CHAPTER TWO

Salt Marsh Succession: Geomorphological Controls
on
Pattern and Process


2.1 INTRODUCTION

In the most traditional sense, the term "succession" refers to the unidirectional change in the composition of an ecosystem where one group of organisms replaces another until a climax community is attained. Since the first ideas of Cowles were published in 1901, nearly a century ago, alternative definitions of succession have been discussed in the ecological literature. In 1969, Odum presented a definition of succession that seems to be generally accepted today. He stated that: 
The purpose of this paper is to examine parts (i) and (ii) of Odum's definition in relation to succession in a salt marsh. First, the aim is to characterize salt marsh succession, especially as it relates to the creekbank region of the marsh. Second, the aim is to identify and discuss the modification of the environment by the vegetation during succession. Third, the aim is to characterize what aspects of the physical environment determine the patterns and the rate of change within the system. 

2.1.1 Salt marsh succession 

The first question to be addressed is: does succession occur in a salt marsh? There is no succession in the classic sense of the replacement of one community by another. There is only a single macrophyte, the salt marsh cord grass, Spartina alterniflora, which grows predominantly in monospecific stands. The lack of a progression in plant species over time was recognized by Johnson and Raup in 1947. In an examination of Massachusetts salt marsh peat, they found that there was only detritus from cord grass. The development of the marsh ecosystem cannot be characterized by the change in species composition, as is possible in many terrestrial systems; thus, marsh succession must be defined by changes of a different nature. Cowles (1901) noted that the character of the soil on a sand dune changes along with the species composition; and Olson, in an examination of these same soils 50 years later discussed the change in the sorptive capacity of the soils (1958). Thus, succession in a salt marsh can potentially be described by changes in the physico-chemical properties of the substrate. Indeed, as a marsh ages: (1) the organic content of the sediment increases (Craft et al 1988a, Osgood & Zieman 1993b), (2) the nutrient reservoirs increase (Osgood & Zieman 1993b, Craft et al. 1988a), and (3) the redox potential (Eh) decreases (Osgood & Zieman 1993b). 

Further, in their 1973 review of successional work to date, Drury and Nisbet concluded that traditional theories of succession did not adequately describe the existing field data, and that the changes observed in the patterns and processes are not necessarily associated with a change in the species composition. They suggested that a "comprehensive theory of succession should be sought at the organismic or cellular level, and not in emergent properties of communities." This idea was also offered earlier by Shelford (1911) where he discussed that ecological succession is based upon "physiology, habits, behavior, mode of life…" of the species, and not necessarily on the change in species present. In a salt marsh, the primary production increases and stabilizes over time (Broome et al. 1986). The quantity of S. alterniflora biomass may be a useful proxy in characterizing succession in a salt marsh. In addition, with the changes in the physico-chemical nature of the substrate, the physiology of the plants may change as well. This study examines some of the physiological changes that take place within the organism during the successional process. 

2.1.2 "Reaction" during salt marsh succession 

In 1916 Clements presented the idea of 'reaction', which is "the effect which a plant or a community exerts upon its habitat". Reaction is explained in the second part of Odum's definition as: "the modification of the physical environment by the community" (1969). Reaction theory was generated in order to explain the progression from one sere to the next, with the idea that one sub-climax sere alters the environment such that it is prepared for the species of the next sere. Figure 2.1 is a schematic diagram illustrating some of the potential pathways whereby the vegetation, S. alterniflora, interacts with the environment in a series of biotic-abiotic feedbacks. There are certain biotic-abiotic interactions which act to change the physico-chemical nature of the substrate, and which will thereby affect the rate at which the marsh develops. The role of these feedbacks is examined within the framework of successional changes within the marsh. 

2.1.3 Physical controls on the rate of succession 

Salt marshes are physically driven systems. In these systems, the landscape influences the hydrology, which influences the chemistry, which controls the productivity. Thus, it follows that the rate of succession of a salt marsh will be controlled to a great extent by the physical forces which create the patterns that we see. The landscape of a marsh is often dissected by meandering tidal creeks. In a mature marsh, the creekbank is generally associated with tall form S. alterniflora. The relationship between the tall morph of S. alterniflora and the creekbank has been attributed to the increased pore water movement in this region (Wiegert et al. 1983). The increased flushing has, in turn, been linked to increased sediment aeration (Agosta 1985), increased nutrient supply (Valiela et al. 1978), decreased salinity (Phleger 1971) and decreased sulfide (King et al. 1982). Salinity and sulfide have both been shown to inhibit N uptake by S. alterniflora (Bradley & Morris 1990, Bradley & Morris 1991, Morris & Dacey 1984). Thus, the plants in this region will be better able to take up nitrogen because the concentrations of salts and sulfide will be lower. Howes et al. (1994) demonstrated that a significant amount of inorganic nitrogen and DOC are lost from the creekbank through seepage at low tide. In a mature marsh, these losses are balanced by the benefits of increased flushing, such that the S. alterniflora grows vigorously in these regions. In addition, Osgood et al. (1995) have put forth the idea that in younger marshes, where the substrate is sandy, the increased hydraulic gradient imposed by the slope at the creekbank will cause greater throughflow of nutrients. Even though measured concentrations are low, the total amount of N available to the plant is high. Thus, the morphological variation at the creekbank leads to a change in the subsurface hydrology. This, in turn leads to changes in the porewater chemistry, which affects the production of the plants. As illustrated in Figure 2.1, when the environment acts to change the production of the plants, there is a feedback whereby the plants then act to alter the physical environment. As the final aim of this study, the environmental controls on the marsh ecosystem will be examined in order to determine the role that they play in the rate of successional change in the marsh. The creekbank region of the marsh will be examined because it is an area where there is great spatial variability in the landscape, and thereby in the hydrological and chemical environment. 

