0.05 level are shown in bold
A. Log10 (mean height). R2 = 0.141
Clone × Location80.800.603
Means (1 SE)
A0.839 (0.181)0.528 (0.187)0.469 (0.187)
B0.703 (0.181)0.286 (0.181)0.215 (0.181)
C0.409 (0.222)0.194 (0.222)0.266 (0.222)
D0.187 (0.211)0.415 (0.211)0.239 (0.211)
E1.275 (0.405)0.527 (0.405)0.034 (0.405)
B. Log10 (total stem length). R2 = 0.143
Clone × Location80.7220.672
Means (1 SE)
A0.929 (0.193)0.545 (0.200)0.508 (0.200)
B0.722 (0.193)0.286 (0.193)0.231 (0.193)
C0.499 (0.236)0.195 (0.236)0.267 (0.236)
D0.187 (0.225)0.415 (0.225)0.239 (0.225)
E1.277 (0.432)0.685 (0.432)0.036 (0.432)
C. (No. stems m−25)−0.5. R2 = 0.103
Clone × Location81.030.414
Means (1 SE)
A0.17 (0.091)0.25 (0.095)0.30 (0.095)
B0.18 (0.091)0.11 (0.091)0.15 (0.091)
C0.16 (0.112)0.09 (0.112)0.12 (0.112)
D0.03 (0.107)0.39 (0.107)0.11 (0.107)

Salicornia was shorter and had greatly reduced stem densities and total stem lengths in Spartina clone centres, although heights in clone edges were greater, possibly as a response to shading by the expanding Spartina (Table 5). Salicornia height and total stem length also varied with Spartina genotype and both these variables and stem density had significant clone–location interactions (Table 5).

Aster had much higher mean heights and total stem lengths within Spartina clones compared with outside clones (Table 6), and all three response variables differed significantly among Spartina genotypes. Aster mean height and total stem length were greatest in Spartina clone centres, but stem density did not quite vary significantly with location (P = 0.083, Table 6).

Atriplex was taller and had the greatest total stem length within Spartina clones (Table 7), but did not differ among Spartina clone genotypes. Atriplex mean height, total stem length and density increased with increasing elevation (Table 7).


Biotic interactions such as competition and facilitation play an important role in determining the mixture of species, diversity and dominance in ecological systems (Connell & Slatyer 1977; Chesson & Huntly 1997; Day & Young 2004; Pennings et al. 2005). In many systems, such as forests, grasslands and coral reefs, some species may exert strong influences on others by imposing physical structure and new suites of environmental conditions that do not exist in their absence (Bruno & Bertness 2001). We suggest that in such foundation or matrix or keystone species, differences in morphology or growth among clonal genotypes will increase the type and/or magnitude of effects on other species.

The presence or absence of the grass Spartina alterniflora has strong influences on other species in a young, developing marsh. The prior dominant species, the annual Salicornia bigelovii, is displaced by S. alterniflora, probably due to reduction in light penetration of the grass canopy, and is possibly further inhibited by reduction in water-borne seed transport to clone centres because of blockage by the dense clone edges. However, a number of other high marsh species recruited better into S. alterniflora experimental clones than non-grassed areas. Similar interactions among salt marsh plants have been reported in a number of studies. For example, Bertness & Ellison (1987) found that interspecific competition and disturbance were primary agents structuring a salt marsh community. Bertness & Ewanchuk (2002) reported greater survival and growth of certain fugitive salt marsh species when growing with dominant or matrix plant species than when grown in non-vegetated surroundings. Levine et al. (1998a) reported that nutrients could reverse competitive outcomes among salt marsh plants. Huckle et al. (2002) found that competitive interactions between salt marsh plant species were primarily above ground rather than below ground, and that competition was reduced under conditions of increased environmental stress.

Our study is the first to document differences in species–species plant interactions due to varying effects of different genotypes of S. alterniflora, both in terms of numbers of species recruiting and size and density of some of the more abundant species. Different shade regimes may be one of the mechanisms behind the biological effects of the genotypes, and contribute to the patchiness often observed for some less abundant or fugitive salt marsh species, but soil moisture, salinity and nutrients were unaffected. Thus, our findings agree with Seliskar et al. (2002), who reported that different genotypes of S. alterniflora produced changes in the marsh environment and exerted strong control on certain ecological functions.

The extent and ecological importance of these types of effects may be related to genotypic and possibly genetic diversity of the population of the dominant species. Genetic diversity tends to be relatively high in this species (Travis et al. 2002) and the diversity of clonal genets first increases then decreases with marsh age (Travis et al. 2004). However, despite some reduction, considerable genetic and genotypic diversity remains in populations over 1000 years old (Travis & Hester 2005). Genotypic variation in S. alterniflora results in variation in stem density, height and clonal morphology (Proffitt et al. 2003, present study). The high clonal and genetic diversities in S. alterniflora populations of the northern Gulf of Mexico marshes are maintained by high levels of outcrossing coupled with severe inbreeding depression (Travis et al. 2004). Clones in natural marshes of this region range considerably in size (maximum diameter < 20 to > 500 m) depending in part on marsh age (Travis & Hester 2005). This suggests that many marshes will consist of populations of many more different clonal genotypes than the five tested here, and their substantial variation in morphology, size, onset of senescence, etc., will greatly increase the potential for genotypic and genotype × environment effects in S. alterniflora-dominated marshes.

