Assessing Dynamic Soil Properties in Southern New England Using an Ecological Site Framework

An ecological site is defined as a distinctive kind of land based on recurring soil, landform, geological, and climate characteristics that differs from other kinds of land in its ability to produce distinctive kinds and amounts of vegetation and in its ability to respond similarly to management actions and natural disturbances. The primary objective of this study was to initiate provisional ecological site concepts for upland, riparian, salt marsh, and subaqueous soils in southern New England by comparing sites that share similar geomorphic settings, but differing soil types. For each system, I also determined how a specific disturbance or management scenario affected dynamic soil properties. In uplands, Merrimac (sandy) and Enfield (silty) soil components were compared to determine whether or not these soils are different ecological sites. My preliminary investigation showed that forest stands on these soils could be coniferous or deciduous. Therefore, within each upland soil type, three deciduous and three coniferous sites were investigated. Within the upper 50 cm, Merrimac soils averaged 61% sand, which was significantly greater than the 26% recorded for Enfield (p<0.01). Although this supports that these soils differ in drainage, soil texture did not seem to influence the 50 cm soil organic carbon pools between Merrimac (109 Mg C ha) and Enfield (101 Mg C ha; p=0.66). Even though the Merrimac soils are sandier and thus better drained than Enfield, the similarity in vegetation composition and tree productivity indicate that these soils have similar ecological potential. 15 years after the selective harvest of sites with either Enfield or Merrimac soils, soil carbon pools were determined to be resilient to change. I concluded that the 50% removal of overstory trees decreases carbon additions from litter by 28% (p=0.036), but that this reduction did not significantly impact the distribution of soil carbon within the soil profile in both Merrimac and Enfield soils. For riparian ecological sites, I aimed to develop concepts to differentiate poorly drained (Walpole) and very poorly drained (Scarboro) soils. Both the Walpole and Scarboro riparian sites had stands of Acer rubrum, but there were observable differences in the understory species composition that support separate ecological sites for these soil systems. Carex stricta and Symplocarpus foetidus were the two species that seemed to indicate the very poorly drained conditions of the Scarboro soils. Within the upper 50 cm, Scarboro soils averaged 210 Mg C ha, which was greater than the 116 Mg C ha recorded for Walpole (p=0.17). The higher water table found at the Scarboro sites is the likely cause of increased organic matter accumulation and thus the higher SOC pool that was observed in comparison to the other soils used in this study. In a plot enrichment study, I compared two levels of nitrogen additions (7.5 and 15 g N m yr) with a control to determine whether nitrogen enrichment alters dynamic soil properties in riparian sites with Scarboro soils. Root biomass, measured in the upper 20 cm, was 4.6 times greater in the high treatment when compared to the control (p=0.006). The low treatment showed a similar trend with 1.6 times more root biomass than the control (p=0.135). Thus, N may be a limiting nutrient for plant growth in these riparian soils. Although there were significant root biomass differences, above ground biomass values were similar across treatments. In salt marshes, Ipswich and Matunuck soils were investigated to determine how these soils respond to ditching and whether or not they are different ecological sites. The main difference between Ipswich (Histosols) and Matunuck (Entisols) soils is the thickness of organic materials. Based on the kind of vegetation present and the response of the vegetation to salt marsh ditching, these soils are the same ecological site. On both soils, Spartina patens and tall Spartina alterniflora were most common at or near the edge of the ditch and short S. alterniflora and salt marsh pannes occupied zones inward from the ditch. The productivity and distribution of individual salt marsh species is based on several factors including soil salinity, which is often a function of the distance of the pedon to the marsh-water interface. Four passive opentopped warming chambers (OTCs) were installed on an Ipswich soil to determine how increased temperature will effect soil carbon dynamics. I concluded that OTCs can successfully increase air temperatures, but modifications to the design used in this study may be necessary to achieve projected (1.5-4 °C) temperature increases. Postseason biomass was 32% greater in the OTC plots in 2012 (p=0.06) and 91% more in 2013 (p=0.01), suggesting higher temperatures could increase productivity in salt marshes. However, potential increases in carbon additions to the soil may be offset by increased decomposition. I used macroinvertebrate distributions to compare Massapog and Pishagqua soils to illustrate that subaqueous soils can be viewed through an ecological site framework. Massapog soils are part of the flood-tidal delta, a high energy environment near the estuary’s inlet. These soils are sandier and have less SOM compared to the Pishagqua soils, which form on the bay floor, an area protected from high energy deposition. Because of their different geomorphic settings, 94% of the invertebrate community sampled from the Massapog soils were filter feeders, while in the Pishagqua soils the benthic community mostly consisted of deposit feeders (78%). Invertebrate density was reduced in dredged sites by 97 and 71% for the Massapog and Pishagqua soils, respectively. In the Massapog soils, dredging increased water depths promoting eelgrass colonization. This change induced a shift from dominantly filter feeding organisms such as Mya arenaria and Clymenella torquata to deposit feeders including Nephlys picta and species in the Ampeliscidae family. The invertebrate community in the Pishagqua soils was similar between the dredged and control site, indicating that these soils likely respond differently to dredging. I found that water depth strongly influences the presence of eelgrass, likely because depth influences light availability. I believe that in most cases dredging lagoon bottom soils will inhibit their ability to support eelgrass because depth will be too great. In contrast, dredging in the floodtidal delta could inhibit or induce eelgrass presence. For both Massapog and Pishagqua dredging increased depth which resulted in finer textures and greater SOC accumulation.