2.1.4 Introduction to the current study 

Past studies of marsh succession have been primarily associated with restored and created marshes (e.g. Broome et al. 1986, Sinicrope et al. 1990, Craft et al. 1988a, Craft et al. 1988b). These studies have examined changes in primary production (Broome et al. 1986) species composition (Sinicrope et al. 1990), nutrient and organic carbon reservoirs (Craft et al. 1988a), and the sources of organic matter in these marshes (Craft et al. 1988b). Few studies have focused on naturally developing marshes (e.g. Redfield 1965, 1972, Osgood & Zieman 1993a, Osgood & Zieman 1993b, Osgood et al. 1995). While the geomorphological evolution of tidal creeks and tidal creek networks has been extensively studied (e.g. Pestrong 1965, Garofalo 1988, French & Stoddart 1992, Knighton et al. 1992, Shi et al. 1995), the relationship between the tidal creek and the development of the vegetated portion of the marsh has not been examined. 

This study takes place on a chronosequence of marshes, where marshes ranging from 1-150+ years exist side by side (J. Walsh, pers. comm.). The Hog Island marsh chronosequence site provides a natural experiment where instantaneous observations of the changes that take place as a marsh matures can be made through a space for time substitution. In addition, the geomorphological variability present at the site allows for comparison studies to be made with out a great deal of damaging experimental manipulation. In instituting this space for time substitution, we are making the assumption that the 'snap-shots' of ecosystem pattern and process at difference ages represent real trends, and that all of the marshes in this chronosequence are following the same developmental trajectory. This provides a test of the idea that succession is a unidirectional process, and this will be discussed in light of the results. 

2.2  METHODS


2.2.1 Site description 

This study takes place on Hog Island (Figure 2.2), a barrier island off of the coast of the Delmarva Peninsula, Virginia. It is a part of the Virginia Coast Reserve Long Term Ecological Research project. The island is a highly dynamic environment, which is subject to frequent disturbances (Hayden et al. 1991). This is exemplified by the fact that the town of Broadwater, a town with nearly 250 inhabitants at the turn of the century, is now underwater some distance from the present shoreline of the island (Badger & Kellam 1989). In March of 1962 a large and powerful coastal storm caused a washover event which deposited nearly a meter of sand on the mature marshes on the eastern side of the southern end of the island (Stewart 1962). Since this overwash event, the fringing S. alterniflora marshes have gradually regrown on the bay side of the island. 

Five sites were selected for this study. There are four creeks, and one site at the edge of the lagoon. The approximate ages of the creek sites are 5, 6, 15, and 150+ years as of 1996. The fifth site is approximately 3-4 years old, and is included to be able to compare a site that has developed at the edge of lagoon with sites that developed at the edge of a creek. The ages of the sites were determined by inspection of aerial photographs. I have defined age as the time since S. alterniflora has been visible on these photographs, and it is thereby the ecological age of the site. All of the young sites are the result of the 1962 washover event, and therefore have the same geological age. The 150+yr site was not affected by the 1962 washover. The age of this site is not known precisely, but it was present as a marsh on maps dating back to the 1840s (J. Walsh, pers. comm.). It may be older than 150yr, but it is most likely that it has the same sedimentary origin, in that the marsh platform was probably lain down by an earlier washover. Figure 2.3 is a map illustrating the positions of each site. The two young creeks, hereafter referred to as Y1 and Y2, which are 0.5-.75 m across, become discontinuous with the lagoon after the water has fallen past a certain level. The intermediate aged creek, hereafter referred to as I, is 1-1.5 m across and becomes discontinuous with the lagoon on the majority of tides. The old creek, referred to O, is several meters across and remains continuous with the lagoon throughout the tidal cycle. The discontinuity in creek and lagoon water at low tide is important to note because it indicates that once the lagoon water has fallen below a certain level on the ebb tide, all remaining water draining from the marsh is trapped within the creek bottom. 

2.2.2 Site instrumentation  

At each creek, two-3 m transects were established perpendicular to the creek. Past studies indicate that the influence of creekbank morphology on the subsurface hydrological processes does not extend beyond approximately 2.5 m (Howes et al. 1994, Harvey et al. 1987, Nuttle 1988, Agosta 1985) to 15m (Nuttle 1988, Nuttle and Hemond 1988) of the creekbank. Given the small size of these creeks, it is likely that 3 m covers the range of influence. Four sites were established along each transect at 0-0.5 m, 0.5-1 m, 1-2 m and 2-3 m. Hereafter, the term 'site' is used to describe each of these distances. Thus, there are 8 sites at each of the 5 marshes. A suction lysimeter ("sipper", Chambers & Odum 1990) was placed at 0.5 m, 1 m, 2 m, & 3 m along each transect. The elevation relative to mean sea level of each sipper location was determined using Pentax Total Station survey equipment. The survey was tied in to five benchmarks which are part of the LTER benchmark system, which was established using Trimble Survey Grade GPS, and is tied to USGS benchmarks on the mainland. The reference datum is the USGS 1967 reduction level. The overall accuracy of the survey is within 3-4 cm of the 1967 USGS reduction level, and the within-survey accuracy is 1.5cm in the z-direction. Figure 2.4 shows the site instrumentation and the elevations of each site. 