Here, we show that genet-specific differences in S. alterniflora plant size, density and degree of senescence of clone centres introduce heterogeneity that results in increases in the rate recruitment of a number of high marsh plant species at mid-tidal elevations and in size and/or density of the fugitive annual plant species Aster subulatus, as well as affecting replacement of Salicornia bigelovii. Some authors have suggested that higher order properties arising from such interactions may themselves be heritable, thus constituting a discipline of community genetics (Antonovics 1992; Whitham et al. 2003; but see Ricklefs 2003). If true, then genetic and genotypic variation in dominant species would have even greater, higher order, consequences for marsh ecology.

Senescence in the centre of S. alterniflora clones may be a function of age-related ramet mortality but, unlike some plant species (Bell 1974; Harper 1977), it is probably not related to directional growth of rhizomes (Metcalfe et al. 1986). Some grass patches senesce in the centre because they have vertical orientation of crowns and tend to grow themselves out of the ground (Dahl 1995); however, this is not known to be the case for S. alterniflora. Nutrient values measured in clones were similar to those in natural marshes in the area (Edwards & Proffitt 2003; K. R. Edwards & K. Mills, unpublished data). Also, fertilization of clone centres and edges produced no response in this marsh (A. B. Owens & C. E. Proffitt, unpublished data). This suggests that nutrient depletion is unlikely to be responsible for central senescence of the S. alterniflora clones or the different effects of its genets on other plant species. However, manipulative experiments on nutrients and senescence need to be done to support this contention.

Falinska (1995) reported that senescence of the upland plant Filipendula ulmaria influenced succession by giving rise to open space that was subsequently colonized by other plants. McConnaughay & Bazzaz (1990) found that there was greater plant performance in larger gaps and that plant–plant interactions were independent of gap size in a herbaceous old field community. Their work showed the species-specific nature of interactions, as indicated here by competition and facilitation among plants colonizing the senescent centres of S. alterniflora.

The patterns of colonization and seedling growth noted here are important in the development of community structure at mid and higher elevations. Several of the more abundant colonizers, such as Atriplex, Aster and Baccharis, have seeds that are adapted for wind dispersal, although in these tidal marshes they are probably also secondarily moved by water (Huiskes et al. 1995). Iva and Salicornia produce large numbers of very small seeds that are water dispersed. The substantially larger caryopsis of S. alterniflora is also dispersed by water movement. Dispersal of Salicornia europaea (an annual prevalent in the north-eastern United States) appears to be generally limited to a few metres (Ellison 1987) and, at our site, S. bigelovii seeds occurred on the soil surface mainly within a metre or two of parent plants. However, longer distance dispersal of seeds and/or seedlings clearly occurs because three to five successive years of seed set and colonization leads to a dense population of S. bigelovii across essentially the entire area of the sites (Proffitt & Young 1999; C. E. Proffitt, field observations). Whether the S. alterniflora clonal genotypes have direct effects on the other plant species or mediate plant responses through effects on microenvironmental parameters, or both, is not discernable from our study. Nor can we tell whether colonization per se is different in different S. alterniflora clonal types, or whether seedling survival and growth are different, or both. However, Egerova et al. (2003), working in this same marsh, did not find that S. alterniflora clones trapped greater numbers of seeds of Baccharis, but that they did enhance survival and growth of that woody high-marsh species.

Huckle et al. (2002) reported that Spartina anglica facilitated the growth of Puccinellia maritima at low nutrient concentrations, and that P. maritima outcompeted S. anglica. Bertness & Hacker (1994) and Bertness & Ewanchuk (2002) reported that in years with higher soil salinities, interactions among plant species tended to be facilitative because some species could ameliorate soil conditions that could adversely affect more sensitive plant species. However, Levine et al. (1998b) noted that for one salt-intolerant marsh species this effect was eliminated by fertilization. We found similar reductions in temperature to those reported by Bertness & Ewanchuk (2002) but, unlike Bertness & Ewanchuk (2002) and Bertness & Hacker (1994), we did not find significant effects of higher vegetation cover on soil pore water salinity. It is possible that, as the assemblage is developing on a newly deposited mud flat, nutrients are not limiting and consequently that the effects of salinity are somewhat reduced.