reduction did not significantly impact the distribution of soil carbon within the soil profile in both Merrimac and Enfield soils.
For riparian ecological sites, I aimed to develop concepts to differentiate poorly drained (Walpole) and very poorly drained (Scarboro) soils. Both the Walpole and Scarboro riparian sites had stands of Acer rubrum, but there were observable differences in the understory species composition that support separate ecological sites for these soil systems. Carex stricta and Symplocarpus foetidus were the two species that seemed to indicate the very poorly drained conditions of the Scarboro soils.
Within the upper 50 cm, Scarboro soils averaged 210 Mg C ha -1 , which was greater than the 116 Mg C ha -1 recorded for Walpole (p=0.17). The higher water table found at the Scarboro sites is the likely cause of increased organic matter accumulation and thus the higher SOC pool that was observed in comparison to the other soils used in this study. In a plot enrichment study, I compared two levels of nitrogen additions (7.5 and 15 g N m -2 yr -1 ) with a control to determine whether nitrogen enrichment alters dynamic soil properties in riparian sites with Scarboro soils. Root biomass, measured in the upper 20 cm, was 4.6 times greater in the high treatment when compared to the control (p=0.006). The low treatment showed a similar trend with 1.6 times more root biomass than the control (p=0.135). Thus, N may be a limiting nutrient for plant growth in these riparian soils. Although there were significant root biomass differences, above ground biomass values were similar across treatments.
In salt marshes, Ipswich and Matunuck soils were investigated to determine how these soils respond to ditching and whether or not they are different ecological sites. The main difference between Ipswich (Histosols) and Matunuck (Entisols) soils is the thickness of organic materials. Based on the kind of vegetation present and the response of the vegetation to salt marsh ditching, these soils are the same ecological site. On both soils, Spartina patens and tall Spartina alterniflora were most common at or near the edge of the ditch and short S. alterniflora and salt marsh pannes occupied zones inward from the ditch. The productivity and distribution of individual salt marsh species is based on several factors including soil salinity, which is often a function of the distance of the pedon to the marsh-water interface. Four passive opentopped warming chambers (OTCs) were installed on an Ipswich soil to determine how increased temperature will effect soil carbon dynamics. I concluded that OTCs can successfully increase air temperatures, but modifications to the design used in this study may be necessary to achieve projected (1.5-4 °C) temperature increases. Postseason biomass was 32% greater in the OTC plots in 2012 (p=0.06) and 91% more in 2013 (p=0.01), suggesting higher temperatures could increase productivity in salt marshes. However, potential increases in carbon additions to the soil may be offset by increased decomposition. I used macroinvertebrate distributions to compare Massapog and Pishagqua soils to illustrate that subaqueous soils can be viewed through an ecological site framework.
Massapog soils are part of the flood-tidal delta, a high energy environment near the estuary's inlet. These soils are sandier and have less SOM compared to the Pishagqua soils, which form on the bay floor, an area protected from high energy deposition.