2.2.3 Creekbank study 

2.2.3.1 Sediment characteristics 

Samples for grain size and organic content analysis were collected in September of 1995. Three replicate 10 cm were cores were taken from each site and combined. The fractions of sand, silt and clay present at each sipper site were determined using the method described in Brower and Zar (1984); the organic content of the sediment was determined gravimetrically by loss upon combustion. The sediment nitrogen content was determined in April, June, August and October of 1995 and 1996. One cm deep cores were taken using a 5cc syringe corer. Three replicate cores were taken from each site and combined into a single sample. The samples were lyophilized and ground to homogeneity using a mortar and pestle. Prior to grinding, shells and large roots and rhizomes were removed. Fine roots were left in the samples. The nitrogen content was determined using a Carlo Erba NA1500 Elemental Analyzer. Two replicates were run of each sample to account for the analytical error of the instrument and the heterogeneity of the sample. 

2.2.3.2 Pore water collection, nutrient and salinity determination 

Pore water samples were collected from the sippers on a monthly basis from June to October 1995 and April to October 1996. Prior to collection of pore water, any old water in the sipper was evacuated using N2 gas to maintain an anoxic environment. A slight vacuum pressure was applied to the sipper using a hand-pump, and pore water was allowed to refill the sipper for a few hours. The water was extracted from the sipper using a syringe. Redox potential (Eh, or platinum electrode potential), pH, and temperature were determined immediately following collection by injecting the sample into an anaerobic chamber fitted with Corning electrodes and a temperature probe attached to a Beckman 12 pH/ISE meter. The measured redox potential, in millivolts, was adjusted by adding 199mV to correct for the Ag/AgCl, saturated KCl electrode used. Salinity was measured using a temperature compensating hand-held refractometer. 

Water for NH4+ and PO4-3 analysis was immediately filtered through 0.45m membrane filters into vacutainer tubes containing 0.1ml 6N HCl, and kept on ice. The acid was added to prevent the volatilization of NH4+ during the removal of H2S. The samples were brought back to the field laboratory, and analyzed within 6 hours. Because H2S interferes in the analysis of both NH4+ and PO4-3,each sample was bubbled for 5-10 minutes with N2 gas in order to drive off the H2S. Immediately prior to pipeting the samples for analysis and addition of the color reagents, the pH of the samples was readjusted with 0.1ml 6N NaOH. From time to time the pH of the sample was tested using a pH meter to ensure that it was in the proper range for color development. Both NH4+ and PO4-3 were determined spectrophotometrically as described in Grasshoff et al. (1983). Ammonium was determined by the addition of 0.3 ml each of trisodium citrate, phenol/nitroferricyanide, and hypochlorite reagents to 5ml of sample and standard. The samples were incubated in the dark for at least 6hr, and the absorbance was determined on a spectrophotometer at 630 nm. The pH of the PO4-3 samples was adjusted to 8.0 by titration with 6N NaOH and 1N H2SO4 using phenolphthalein as an indicator. One ml of a combined color reagent containing ammonium molybdate, sulfuric acid, ascorbic acid and potassium antimonyl tartrate solutions was added to 5.0 ml of sample and standard. Color development was allowed to proceed for 30 min, and the absorbance was read at 885 nm. 

Nitrate + nitrite (hereafter referred to as nitrate) was analyzed using the method described in Jones (1984). The nitrate in 10 ml of sample was reduced to NO2- by shaking with spongy cadmium in the presence of 5ml aluminum hydroxide suspension. One half ml of the color reagent, which consisted of sulfanilimide, N-(1-naphthyl) ethylenediamine dihydrochloride and phosphoric acid, was added. The color was allowed to develop for at least 10 min, and the absorbance was read at 540 nm. The detection limits of this method are 2 mol/l. 

Sulfide concentration was determined using a method described by Cline (1969) as modified by Otte and Morris (1994). Five ml of unfiltered porewater was added immediately to a vacutainer tube containing 5 ml ZnAc. If necessary, dilutions were made, and 0.4ml N,N-dimethyl-p-phenelynediamine sulfate + ferric chloride dye was added to each sample and standard. At least 20 min was allowed for color development, and then the absorbance was measured spectrophotometrically at 670 nm. 

2.2.3.3 Spartina alterniflora production and tissue element composition 

Production was measured at each site in September 1995 and 1996, which is the end of the growing season. At each site three 1/16th m2 haphazardly placed quadrats were thrown. The species composition was determined, and all of the above ground S. alterniflora was clipped. The plants were brought back to the laboratory and frozen until biomass determinations were made. Each plant was cleaned of sediment and all dead leaves removed. The plant height was measured to the tallest point (leaf or flower), oven dried and weighed individually. Using the density and the average weight per plant, the biomass per m2 of marsh surface was calculated. 

