*Present address and correspondence: Steven C. Pennings, Department of Biology and Biochemistry, University of Houston, Houston, TX 77204, USA (tel. +1 713 7432989; fax +1 713 7432636; e-mail email@example.com).
1We investigated the factors producing zonation patterns of the dominant plants in south-eastern USA salt marshes where Juncus roemerianus dominates the high marsh, and Spartina alterniflora the middle and low marsh.
2Juncus did not occur naturally in the Spartina zone and performed poorly when transplanted there, irrespective of whether neighbours were present or removed, indicating that its lower limit was set by physical stress.
3In contrast, although Spartina occurred naturally at low densities in the Juncus zone, it performed well if transplanted there only if neighbours were removed, indicating that its upper limit was set by competition.
4Parallel laboratory and field manipulations of flooding, salinity and competition indicated that the lower limit of Juncus was mediated by both flooding and salinity, but not by competition.
5The general mechanisms producing zonation patterns of vegetation in coastal salt marshes may be universal, as suggested by previous studies, but the importance of particular factors is likely to vary geographically. In particular, salinity stress probably plays a much more important role in mediating plant zonation patterns at lower latitudes.
6Our results suggest that the nature of ecological interactions is likely to vary geographically because of variation in the physical environment, and this variation must be taken into account in order to successfully generalize the results of field studies across geographical scales.
Salt-marsh plant communities are characterized by striking zonation patterns across elevational gradients (Chapman 1974). Over recent decades, a series of experimental studies (Snow & Vince 1984; Bertness & Ellison 1987; Bertness 1991a,b; Pennings & Callaway 1992) have led to an emerging paradigm about the forces that mediate these patterns. According to this paradigm, there is an inverse relationship between competitive ability and stress tolerance, such that competitively superior plants occupy the least stressful zones of the salt marsh and displace competitively inferior plants to more stressful zones (Bertness 1992; Pennings & Bertness 2001). This paradigm about zonation patterns in salt marshes is consistent with results from non-saline wetlands (Grace & Wetzel 1981; Grace 1989; Keddy 1989) and with general ideas about inherent trade-offs in plants between competitive ability and stress tolerance (Grime 1977; Grime 1988). In marshes that are irregularly flooded, however, this paradigm may not apply because there may not be a consistent gradient in physical stress across the marsh (Costa et al. 2003).
Disentangling the role of flooding and salinity in producing zonation patterns is particularly important in low-latitude habitats. In New England, where most experimental studies of plant zonation in salt marshes have been conducted (Bertness & Ellison 1987; Bertness 1991a,b, 1992; Bertness et al. 1992), both flooding and salinity increase in severity from the marsh upland border to the water's edge (Pennings & Bertness 1999, 2001). Because most marsh plants can tolerate salinities typical of undisturbed New England marshes, previous studies concluded, albeit without explicit tests, that flooding was the primary stress mediating zonation patterns (Bertness 1991b; Bertness et al. 1992). In contrast, in hotter, low-latitude marshes on the east and west coasts of the United States, salinities reach a peak in mid marsh zones (Pennings & Bertness 1999, 2001); consequently, salinity and flooding gradients are not parallel, and workers have speculated that both stresses could play an important role in mediating plant zonation patterns (Pennings & Callaway 1992). Thus, a more general understanding of salt marsh plant zonation patterns may require a better understanding of geographical variation in the roles of different physical factors.
Here, we focus on the zonation of Spartina alterniflora Loisel and Juncus roemerianus Scheele (henceforth referred to as Spartina and Juncus) in salt marshes in Georgia, USA. These two plant species comprise the vast majority of the plant biomass in south-eastern USA salt marshes along both the Atlantic and Gulf coasts (Eleuterius 1976a; Stout 1984; Wiegert & Freeman 1990). Interactions between these two species have been experimentally examined in only one study (Stanton 1998). Along the southern Atlantic Coast, Spartina typically occupies salty soils at lower marsh elevations, and Juncus less salty soils at higher elevations (Wiegert & Freeman 1990), although considerable overlap occurs in these environmental variables (Woerner & Hackney 1997). We asked three questions using a combination of field and glasshouse studies. First, what is the role of competition and physical stress in creating this zonation pattern? Secondly, what is the relative importance of salinity and flooding in excluding Juncus from the Spartina zone? Thirdly, do field and glasshouse studies provide similar insights into the nature of these interactions?