Because of their different geomorphic settings, 94% of the invertebrate community sampled from the Massapog soils were filter feeders, while in the Pishagqua soils the benthic community mostly consisted of deposit feeders (78%). Invertebrate density was reduced in dredged sites by 97 and 71% for the Massapog and Pishagqua soils, respectively. In the Massapog soils, dredging increased water depths promoting eelgrass colonization. This change induced a shift from dominantly filter feeding organisms such as Mya arenaria and Clymenella torquata to deposit feeders including Nephlys picta and species in the Ampeliscidae family. The invertebrate community in the Pishagqua soils was similar between the dredged and control site, indicating that these soils likely respond differently to dredging. I found that water depth strongly influences the presence of eelgrass, likely because depth influences light availability. I believe that in most cases dredging lagoon bottom soils will inhibit their ability to support eelgrass because depth will be too great. In contrast, dredging in the floodtidal delta could inhibit or induce eelgrass presence. For both Massapog and Pishagqua dredging increased depth which resulted in finer textures and greater SOC accumulation.
x LIST OF TABLES  either deciduous or coniferous cover were compared to determine whether or not these soils are different ecological sites. Within the upper 50 cm, Merrimac soils averaged 61% sand, which was significantly greater than the 26% recorded for Enfield (p<0.01). Although this supports that these soils differ in drainage, these soils had similar organic carbon pools (109 and 101 Mg C ha -1 for the Merrimac and Enfield, respectively; p=0.66). Similarity in vegetation composition and tree productivity suggest that these soils have similar ecological potential. Selective harvest of 50% of overstory trees decreased carbon additions from litter by 28% (p=0.036), but this reduction did not significantly impact the distribution of soil carbon within the soil profile suggesting these soils were resilient to change. For riparian ecological sites, I aimed to develop concepts to differentiate poorly drained (Walpole) and very poorly drained (Scarboro) soils. Both the Walpole and Scarboro riparian sites had stands of Acer rubrum, but there were observable differences in the understory species composition that support separate ecological sites for these soil systems. Carex stricta and Symplocarpus foetidus were the two species that seemed to indicate the very poorly drained conditions of the Scarboro soils. Within the upper 50 cm, SOC pools for Scarboro sites averaged 210 Mg C ha -1 , which was greater than the 116 Mg C ha -1 recorded for Walpole (p=0.17). The higher water table found at the Scarboro sites is the likely cause of increased organic matter accumulation and thus the higher SOC pool that was observed in comparison to the other soils used in this study. In a plot enrichment study, I compared two levels of nitrogen additions (7.5 and 15 g N m -2 yr -1 ) to determine whether nitrogen enrichment alters dynamic soil properties in riparian soils. Root biomass, measured in the upper 20 cm, was 4.6 times greater in the high treatment when compared to the control (p=0.006). The low treatment showed a similar trend with 1.6 times more root biomass than the control (p=0.135).This finding supports that N may be a limiting nutrient for plant growth in forested riparian systems. Thus, N may be a limiting nutrient for plant growth in these riparian soils.
Although there significant root biomass differences, above ground biomass values were similar across treatments.

INTRODUCTION
Soil-based interpretations are an effective decision-making tool for land use and management. Commonly used soil interpretations include suitability of the land for building roads, supporting houses with basements, and siting for septic tank absorption fields (Soil Survey Staff, 2008). Over the past decade, the Natural Resource Conservation Service (NRCS) and soil survey activities are transitioning toward a more ecological approach to soils and soil interpretations (Herrick et al., 2006). Herrick et al. (2006) noted that this change in approach is the result of an increase in our understanding of how ecosystems function. In addition, the demand for consistent management and monitoring across regions has increased in order to achieve broad scale management goals (Herrick et al., 2006). Ecological site descriptions (ESDs) are a tool that has been developed for monitoring and documenting the condition of ecosystems across regions with similar landscapes. An ecological site is defined as a distinctive kind of land based on recurring soil, landform, geological, and climate characteristics that differs from other kinds of land in its ability to produce distinctive kinds and amounts of vegetation and in its ability to respond similarly to management actions and natural disturbances (NRCS, 2013). ESDs provide a consistent framework for describing soil, vegetation, and abiotic features; delineating landscape scale units that share similar responses to management activities or disturbance processes; and estimating ecosystem services that can be expected from particular soil/vegetation combinations (Townsend, 2010).
Unlike typical vegetation surveys, ESDs provide land managers with an understanding of the potential vegetation that may exist under certain management conditions rather just a snapshot of the existing vegetation. To document several potential vegetation communities that may exist on a given site, each ESD has an associated state and transition model (S&TM) which describes a reference state of vegetation and a series of alternative states that have transitioned from the reference community through management or disturbance (Briske et al., 2005). In the past, the reference community has been defined as the plant community that existed at the time of European immigration and settlement (NRCS, 1998).