Spartina alterniflora leaf samples were collected from each site in June, August and October 1995 and 1996 for tissue element determination. Ten to fifteen leaves, from different plants, were collected and combined into a single sample from each site. The leaves were washed free of sediment, and stored in a freezer. The samples were lyophilized, and ground to homogeneity using a Krups coffee mill. The carbon and nitrogen composition of the tissue was analyzed using a Carlo-Erba NA1500 Elemental Analyzer. Three replicates of each sample were run in order to assess the analytical error of the instrument and the homogeneity of the samples. 

2.2.3.4 Data analysis 

Between marsh comparisons were made using SPSS General Linear Model - General Factorial function. When significant differences were found between marshes, post hoc tests were run. When variances were equal, Tukey's HSD was used; when variances were unequal, the Games-Howell test was used. A Principal Components Analysis was run on the variables shown in Table 2.1 using SPSS. Between and within marsh comparisons were made on the factor scores generated by the PCA using SPSS General Linear Model - General Factorial function. 

2.2.4 Tidal cycle study: creek water collection and analysis 

Creek water was sampled on two spring tides (14 July & 23 August, 1995) and two neap tides (16 September and 15 October, 1995). Samples were collected from each creek and the flooding lagoon at hourly intervals over an entire tidal cycle (12 hr) on each date. Salinity and pH were measured as described above. Particulate carbon and nitrogen were measured by filtering water through 13 mm pre-combusted glass fiber filters. The filters were inserted into tin capsules, dried at 60C, and analyzed using a Carlo-Erba NA1500 Elemental Analyzer. Particulate matter was determined by filtering creek water through pre-weighed, pre-combusted 47 mm glass fiber filters. The filters were dried at 60C and weighed. The filters were then combusted at 450C for 6 h, and weighed again to obtain the percent organic content of the particulate matter. Nutrient samples were filtered as described for the porewater samples, and stored in 35 ml brown Nalgene bottles on ice. The samples were analyzed for NH4+, NO3- and PO4-3 as described above. Data from the 4 tidal cycles were pooled, and averages were made at hourly intervals based on the time since high tide. 
2.3 RESULTS

The lagoon site was scoured considerably during the winter of 1995-96. Destruction of the below-ground S. alterniflora biomass was evident in the early spring of 1996, and there appeared to be elevational changes. In addition, following the establishment of this site, we found that the old marsh, which was buried in 1962, was only about 15 cm below the surface of the newly established marsh. Following the winter, it was even closer to the surface. This marsh seemed to be supplying high concentrations of PO4-3 to the porewater here. In addition, at the sites 0.5 m from the edge of the lagoon, the redox potential was very low, most likely due to the degradation of organic matter in the buried marsh. For these reasons, this marsh was not included in the majority of the data analyses presented here. It was, however, used in the PCA so that a marsh which develops in the absence of a creek can be compared with those that have creeks. 

With the exception of salinity, there were no significant differences between months, or years, so data for all months were pooled. Figures describing all data monthly, by year, can be found in Appendix A. The comparisons presented for all creekbank variables will be two-fold: (1) the means for each marsh and the associated statistical differences are presented on one figure, and (2) values based on distance from the creek for each marsh; these are presented as a separate figure. 

2.3.1 Creekbank study 

2.3.1.1 Porewater chemistry 

The porewater chemistry is presented in Figures 2.5, 2.8 and 2.9. Porewater salinity was higher in August and September, most likely due to the increased ET resulting from higher temperatures and greater S. alterniflora biomass. As illustrated in Figure 2.5A there is no difference in salinity between I and O, but there is a significant 10ppt difference between Y1 and Y2. This difference is evaluated in the following chapter. There is a positive relationship between the salinity and the distance from the creek in all marshes (Figure 2.8A). There is no trend either between (Figure 2.5D) or within (Figure 2.8B) marshes for temperature. The mean pH at Y2 is 6.6, and at all other sites it is ~6.9. The reason for this is unknown. There is no difference in redox potential between Y1 and Y2, and between I and O, however these two groupings are significantly different from each other. In the two young marshes, there is an increase in Eh with distance from the creek, and in the older marshes the maximum Eh appears at 1 m from the creek (Figure 2.8C). In the two young marshes the redox potential appears to increase with distance from the creek. The sulfide concentration in the old marsh is inversely related to the redox potential, and is highest where the redox is lowest, further from the creek. 

Mean ammonium concentration (Figure 2.5E) increases with age. There does not appear to be a trend associated with the location within each marsh (Figure 2.9A). Nitrate is consistently low (<3 M), and is only significantly different between the two youngest marshes, and the two older marshes (Figure 2.5F). However, most of these values are below the limits of detection, and these results should be treated accordingly. PO4-3 is lowest at Y2, intermediate at Y1, and is highest at I and O, as shown in Figure 2.5G. Thus there appears to be a significant trend with age. In the youngest marshes there appears to be a trend of decreasing P with increasing distance from the creek, but it is slight (Figure 2.9C). There is no apparent trend in the older marshes. 