study site and zonation patterns
Fieldwork was conducted at Sapelo Island, GA (31°27′ N, 81°16′ W). Plant zonation patterns in salt marshes around Sapelo Island are typical of coastal salt marshes throughout the South Atlantic Bight (Wiegert & Freeman 1990). We worked at three sites on the west side of Sapelo Island (north to south: Keenan Field, Airport Marsh, Marsh Landing) in locations where the Juncus and Spartina zones directly abutted each other, because this zonation pattern is the most typical of the region. At some sites with extremely high soil salinities, this typical zonation pattern is interrupted by unvegetated salt pans and associated ‘salt meadow’ communities comprised of highly salt-tolerant plants (Wiegert & Freeman 1990); we did not consider these more complicated zonation patterns in this study.
Spartina alterniflora is a C4 grass that grows as upright shoots connected by underground rhizomes. It is highly tolerant of flooding and anoxic soils, and moderately tolerant of high salinities (Mendelssohn & Morris 2000). Juncus roemerianus is a C3 rush that grows as horizontal shoots with upright leaves (Eleuterius 1976b). It is thought to be less tolerant of flooding and salinity than Spartina (Eleuterius 1976b; Wiegert & Freeman 1990).
To document a typical zonation pattern of Spartina and Juncus, 1 × 1 m quadrats were centred 1, 2, 4 and 6 m on each side of the border between the two species at Keenan Field (n = 8 transects, ≥ 5 m between transects) in October 1996. We counted the number of shoots (Spartina) or leaves (Juncus) of each species in each quadrat. Data (means ± SE) are presented for visual inspection of the patterns without formal statistical analysis.
Edge removal experiment
We established 30 0.5 × 0.5 m plots on each side of the border between Juncus and Spartina at Airport Marsh in March 1994. Plots were located with one side along the border between the two species. Measurements of each species were made in three conditions: in its own zone, in the other species’ zone and in the other species’ zone with the other species removed to assess the impact of competition (n = 10 individuals treatment−1). In the removal treatment, we removed the zonal dominant by clipping at the soil surface bi-weekly. Because the border was not perfectly abrupt, both species were initially present in the other species’ zones. These individuals were not removed in either the control or ‘zonal dominant removal’ treatments. Treatments were fully interspersed and initial conditions were similar across all replicates of all treatments within each zone. The central 0.25 × 0.25 m of each plot was harvested in November 1995, and all above-ground live plant material was dried for 3 days at 60 °C and weighed. In this, and all the following experiments, we focused on above-ground biomass because it was very difficult to excavate completely the below-ground portions of these highly clonal plants and to separate accurately the roots from the soils. For similar reasons, most previous studies of salt-marsh plant zonation patterns have also focused primarily on above-ground biomass (Vince & Snow 1984; Scholten & Rozema 1990; Bertness 1991a,b; Pennings & Callaway 1992; Huckle et al. 2000). Data on above-ground biomass for each species were compared among treatments with a one-way anova.
Individual Spartina and Juncus culms with associated soil blocks (20 × 20 × 20 cm) were transplanted into their own and the other species’ zone at Airport Marsh in April 1994. We removed the vegetation surrounding half of the transplants into the other species’ zone by clipping a 0.25-m radius border around the plant at the soil surface. Clipping treatments were maintained by repeated clipping every 2 weeks as needed. Unmanipulated culms were tagged and left as transplant controls. All treatments were replicated eight times. Measurements with electronic surveying equipment documented that transplanted plants in the Spartina zone were located 190 m seawards from the transplanted plants in the Juncus zone, and 6 cm lower in elevation. To document edaphic patterns in both zones we measured soil organic content in September 1995 by ashing at 450 °C (n = 12 zone−1), and soil pore water salinity in July and October 1994 and September 1995 by rehydrating dried soil samples in a known volume of distilled water, mixing thoroughly, measuring the salinity of the supernatant after 48 hours, and calculating the salinity of the original pore water based on the original gravimetric water content of the individual soil samples (n = 8–12 zone−1 date−1). Water content of the soil (grand mean 62.2%) varied among dates with a significant date × treatment interaction (data not shown). Above-ground biomass of surviving plants was estimated in October 1994 by summing the length of all the shoots (Spartina) or leaves (Juncus) within the transplant block. For these species, the total length of shoots or leaves, respectively, correlates highly with biomass (Pennings & Callaway 2000). Five Spartina plants that gradually shrank in size and then died (presumably due to competition) were included in the analysis with total shoot length set to zero. Three Spartina plants that were eaten by deer were excluded from the analysis. Proportional data (organic content) were arcsine (square root) transformed before analysis. Edaphic data (organic content, salinity) were compared among zones, or zones and dates, using one- or two-way anova, respectively. Above-ground size (total shoot length) of plants of each species was compared among treatments using a one-way anova.