As noted, Spartina alterniflora clonal genotypes played a role in the extent of suppression of the annual dominant Salicornia bigelovii. Despite higher light levels, the very low densities and small plant sizes of S. bigelovii indicate that senescent clone centres did not serve as effective refugia for this species. In New England, Salicornia europaea, which occurs in non-grassed patches produced through wrack disturbance, is similarly competitively inferior to the marsh grasses (Bertness & Ellison 1987; Ellison 1987), and is suppressed by other marsh matrix species (Bertness & Ewanchuk 2002). However, in a cobble beach system, Bruno (2000) found that the presence of S. alterniflora facilitated establishment of S. europaea immediately landward by reducing water velocity and thus stabilizing the substrate. S. bigelovii remained a dominant in areas throughout our 200-ha created marsh site and was especially tall and abundant in interior, higher elevation regions. However, S. bigelovii, coming up new from seed each growing season, suffered greatly reduced densities after encountering expanding S. alterniflora clones. Boyer & Zedler (1999) found different outcomes when S. bigelovii was grown in competition with Spartina foliosa in a fertilizer experiment. Their results showed that S. foliosa, which is much shorter than S. alterniflora, only responded with increased growth to nutrient addition in the absence of S. bigelovii, while S. bigelovi grew to the size of S. foliosa when fertilized and produced > 1 × 106 seeds m−2. Salicornia virginica has also been reported to out-compete S. foliosa for nitrogen in California (Covin & Zedler 1988). Such experiments have not yet been done in our young, clay-soil marshes, although available data suggest that nutrients are not generally limiting (Edwards & Proffitt 2003; K. R. Edwards & K. Mills, unpublished data).

Bertness & Ewanchuk (2002) found more facilitative species–species interactions in New England salt marshes south of Cape Cod and more competitive interactions northwards. This was linked to warmer conditions in the more southern sites and in both sites in warm years. Our study included some of the same marsh genera and species as Bertness & Ewanchuk (2002) in a much more southerly locale (coastal Louisiana). Similar to their results for S. europaea, we found that S. alterniflora suppressed Salicornia bigelovii, but facilitated colonization and/or growth of a number of other species.

Numerous species are colonizing the large, initially vacant, mudflats created from dredged sediment at the Sabine National Wildlife Refuge (Proffitt & Young 1999; Edwards & Proffitt 2003; Egerova et al. 2003; Proffitt et al. 2003). However, the development of the dominant perennial S. alterniflora is an important agent of other changes within the vegetation assemblage. S. alterniflora colonizes essentially the entire elevation range of these sites when not in competition with other perennials (Proffitt & Young 1999; Proffitt et al. 2003) and enhances the recruitment and growth of a number of annual and woody plant species (Egerova et al. 2003; present study). A recent model suggests that S. alterniflora will become dominant over a large elevation range when nutrients are not limiting (Bertness 1991a,b; Huckle et al. 2002). Our results on the interactions between S. alterniflora and S. bigelovii tend to support this model although we found a substantial lag time because of the time necessary for S. alterniflora colonization and clonal growth in this young site (Proffitt et al. 2003; present study). However, our results also show that differences in clonal genets and central senescence lead to patchiness that can promote the recruitment and growth of some plant species. Whether or not long-term changes will result in the loss of some of these colonizing species through interspecific competition with the dominant grass will depend on the long-term clonal population dynamics of existing S. alterniflora clones and on the colonization of senescent centres by S. alterniflora seedlings and the ensuing establishment of new clones (Travis et al. 2004).

Spartina alterniflora has been described as a keystone species (Seliskar et al. 2002), a keystone or ecosystem engineer (Bruno 2000), and a foundation species (Pennings & Bertness 2001), because of the profound roles that it plays in creating habitat for other species, providing considerable productivity for marsh and nearshore food webs, buffering storm surges, and sequestering organic matter in the sediment. Despite decades of ecological studies in salt marshes, the importance of population biology, genetics and morphological variation for ecological function has only recently begun to be established (Seliskar et al. 2002; Proffitt et al. 2003; Travis et al. 2004; present study). Not only is an adequate understanding of the effects of genotypic and genetic variation on ecologically important plant traits critical to a full understanding of basic salt marsh ecology, it may also be fundamental to the conservation and restoration of fully functional populations of this dominant plant species.

Crane et al. (2004) stated that plants are not just ‘found in habitats where they grow best’, but rather that biotic interactions in association with environmental factors are important in producing plant assemblages. We suggest that populations of foundation species, such as S. alterniflora, that have greater clonal diversity will also have a broader array of ecologically important interactions with other plant species. This certainly appears to be the case in young marshes developing on mudflats. Further work will be required to determine if clonal diversity is important in biological interactions in very old marshes.


K. Mills, J. Kemmerer and J. Egerova assisted in the field. Statistical advice was provided by D. Johnson, S. Mopper, P. Leberg and K. Krauss. K. Edwards was supported by a grant from the USEPA. We thank M. D. Bertness, J. L. Gallagher, S. Pennings, and D. M. Seliskar for visiting the site with us and providing helpful suggestions and comments. We appreciate helpful comments on the paper by J. B. Grace, D. J Devlin, B. Middleton, K. L. McKee and B. Vairin. A. Anteau and K. McKee performed the laboratory analyses for nutrients. Primary financial and logistical support and access to the study site was graciously supplied by C. Pease and his staff of the Sabine National Wildlife Refuge (USFWS).