In New England, reforestation of previously cleared land, five centuries of extensive land use, and changes in aspects of environmental conditions have shifted regional forest communities from long lived, shade tolerant species to secondary, shade intolerant species . Changes in the ecology of the New England forests, including increased atmospheric acid deposition, nitrogen loading, and disease have favored the colonization of species that were not dominant during pre-colonial times (Bromley, 1935;Johnson and Siccama, 1983). Specifically, species such as beech, hemlock, elm, hickory, and chestnut have decreased since pre-European settlement, while highly productive and widely dispersed species such as oaks, maples, and pines have increased Hall et al., 2002).
Therefore, finding sites that represent the state that existed before colonial times is essentially impossible. As such, mature plant communities that represent current climatic and environmental conditions, and are common throughout the landscape, are the key to understanding dynamic soil properties under recent environmental conditions (Duniway, 2010).
Forest land cover and productivity in the eastern US has increased dramatically over the last century due to agricultural abandonment, shifts from public to private land ownership, and the reduced demand for fuel wood and lumber (Clawson, 1979).
Over 59% of southern New England, which includes Connecticut, Rhode Island, and Massachusetts, is currently forested (Butler et al., 2007). Although New England has experienced multiple land-use shifts over the last 400 years, most of the work focused on ESDs has taken place in rangeland in the western states of the US. Townsend (2010) reported that less than 10% of the 7000 ESDs that had been recorded were made for forested ecosystems, and no ESDs had been made in southern New England.
The current goal for NRCS and partnering agencies is to have provisional ecological site descriptions across the country within the next five years (Brown, 2015;personal communication). In order to meet this goal, concepts to distinguish forested ecological sites in New England must be developed.
Soil properties used to differentiate ecological sites are typically inherent such as soil texture or parent materials (Duniway et al., 2010 (Duniway et al., 2010). Soil map unit components provide the best opportunity to identify soils with similar ecological potential which can be used to develop ecological site concepts (Duniway et al., 2010).
Unlike the inherent properties used for mapping soils, dynamic properties are those with potential to change with management or disturbance. In New England, soil carbon and nitrogen are dynamic in respect to changes in land use such as agricultural abandonment (Compton and Boone, 2000;Stolt et al., 2010). Understanding the resistance and resilience of these properties to change is important for making land use decisions, especially now that we recognize their role in ecosystem services.
The primary objective of this study was to initiate provisional ecological site concepts for forested riparian and upland glaciofluvial soils in southern New England by comparing sites that share similar geomorphic settings, but differing soil types.
Specifically in uplands, I wanted to test if soils with contrasting particle size classes differ ecologically. For riparian ecological sites, I aimed to develop concepts to differentiate poorly drained (mineral) and very poorly drained (organic) soils. To begin to develop an understanding of the underlying ecological processes that lead to changes in soil properties with disturbance, I quantified the effects of two drivers of change to forest soil dynamic properties: selective harvesting in upland soils and increased nitrogen loading in riparian soils. These disturbances ultimately represent a transition from a reference state in the state and transition model (S&TM).

Forested Uplands and Selective Harvesting
Numerous studies have focused on environmental and historical influences on forest community composition and dynamics on a regional scale , Fuller et al., 1998Hall et al., 2002). Most of these studies document long-term and widespread trends in forest composition. In contrast, other studies have focused on the response of specific soil or vegetation properties to management. For example, Compton and Boone (2000) studied changes in New England forest soils resulting from historic logging and cultivation, but neglected the response of vegetation.
Developing an understanding of regional forest dynamics and specific biotic or abiotic properties does not provide land managers with an effective tool to predict the effect of disturbance at a local scale. Instead, an ecological approach that considers changes in both vegetation and soils should be used to develop an understanding of how use and management affect the ecosystem functions and values.
In some forests, clear-cutting is an inappropriate means of harvesting timber because it can lead to management issues concerning wildlife habitat, soil stability, and water quality (Keenan and Kimmins, 1993). Therefore in many situations, alternative practices including patch cutting and selective harvesting have been implemented to reduce impacts of logging on ecosystem services. Selective cutting is a silvicultural practice in which only desired trees are removed resulting in an uneven forest stand. A study by Brooks and Kyker-Snowman (2008) showed that partial harvesting of the forest canopy has minimal effects on forest soil temperature and humidity, possibly due to the rapid growth of understory vegetation following timber removal. The effects of selective cutting on other soil dynamics, such as carbon distribution in New England forests, is currently in question. A timber product output survey conducted in 2004-2005 claimed that one third of the timber harvested in southern New England was eastern white pine (Pinus strobus); of which 90% was harvested for commercial timber (Butler et al., 2007). In this study, I used a paired site approach to document changes in soil and vegetation dynamics between selectively harvested and uncut stands of P. strobus. Specifically, I wanted to determine how carbon additions from litter, deadfall, and emergent vegetation change with harvesting and if these changes induce a response in the amount and distribution of soil organic carbon.