mediation of competition by edaphic conditions
In order to examine the roles of flooding, salinity and competition on performance of Juncus and Spartina, we grew the two species alone and together in the glasshouse under a variety of edaphic conditions. Plants were collected in May 1995 from an area of Keenan Field where the two species grew highly intermingled. Blocks of soil (20 × 20 × 20 cm) containing one or both species were excavated from within a small (10 × 25 m) collection area that appeared to have homogenous soil conditions throughout. Plants were thinned by clipping at the soil surface to four to five shoots of Spartina and/or eight to twelve leaves of Juncus and potted in 20-L (29 cm wide × 35 cm high) pots lacking drainage holes, using additional soil collected from a single marsh site. Drained and flooded treatments were achieved by drilling small holes in the sides of the pots at the soil surface (flooded) or 10 cm below the soil surface (drained), representative of conditions commonly observed in the Spartina and Juncus zones, respectively. After filling with water to 5 cm above the soil surface, pots took c. 1 hour to drain to the level of the drain holes. Plants were watered three times each week with fresh water or seawater as needed to maintain flooding and salinity treatments. Our aim was to achieve salinities of approximately 10 and 40 p.p.t., representative of conditions commonly encountered in the Juncus and Spartina zones, respectively (Wiegert & Freeman 1990). Salinities around representative plants from each treatment were checked on every watering date to ensure that salinity treatments were being appropriately maintained. The two flooding treatments were crossed with the two salinity treatments and the two competition treatments (alone, with the other species), for a total of eight treatments, each replicated 10 times for each species. Pots were located outdoors under a plastic roof to shelter them from rain, but were otherwise exposed to ambient temperature, humidity and light. All live above-ground biomass was harvested from each pot in November 1995, dried for 3 days at 60 °C and weighed. Plants did not appear to be ‘pot-bound’. Data for each species were analysed with three-way anova, with flooding, salinity and competition as the three main effects.
In order to further examine the roles of flooding, salinity and competition on the performance of Juncus in the field, we transplanted plants into the Spartina zone in a fully factorial experiment that manipulated flooding, salinity and competition. In April 1997 we collected culms of Juncus and transplanted them into the middle of the Spartina zone at Marsh Landing. Single culms were planted inside 20 cm diameter pvc pipe sections. For the elevated treatment, the pipe sections were 12 cm long, and were pressed 4 cm into the soil, so that the plants were elevated 8 cm above ambient. For the not-elevated treatment, plants were planted in a 4 cm long section of pipe that was fully pressed into the soil surface so that plants were level with the ambient soil. In order to manipulate competition, we either left neighbouring vegetation intact (competition) or clipped neighbouring Spartina plants, both inside and around the pipe, at the soil surface within a 50-cm radius of the transplant (no-competition). In order to manipulate salinity, plants were either unmanipulated (salt) or were watered twice a week throughout the entire duration of the experiment with fresh water (fresh), which reduced salinity by 5–20 p.p.t. (effectiveness of watering varied with climate and the lunar component of the tidal cycle) without significantly affecting soil water content. Each treatment combination was replicated 15 times. All above-ground biomass within each pvc pipe was harvested in August 1998, dried for 3 days at 60 °C and weighed. Data were analysed with three-way anova, with flooding, salinity and competition as the three main effects.
The zonation pattern we documented at Keenan Field was typical of what we have observed at other sites within Georgia (S. C. Pennings, personal observations). There was an abrupt transition over 1–2 horizontal metres between dominance by Juncus and dominance by Spartina (Fig. 1). Spartina was present in the Juncus zone, but only at 10–20% of the density that it attained in its own zone. Juncus did not occur at all in the Spartina zone.