2.3.1.2 Sediment characteristics 

Figure 2.6 shows the mean values of the sediment characteristics for each marsh. For all variables there is no significant difference between Y1 and Y2, while I and O are each significantly different from all others. These variables all show a distinct trend with age: decreased elevation, decreased sand, increased silt, clay, organic matter, and nitrogen. In Figure 2.10A it can be seen that the % organic matter is positively associated with increasing distance from the creek edge in the old marsh, and negatively associated in the younger marshes. Figure 2.10E shows that there is also an inverse relationship with %N and increasing distance from the edge, for all marshes. 

2.3.1.3 Spartina alterniflora production 

Figure 2.11 shows the S. alterniflora above ground primary production, density, and plant characteristics for each marsh, with respect to the distance from the creekbank. In all three of the immature marshes, the biomass per m2 and the average weight per plant drops off with increasing distance from the creek. This trend is not observed in the old marsh. Overall, at 0.5m from the creekbank Y1 had the highest biomass (1689.3 g/m2), and the highest weight per plant (3.2 g/plant), and I had the greatest density (695 plants/m2). Height decreases with distance in Y1 and Y2, but remains constant in I and O. There does not appear to be a difference in sexual reproduction (percent of plants flowering) with respect to the distance from the creekbank. Figure 2.7 shows the mean production characteristics for each marsh, and indicates where significant differences exist between marshes. The mean biomass for all distances are 805.8, 270.4, 956.8 and 976.9 g/m2 for Y1, Y2, I & O, respectively. In general, it appears that biomass increases with age, but only Y2 is significantly different from the others. The very high production at Y1 - 0.5 m causes the grand mean for that marsh to be very high. Overall, Y2 has significantly lower biomass, plant weight, and plant height when compared to all other marshes. The old marsh has the tallest plants, and these plants are more likely to be sexually reproducing. Even though the two young marshes are approximately the same age, there is a significantly greater biomass at Y1. The only difference in porewater chemistry between these two sites is salinity. The elevated salinity at Y2 may have a detrimental effect on S. alterniflora growth (Phleger 1971). The cause of this variation in biomass, which is attributed to the difference in the subsurface hydrology at these two sites, is the subject of the next chapter. 

2.3.1.4 Spartina alterniflora tissue element composition 

The relationships between the S. alterniflora tissue element composition and the distance from the creekbank are shown in Figure 2.12, and the mean values for each marsh are shown in Figure 2.7. As illustrated in Figure 2.12, there does not appear to be any relationship between %C and distance. However, for all marshes there is a decrease in %N, and a contemporaneous increase in C:N for plants growing farther from the creek. Figure 2.7 shows that the overall the mean %N increases significantly with age, the %C decreases, although not significantly, and the C:N decreases significantly with age. Thus, in an older marsh, there is a greater amount of above ground nitrogen both per gram of plant tissue, and overall, which is due to the higher biomass. 

2.3.1.5 Principal components analysis 

The principal components analysis resulted in 3 significant principal component vectors. The first one of these (PC1) explains 42.1% of the variability in the data set, and is the only one that will be discussed. The factor loadings associated with each variable are shown in Table 2.1. This principle component is associated primarily with sediment characteristics, although S. alterniflora characteristics are moderately important, and nutrients and pore water sulfide and Eh are associated as well. In the positive direction, PC1 is associated with an increase in organic matter, sediment carbon, sediment nitrogen, silt and clay, and a decrease in sand and elevation. PC1 is also associated with plant height and % flowering, and to a lesser extent to biomass and plant weight. In addition, high factor scores are associated with a high N content and low C:N in plants. Finally, there is a modest relationship between porewater ammonium, phosphate, sulfide and redox potential. All of these characteristics have previously been associated with older marshes, and there is a significant relationship between PC1 and age (r2=0.92, p<0.001). The mean factor scores for each site are -0.56, -0.84, 0.22, and 1.78 for Y1, Y2, I and O, respectively. See Figure 2.13A. These values are all significantly different from each other, indicating that this principal component explains the variability associated with different ages, even though age was not used as a variable. The lagoon site was included in the PCA in order to assess the aging process in a marsh without a creek. The mean factor score for this marsh was -0.59. Figure 2.13B also shows the relationship of the mean factor score for each site to the distance from the creekbank. For the lagoon site 
variable type  variable  factor 

loading 
sediment   % N   .94 
characteristics  % sand  -.93 
% organic matter   .93 
elevation  -.92 
% silt  .91 
% clay  .89 
porewater chemistry  NH4+   .59 
Eh  -.46 
PO4-3   .43 
S-2   .40 
NO3-   -.27 
pH   .17 
temperature   .06 
salinity  .03 
Spartina alterniflora  plant height  .80 
characteristics  % flowering  .73 
% N   .72 
C:N   -.71 
plant weight  .69 
biomass  .63 
density  .20 
% C   -.11 
Table 2-1. Variables used in the Principal Components Analysis, and the factor loadings associated with each variable for principle component 1 (PC1). Variables within each type are ordered in descending factor loading. 

and the old marsh, there is no pattern in factor scores with respect to the distance from the creek bank. However, in the immature marshes, there are significant differences. In all cases, the factor score, which can be taken as a proxy for the ecological or successional maturity of the site, increases as one approaches the creek. It follows, therefore, the marsh is more mature closer to the creekbank, than further away from the creekbank. For example, the mean factor score at Y1 - 0.5 m is the same as the mean factor score at I - 3 m, indicating that these two marshes have a similar functional maturity. 