Forested Riparian Zones
Forested riparian zones occupy the interface between upland and aquatic systems and provide ecosystem functions such as flood mitigation, water quality improvement, and wildlife habitat (Mitsch and Gosselink, 2000). For example, riparian soils act as a sink for nitrogen additions from ground water, precipitation, and surface runoff (Lowrance et al., 1984;Galloway et al., 2003) through the process of denitrification. Addy et al. (1999) found that forested riparian soils exhibit higher ground water nitrate removal rates than herbaceous riparian soils suggesting land use and cover has an effect on riparian zone soil functions.
Additional N that is not removed via denitrification may influence vegetation productivity and microbial activity which affects other soil processes such as soil respiration. Total soil respiration is the result of the production of CO2 from microbial decomposition, diffusion through culms, and root and rhizome respiration (Howes et al., 1985;Wigand et al., 2009). Although respiration from fine roots is a major contributor to soil respiration, decomposition is responsible for increased respiration with nitrogen loading. With increased nitrogen availability root production increases but high root turnover rates may result in less belowground biomass (Valiela et al., 1976;Nadelhoffer, 2000). If this is the case, carbon dynamics in the soil system may be altered.
There has been contrasting reports as to the effect of increased nitrogen loading on short term soil respiration efflux in upland forest soils; with some studies suggesting no effect (Lee et al., 2003), other studies report an increase in respiration (Pregitzer et al., 2000), while still other studies a decrease (Bowden et al., 2004;Mo et al., 2008;Janssens et al., 2010). In tidal wetlands, however, respiration has been shown to increase with nitrogen loading (Valiela et al., 1976;Wigand et al., 2009).
These contrasting findings suggest that the response of soil respiration to increased nitrogen varies between ecosystems and associated soil types. In this study, I used a plot enrichment experiment to clarify the fate of soil respiration and related dynamic soil properties in forested riparian zones resulting from nitrogen additions.

Site Selection
The soils of southern New England are the result of the advance and retreat of last glaciation which occurred between 10,000 and 25,000 years ago (Boothroyd and Sirkin, 2002). Glaciofluvial deposits are stratified soil materials that were deposited via meltwater from receding glaciers (Gustavson and Boothroyd, 1987). These deposits are often capped with silty loess that was deposited over the landscape following glacial retreat (Boothroyd and Sirkin, 2002 (Richardson, 2006). Litter trays were sampled monthly from September through November, and at the end of August during a period when minimal litter deposition occurs (Richardson, 2006). Along with each litter tray, a 1 m 2 plot was delineated at each station to measure emergent vegetation and deadfall (any woody debris greater than 1 cm; Richardson, 2006). Prior to field collection, plots were cleared of existing vegetation and deadfall. After one year, all deadfall and emergent vegetation within the plots was collected for laboratory analysis.
To investigate the fate of soil dynamics in riparian zones, three Scarboro sites (HLS, BZS, VRS) were chosen for the nitrogen enrichment experiment. At each of these sites, three clusters, each containing three 1 m 2 plots were marked to receive different N-addition treatments (control, low, and high). To simulate nitrogen enrichment in riparian soils, urea dissolved in 10 L of water from the adjacent stream was applied to each treatment plot. Water with no added nitrogen was added to the control plots. The low treatment consisted of two additions of urea totaling 7.5 g N m -2 yr -1 . This addition is equivalent to the upper range of atmospheric N-deposition concentrations in the northern hemisphere (Galloway, 2003) and annual N loads in the region (Lowrance et al., 1995;Ettema et al., 1999). The high treatment was applied in two pulses and equivalent to 15 g N m -2 yr -1 . Nitrogen was not applied to the control plots. Prior to the riparian zone nutrient addition experiment, simulated soil peds, also known as in-growth cores, were buried within each plot to measure carbon additions from fine root production (Stolt et al., 1998;Ricker et al., 2014). In-growth cores were constructed in nylon bags with 15-cm length and 4-cm diameter and buried to a depth of 5-20 cm (Ricker et al., 2014). The bags were filled with mineral soil material collected from the upper horizon of a riparian soil similar to the soils present at each site. In-growth cores were retrieved after two growing seasons and sieved to determine root content. In-situ CO2 efflux measurements were made monthly throughout two growing seasons using a Li-Cor 6262 infrared gas analyzer (Li-Cor, Lincoln, Nebraska). One 25 cm diameter PVC collar was installed in the center of each plot two weeks prior to the initial nutrient addition to a depth of 2.5 cm, which was used to create a seal between the Li-Cor analyzer and the soil. The PVC collars were left in place throughout the duration of this experiment (Davis et al., 2010). At the end of each growing season, herbaceous understory vegetation within each plot was clipped at the soil surface and returned to the lab to determine aboveground biomass production.