Edge removal experiment
Above-ground biomass of Spartina on its own side of the border was four times greater than just across the border in the Juncus zone (Fig. 2, anova, F2,27 = 9.72, P = 0.0007). When Juncus was removed, however, biomass of Spartina increased sixfold in the Juncus zone. The absolute above-ground biomass of Spartina in these treatments was low compared with values typically reported in the literature because the experiment was conducted in an area of very short ‘short-form’Spartina. Above-ground biomass of Juncus on its own side of the border was three times greater than just across the border in the Spartina zone (Fig. 2, anova, F2,27 = 18.69, P < 0.0001). Removing Spartina, however, had no effect on Juncus biomass in the Spartina zone.
Soil organic content was higher in the Spartina vs. the Juncus zone (19 ± 1% vs. 10 ± 1, anova, F1,22 = 24.75, P = 0.0001). Soil pore water was sometimes hypersaline in the Spartina zone, and consistently saltier than in the Juncus zone, where the maximum recorded salinities were only marginally hypersaline. Although the pattern of higher salinities in the Spartina zone was consistent, salinity values and the magnitude of the difference between zones varied among dates (Spartina zone: July 1994, 67 ± 2 p.p.t.; October 1994, 31 ± 1; September 1995, 39 ± 1; Juncus zone: July 1994, 38 ± 2%; October 1994, 23 ± 1%; September 1995, 30 ± 1%; anova: date, F2,54 = 159.16, P < 0.0001; zone, F1,54 = 171.57, P < 0.0001; date × zone, F2,54 = 35.51, P < 0.0001). Similar organic content and salinity differences between the Spartina and Juncus zones are consistently found in marshes around Sapelo Island (Antlfinger & Dunn 1979; Wiegert & Freeman 1990; Pennings et al. 2003).
Transplanted Spartina plants performed well in their own zone and in the Juncus zone if Juncus neighbours were removed, but performed poorly in the Juncus zone if neighbours were present (Fig. 3a, anova, F3,25 = 17.40, P < 0.0001). Transplanted Juncus plants performed well in their own zone but tended to perform less well in the Spartina zone, especially if Spartina neighbours were removed (Fig. 3b, anova, F3,28 = 3.59, P = 0.03).
mediation of competition by edaphic conditions
In the glasshouse, biomass of Spartina was significantly reduced by both increased flooding and increased salinity (Fig. 4a, Table 1a). A significant three-way interaction between flooding, salinity and competition, and associated significant two-way interactions between flooding and salinity, and between flooding and competition (Table 1a), indicated different effects of Juncus on Spartina in the two low-salinity treatments, reducing Spartina biomass, probably by competing for resources, when drained, but increasing it, perhaps by increasing oxygenation of soils, when flooded (Fig. 4a).
Table 1. anova tables for experiments examining mediation of competition by edaphic conditions. Significant P-values are highlighted
(a) Biomass of Spartina alterniflora in glasshouse experiment
Flooding × salinity
Flooding × competition
Salinity × competition
(b) Biomass of Juncus roemerianus in glasshouse experiment
Flooding × salinity
Flooding × competition
Salinity × competition
(c) Biomass of Juncus roemerianus in field experiment
Flooding × salinity
Flooding × competition
Salinity × competition
In the same experiment, biomass of Juncus was, like biomass of Spartina, reduced by both increased flooding and increased salinity (Fig. 4b, Table 1b). In contrast to results from Spartina, however, neighbours had no effect on biomass of Juncus, either alone or in interactions.
Biomass of Juncus plants that were transplanted into the Spartina zone was greater in the fresh than the salt treatments, and greater in the elevated than the not-elevated treatments (Fig. 5, Table 1c). A significant interaction between salinity, elevation and competition (Table 1c) indicated that Spartina competed with Juncus in some treatment combinations but facilitated it in others (Fig. 5).