2.3.2 Tidal cycle study 

2.3.2.1 Creek water chemistry 

Figure 2.14 shows the pH and salinity in the creeks and in the adjacent lagoon over the tidal cycle. The concentration in the lagoon is shown in each plot for comparison purposes. In all but Y2, the salinity remains fairly close to the salinity of the lagoon. Y2 has salinities of 50 - 60 ppt at low tide, perhaps indicating that salty water is being discharged into the creek. The pH of the lagoon exhibits a slight depression around low tide, and all of the creeks exhibit a large depression in pH at low tide. The two youngest creeks show the greatest reduction in pH at low tide. The nutrient concentration in the creeks is shown in Figure 2.15. While the nutrient concentrations in the lagoon and the old creek remain relatively constant over the tidal cycle, the three developing marshes show a marked increase in ammonium and phosphate concentration during the falling tide. Nitrate concentration is too low in all of the creeks to indicate any trend. 

2.3.2.2 Creek water particulate matter 

The concentration of particulates remained low in the lagoon and the old creek over the tidal cycle, with the exception of a single point in the old marsh (Figure 2.16A). Although there is a great deal of variability, the concentration of particulates in the younger creeks does appear to increase as the tide ebbs. As shown in Figure 2.16B, the organic content of the suspended matter increases as the tide ebbs in the younger marshes. Figure 2.17 shows the concentration of particulate N and particulate C in the creek water as a functional of the tidal cycle. Again, there is a great deal of variability in the data, but there does appear to be an increase in the concentrations of C and N in the younger creeks as the tide falls. In Y1 it looks as if the C:N of the particulate matter is higher as the tide is falling. Perhaps this indicates an increased deposition of undegraded organic matter into the creek from the marsh surface. Overall, it can be stated with some certainty, that as the tide falls in these young creeks there is an increase in the concentration of particulate matter. This could be explained by undegraded plant material being flushed from the marsh surface into the creek as the tide goes out. It is likely that further degradation of this material takes place while it is in the creek.



2.4 DISCUSSION

2.4.1 Patterns associated with succession 

As discussed previously, succession is considered to be an orderly and predictable process. The first goal of this study is to establish that succession is indeed occurring, even in the absence of a true 'succession' of species. In this natural experiment, where we have a space for time substitution, there are clearly general patterns associated with the increasing age of the marsh. These patterns are shown in Figure 2.13C. Thus, even though there is not a change in the plant community, there are predictable changes in the physico-chemical variables measured in this marsh. In general, for most of the variables that are associated with age, Y1 and Y2 tend to be grouped together. For the sediment characteristics, I is at an intermediate stage between the young marshes and the older marsh. However, for NH4+, I groups with the young marshes, and for PO4-3 it groups with the old marsh. Thus, it appears that different variables reach their 'mature' levels at different points in the aging process. 

While the biomass of S. alterniflora is not clearly associated with age, there are other characteristics of these plants that do appear to be related to the age of the marsh. In an older marsh the plants are taller, which is perhaps a response to light limitation imposed by the longer tidal inundation time. In addition, in an older marsh the plants have a significantly higher N content than younger marsh plants. Finally, plants in an old marsh are more likely to reproduce sexually relative to plants in a young marsh. It is unknown whether there is a difference in the clonal reproduction between young and old marshes. However, it is possible to infer that there is a difference in the strategy of reproduction. Plants in a young marsh are putting more effort into rhizomal reproduction as opposed to seed production, and the reverse is true in an old marsh. If this is the case, this would in a sense, contradict the idea presented by Odum (1969) that r-strategists compete well at earlier successional stages but K-strategists dominate in later successional stages. However, Odum is referring to interspecific competition, while this example illustrates intraspecific competition. Further research is necessary in order to draw any conclusions about the reproductive strategy of these plants. However, the physical and physiological variation present in this species over the chronosequence is evidence that although the species composition doesn't change, the "physiology, habits, behavior [and] mode of life" (Shelford 1911) do change. 

It is necessary to clarify that the patterns described herein are those observed at the creekbank of each of these marshes. While it is possible to assume that similar patterns will be observed within the interior of the marsh, caution must be taken in creating overly generalized theories of marsh succession from these results. 