Our general understanding of plant distributions revolves around inherent trade-offs in plants between competitive ability and stress tolerance (Grime 1977; Grime 1988). Similarly, studies focused on freshwater (Grace & Wetzel 1981; Grace 1989; Keddy 1989) and marine (Snow & Vince 1984; Partridge & Wilson 1988; Bertness 1991a,b; Pennings & Callaway 1992) wetlands have identified a trade-off between competitive ability and stress tolerance as the key factor creating plant zonation patterns. In most cases, however, the physical factors creating stress are not explicitly identified. Below, we first discuss how our results are in broad agreement with current paradigms, then discuss how a more explicit focus on physical stress suggests that the factors mediating plant zonation in salt marshes may change geographically.
Our results indicate that the zonation pattern between Juncus and Spartina in Georgia is maintained by a trade-off between competitive ability and stress tolerance. Although Spartina dominates lower marsh zones, it occurs at low densities in the Juncus zone, and performs well when transplanted there if Juncus is removed. Thus, the upper limit of Spartina appears to be set by competition, not physical stress (but see the discussion of salt pans below). In contrast, Juncus almost never occurs as scattered individuals within the Spartina zone, and it performs poorly when transplanted there whether Spartina is present or removed. Although Juncus plants transplanted to the Spartina zone did not die in the first growing season, Juncus is a slow-growing plant with extensive below-ground reserves, and hence tends to respond slowly to changes in abiotic conditions. The Juncus plants transplanted into the Spartina zone looked unhealthy, and we are confident that they would eventually have died had the experiment been extended for additional growing seasons. Thus, the lower limit of Juncus appears to be set by physical stress, not competition. These differences between Spartina and Juncus in stress tolerance and competitive ability lead to a general pattern in south-eastern USA marshes wherein Spartina dominates lower and middle marsh elevations, which tend to be flooded more often and saltier, and Juncus dominates higher marsh elevations, which tend to be flooded less often and lower in salinity (Wiegert & Freeman 1990). In particular, the peak salinities reached in the Spartina zone (> 60 p.p.t.) are probably highly stressful for Juncus. There is broad overlap in the environmental conditions experienced by each plant, however (Woerner & Hackney 1997), and the details of this zonation pattern vary considerably from site to site (sometimes the Spartina zone is larger than the Juncus zone and sometimes the reverse, and large patches of Juncus occasionally occur within extensive stands of Spartina), but this variation probably reflects heterogeneity in the underlying physical environment (i.e. spatial variation in geology and hydrology creating spatial variation in salinity and flooding) rather than variation in the general mechanisms affecting plant interactions (Buck and Pennings, unpublished data).
Although we have concluded that our results are consistent with general paradigms about zonation of marsh plants (Grace & Wetzel 1981; Grace 1989; Keddy 1989; Bertness 1992; Pennings & Bertness 2001), and about plant distributions in general (Grime 1977; Grime 1988), we now argue that a more explicit examination of physical stress will suggest that the underlying mechanisms driving marsh plant zonation may change geographically. Most of the experimental studies of plant zonation in USA salt marshes have been conducted at high latitudes in New England (Bertness & Ellison 1987; Bertness 1991a,b; Bertness et al. 1992). Salt marshes, however, are the dominant intertidal habitat along the entire east coast of the United States. We argue that the details of how these systems work are likely to vary across latitude because of predictable differences in salinity patterns between high and low latitude marshes on the eastern seaboard of the United States (Pennings & Bertness 1999). Previous studies in New England concluded, albeit based on correlations rather than experiments, that flooding was the major physical stress limiting plant distributions in the absence of physical disturbance (Bertness 1991b; Bertness et al. 1992). In contrast, studies conducted at lower latitudes (Pennings & Callaway 1992), similarly working with correlative data, have speculated that salinity was an additional important physical stress mediating plant zonation patterns. Salinity might assume a greater role at low latitudes because low-latitude salt marshes on the eastern and western coasts of the United States generally have saltier soils than high-latitude marshes on the same coasts.