2.4.2 Processes associated with successional patterns 

The presence of plants on the marsh surface can be linked to many of the increases in sediment and porewater characteristics that exhibit changes over time. Figure 2.1 below illustrates these processes. First, S. alterniflora acts to trap sediments that are brought in by the tides (arrow 11). This will lead to an increase in the silt and clay content of the sediment, and lead to a simultaneous decrease in the sand content. This change in the relative proportions of grain size fractions will result in a decrease in the hydraulic conductivity of the sediment (arrow 12). With a lower conductivity, fewer nutrients will be leached out of the sediment on a falling tide (arrow 13). Over time, S. alterniflora will contribute organic matter to the sediment in the form of wrack, and dead roots and rhizomes (arrow 10). This is likely the cause of the increase in organic matter, as well as the increase in the N content of the sediment. The presence of increasing amounts of reduced organic matter in the sediment will cause a reduction in the redox potential in the sediments and will also lead to anaerobic degradation pathways, which in turn will lead to an increase in the H2S concentration in the sediment (arrow 15). The remineralization of this organic matter will also lead to an increase in N and P in the porewater (arrow 14); as mentioned above, fewer of these nutrients will be lost due to a decreased flushing of the system (arrow 4, negative feedback). The increase in N and P will promote additional S. alterniflora growth. Thus, there is a positive feedback between biotic (S. alterniflora production and decomposition) and abiotic factors (sediment accumulation) which acts to accelerate the rate at which a marsh appears older. This type of feedback loop will continue until the quantity of reduced carbon in the sediment leads to a sufficiently low redox potential that it interferes with the plants ability to function through inhibition of N uptake (Morris & Dacey 1984) and sulfide toxicity (DeLaune et al. 1983 ). 

The factor scores from the principle components analysis can be used as a proxy for the maturity of the marsh, and can used to gauge the rate of succession at each site. Using these factor scores, it appears that in terms of the functional maturity of the marshes, or the successional "age", the sequence is: Y2, L, Y1, I and O, from youngest to oldest (Figure 2.13A). Y2, which has lower S. alterniflora production due to elevated salinity, has a limited capacity to participate in the biotic-abiotic feedback loop described above. Y1, on the other hand, has depressed salinities, and higher production. This higher production allows for greater biotic-abiotic interactions, which thereby accelerates the aging process. While these two marshes have the same ecological age, it is evident that there are other factors (e.g. the salinity) that are controlling the relative rate of succession. 

2.4.3 Spatial patterns and processes at the creekbank 

The production in the region closest to the creekbank in the young marshes is higher than further from the creek. Osgood (1996) postulated that in younger marshes, the hydraulic gradient established by the sloping surface characteristic of these marshes acts to supply plant roots with a greater supply of nutrients over time than is apparent in the porewater at any one time. In the vicinity of the very young creeks on Hog Island the porewater nutrient concentration is very low, and does not appear that it should be able to support the plant biomass measured there. However, the hydraulic gradient set up by the creek may act to augment the supply of nutrients to these plants (arrows 2 and 4 in Figure 2.1). The hydraulic gradient also acts to increase the flushing out of salts and toxins (arrows 5 and 6). The higher biomass at the creekbank intimates that there will be greater sediment deposition and greater organic matter contribution to the sediment. Thus, even within a single marsh the magnitude of the biotic-abiotic feedback processes varies spatially. As described above, these feedback processes lead to an acceleration of the rate of succession, and it can be suggested that succession is occurring more rapidly closest to the creekbank. Again, using the results of the PCA as a proxy for marsh maturity, in the young marshes the region closest to the creekbank does appear to be more functionally mature because it has a higher mean factor score (Figure 2.13B). In the lagoon marsh, there is no trend in the factor scores with relation to the edge of the marsh. This makes it appear that the creek is important in determining the spatial patterns of functional maturity. Thus, the landscape, or geomorphology of the region acts to facilitate the successional process. In addition, there are processes associated with the creek itself that act to accelerate the maturation of the marsh in the immediate vicinity of the creek. 

2.4.4 Processes within the creek 

It appears that as the tide is falling, nutrients are being leached out into the creeks. There was a slight difference in this effect between neap and spring tides. It has been noted that spring tides generally have higher nutrient concentrations than neap tides (Vorosmarty & Loder 1994); however, given that the spring tides, which did have higher concentrations, were sampled during July and August, it is likely that the higher temperatures resulted in greater evaporation from the creeks. As the tide falls, nutrients and salts are leached out of the sediment and into the creeks. Given the higher elevation and higher hydraulic conductivity of the sediments in vicinity of the young creeks, it is likely that more water, and hence more nutrients will be lost per unit area of creekbank in younger compared to older marshes (Harvey et al. 1987, Howes et al. 1994). These processes correspond to arrows 2 and 4 in Figure 2.1. However, once the tide is lower than the creek bottom, all further leachates are trapped in the creek bottom. This is true for all material that is washed from the surface of the marsh as well. These processes are shown schematically in Figure 2.18A. 

According to general theories of ecosystem development, young ecosystems are inherently leaky (Odum 1969, Vitousek & Reiners 1980). As an ecosystem matures, it has more developed biogeochemical cycling such that it only loses what it can afford to lose. Given the observed concentrations of particulates and nutrients found in the young creeks on Hog Island, it appears that these young systems are leaky. It is unknown what quantity of nutrients and particulates is lost from the older marsh because dilution takes place in this large creek. In the past two decades, ideas regarding marshes as a source or a sink of nutrients and particulates to the surrounding estuaries have been extensively examined (e.g. Redfield 1972, Wolaver & Spurrier 1988, Wolaver et al. 1988, Dame et al. 1992, Dame & Gardner 1993, Childers 1994). Childers (1994), in a review of a number of studies on the flux of materials between marshes and estuaries, concluded that in systems where the tidal range is greater than 1m, young marshes take up nutrients and organic matter from the flooding water at greater rates than more mature marshes. The tidal range on Hog Island is 1.3 m. 

At low tide, the nutrients and particulates are trapped within the small creeks because no further drainage into the lagoon occurs. Within the creek, the nutrients are concentrated due to evaporative processes, and further degradation of organic particulates can take place in this nutrient rich environment. As the tide rises, the nutrients and particulates are brought back up over the surface of the marsh, where the nutrients become available for uptake by plants, and the particulates are deposited on the surface of the marsh. Figure 2.18B illustrates these processes. These processes correspond to the arrows labeled 1 and 3 in Figure 2.1. Thus, although these systems are "leaky", there does not appear to be an effective loss of nutrients through the creekbank, or particulates over the surface. Reed (1988) discussed that creeks can serve as a temporary storage trap for particulates: the particulates are held within the creeks during "normal" tides, and returned to the marsh on exceptionally high tides. This would indicate that this process of returning nutrients and particulates does not necessarily occur during every tide, but does occur on exceptionally high tides and on tides with a greater flooding velocity. The young creeks are functional structures whereby nutrients and organic compounds are retained within the system. 

Given the sediment organic matter distribution, it appears that the particulates are predominantly deposited close to the creekbank. Movement further towards the interior of the marsh is probably impeded by the S. alterniflora stems. That being the case, the organic content and grain size of the region closest to the creek bank that is vegetated will begin to attain values characteristic of an older marsh more rapidly than areas further towards the marsh interior. Additionally, the increase in organic matter, and the concentration of nutrients in the creek bottom, and subsequent return of these materials to the marsh closest to the creek will act to 'fertilize' this portion of the marsh. Thus, the creek acts to supply nutrients to the plants indirectly, as previously discussed by (1) increasing the flow past the roots, and (2) increasing the plants' capacity for uptake, and directly, by concentrating the nutrients, and returning them to the surface of the marsh. 

2.4.5 Conclusions 

There are predictable trends that take place in a salt marsh as it ages. There is the transition from a sandy substrate to a more highly organic, muddy substrate. The porewater becomes enriched in nutrients, and the sulfide concentration increases. The overall biomass increases, and the plants exhibit different reproductive patterns, and tissue element compositions. Thus, a varietal form of succession, although not the classical one, does occur in these marshes. The rate at which this succession occurs seems to vary spatially within the marsh; the highest rates are closest to the creekbank. There seems to be a positive feedback between the differential geomorphology associated with the creekbank, which acts to promote S. alterniflora growth, and the rate of development of the marsh. In addition, the hydrological processes within the creek itself, which are associated with the rising and falling of the tides, act to accelerate the development of the marsh by preventing the loss of nutrients and particulates. 

There are some potential limitations to the applicability of this study to other systems. First, this is a back-barrier marsh. The sedimentary and geological history is different than other types of marshes, and these marshes are much less expansive than typical mainland marshes. Second, the 'space-for-time' substitution provided by this chronosequence requires the assumption that all of these marshes are following the same successional pathway, and are developing at similar rates. While they may be following the same pathway, the rates of development are clearly unequal. This is shown by the difference in the functional 'maturity' (as defined by the PCA) between Y1 and Y2. While these marshes have the same ecological age, there are factors external to the marsh proper, such as the properties of the catchment of which they are a part of, which leads to a difference in macrophyte biomass. At Y2, the interaction between the biotic and abiotic factors, or the "reaction", which defines the successional process, is diminished, and thus the rate of maturation is likewise diminished. Thus, although the successional process initiated at the same time, the two systems have not progressed identically. 

Despite the idea that not all young marshes are 'created equal', there does appear to be a general pattern of process associated with succession which depends on the geomorphology of the system. The tidal creeks on Hog Island act to accelerate the aging process in the vicinity of the creek. The creeks act to prevent the loss of nutrients and particulate organic matter from the marsh by trapping them in the creek bottom. The marsh is then fertilized with these nutrients. This factor, along with the creation of a lower salinity, higher oxygen environment, leads to increased S. alterniflora production. Through the biotic-abiotic feedback loops discussed above, increased production leads to an increased rate of marsh maturation. The marsh becomes functionally more mature in areas where the production is higher. In this fashion, a small chronosequence is developed at the edge of the creek, and the marsh spreads from the creekbank outwards. 

2.4.6 Implications 

Salt marshes are an important ecological and economic resource (Lugo & Brinson 1979). In recent years, there has been an increase in efforts to restore and create marshes for mitigation of destroyed marshes. Understanding the patterns and processes associated with marsh development is crucial to planning successful restoration efforts. As discussed by Mitsch and Wilson (1996), one of the greatest needs in wetland mitigation projects is a better understanding of wetland function. Many mitigation projects fail because of a lack of communication between the engineers who design and restore wetlands and the ecologists who study them. In an evaluation of the function of a restored marsh, Moy and Levin (1991) suggested that increased creek frontage may have increased the utilization of the marsh by macrofauna. This study suggests that in a naturally developing marsh the creekbank plays an important role in enhancing both primary production and the rate of succession. In addition, these small creeks help to retain nutrients within the system. If, indeed, tidal creeks do accelerate the formation of a mature marsh, restoration could be designed such that the ratio of creek bank to marsh interior will maximize the rate of marsh development.