In high-latitude marshes on both US coasts, salinity levels in undisturbed vegetation tend to decline from the low to the high marsh, and rarely are much greater than levels found in seawater (Pennings & Bertness 1999). In contrast, in low-latitude marshes on both coasts, salinity levels often increase to a peak in the middle or high marsh because of increased evaporation, which concentrates salts in the soil, and salinities may reach levels several times those found in seawater (Pennings & Bertness 1999, 2001). Our field and laboratory studies consistently indicated that salinity and flooding were both important in excluding Juncus from the Spartina zone (Figs 4 and 5). Thus we concluded that, in contrast to the conventional wisdom from studies in New England, salinity stress was important in setting the lower distribution limit of Juncus in Georgia. Similarly, previous work in a low-latitude marsh in California pointed to the importance of both flooding and salinity in mediating plant zonation patterns (Pennings & Callaway 1992). Although the conventional wisdom that salinity does not play a major role in setting zonation patterns in New England salt marshes is consistent with a large body of research, experimental manipulations of salinity conducted in parallel in both geographical regions are needed to provide a rigorous test of our proposal that the role of salinity changes geographically. We plan to report the results of such studies in the future. Whether a similar transition occurs across latitude in other geographical regions of the world is an interesting question that begs for further coordinated field studies.
The most extreme evidence of geographical variation in the role of salinity in mediating salt-marsh plant patterns lies in the occurrence of ‘salt pans’ in low-latitude marshes (Pennings & Bertness 1999, 2001). Salt pans, unvegetated expanses of the marsh that occur where soil salinities exceed levels that plants can tolerate, are a common feature of low-latitude marshes. In contrast, unvegetated areas that occur in New England marshes typically result from disturbance or waterlogging, rather than from high salinities (Pennings & Bertness 2001). Salt pans in south-eastern US marshes are typically surrounded by ‘salt meadows’, zones of highly salt-tolerant plants, such as Batis and Salicornia (Wiegert & Freeman 1990). In areas of moderately elevated salinities, salt meadows may occur without salt pans. We avoided sites where the typical zonation pattern of Spartina and Juncus was interrupted by salt meadows and/or salt pans; however, studies at such sites would probably have indicated an even stronger role for salinity in mediating zonation patterns. In particular, at many such sites, salinity probably plays a role in setting the upper distributional limit of Spartina.
The performance of Juncus in the various experimental treatments in the field was highly correlated with its performance in the analogous treatments in the glasshouse (Fig. 6). The broad agreement between these two approaches lends an extra level of confidence to our conclusions. It also suggests that, despite potential concerns (Davy & Costa 1992; Davy et al. 2000), laboratory studies of wetland systems can be highly informative if field conditions are mimicked well and more than one environmental factor is tested in combination with competition.
One interesting result that occurred in the glasshouse experiment was that Spartina performed better under freshwater flooded conditions when Juncus was present than when it was absent (Fig. 4). Although we are reluctant to speculate too much about this result because we did not measure redox potentials in this experiment, one interpretation of these results would be that Juncus was aerating the soil and thus facilitating Spartina. In New England Juncus gerardii has been shown to aerate marsh soils and increase performance of coexisting plants under certain conditions (Bertness & Hacker 1994; Hacker & Bertness 1995, 1999). Similar effects might not have occurred in the salt treatments because Juncus was more stressed, and might have been overwhelmed in the freshwater drained treatment (best conditions for Juncus) by a strong competitive effect of Juncus on Spartina.
Field experiments in ecology are highly informative, but are also labour intensive and expensive. Consequently, there is the natural desire to extrapolate their lessons as far as possible. Extrapolation, however, is risky (Underwood & Denley 1984) because ecological interactions may change with changes in abiotic conditions (Dunson & Travis 1991; Pennings et al. 2003). Comparing studies of salt-marsh plant communities conducted at low and high latitudes suggests that, in coastal wetlands, the basic trade-off between competition and stress tolerance that creates vegetation pattern is universal. The relative importance of different physical stresses, however, is likely to vary geographically as a function of geographical variation in the physical environment (Pennings & Moore 2001; Bertness & Ewanchuk 2002; Ewanchuk & Bertness 2004).
We thank Ray Callaway, Darrin Moore and Ben Nomann for assistance in the field, and Lee Stanton for discussion and comments on the manuscript. We thank Jelte van Andel, Lindsay Haddon, Robert Jefferies and the anonymous referees for constructive comments that improved the manuscript. MBG was supported by a Brown University Undergraduate Teaching and Research Assistantship. We thank DOE (NIGEC), the Andrew Mellon Foundation and NSF (OCE-99–82133, GCE-LTER program) for financial support. This is contribution number 938 from the University of Georgia Marine Institute.