Laboratory Analysis Soil organic matter, particle size, bulk density, and carbon and nitrogen content were measured for each soil sample. Soil bulk density was measured by dividing the soil dry weight (105 ºC) by a known volume taken from each soil horizon (Blake and Hartge, 1986). Soil organic matter content was measured via the loss on ignition method (Nelson and Sommers, 1996) Acer rubrum, annual rings were distinguished using a phloroglucinol dye solution (NRCS, 2004;Richardson and Stolt, 2013). Age correction factors were used to add the number of years for the tree to reach breast height (NRCS, 2004) and used to calculate the total age (Carmean et al., 1989). All tree data for Quercus spp. were grouped for analysis. For this study, stand age was reported as the average age between dominant tree species. Roots from the in-growth cores were separated using tweezers, shaken for 12 hours in 0.5 g L -1 sodium hexametaphosphate to remove soil material, and rinsed (Ricker et al., 2014). All litter, deadfall, and plant biomass samples were oven dried at 60 °C, and weighed. It was assumed that half of the oven dry weight of all plant samples was carbon (Nelson and Sommers, 1996).

Statistical Analysis
Soil properties were weighted by horizon thickness and averaged for the upper 50 cm. Total soil carbon in the upper 50 cm was calculated and compared between soil types. The number and proportion of species within each wetland indicator category (Lichvar, 2012) and total species richness were calculated by soil type for comparison.
For the upland sites, two-way Analysis of Variance (ANOVA) tests were used to determine the effects of soil and cover type on vegetation, site attributes, and soil properties. When a significant difference was detected, Tukey's test was used to determine which means differed. Data were compared between Scarboro and Walpole riparian sites as well as upland harvested and control sites using paired t-tests. For the riparian enrichment experiment, ANOVA was used to test for differences between the two fertilizer treatments and the control. When differences were detected, a pairwise multiple comparison test (Holm-Sidak) was used to determine which treatments differed from the control.

Upland Soils
Precipitation, temperature, and elevation were similar between all upland and riparian sites (p >0.05). Of the twelve upland sites chosen for vegetation comparison ( Figure 1.1), six were representative of the Merrimac series ( which is important in plant growth and may support differentiating these soils as separate ecological sites. Average pH was higher in the Enfield than Merrimac soils. The higher pH is likely because of the higher buffering capacity associated with the higher clay content in the Enfield soils. The rest of the soil properties I measured, moisture content, O horizon thickness, bulk density, SOM, SOC, and nitrogen contents, showed no significant differences between the soils (Table 1.2). Cover type (coniferous vs. deciduous) and the interaction between soil and cover had no effect on any of soil properties measured.

Riparian Soils
The six riparian sites used in this study (Figure 1.2) were mapped as either Walpole or Scarboro soils (  (Mausbach and Richardson, 1994).

Soil Carbon Pools
Upland sites (M = 104, SD = 27) had significantly less carbon in the upper 50 cm than riparian sites (M = 163, SD = 80; p = 0.03; Figure 1.3A). This finding is consistent with a study by Davis et al. (2010) who found that carbon pools increase as soils move toward a wetter class (i.e. moderately well drained to poorly drained). The very poorly drained Scarboro sites had a greater SOC pool than poorly drained Walpole, but the difference was only significant when the Scarboro taxadjunct (VRS) was excluded from the data (Figure 1.3B). Carbon pools were similar between Merrimac and Enfield soils (p = 0.927; Figure 1.3A). Davis et al. (2010) found that excessively drained outwash soils of the Windsor series had higher SOC pools than Enfield (well drained). Although Merrimac (somewhat excessively drained) is better drained than Enfield, the difference in hydrology does not appear to affect carbon pools within the upper 50 cm. McLauchlan (2005) found that soil texture is not a significant factor in SOC accumulation across several sites with grassland vegetation.
No significant differences in SOC were detected between upland sites with deciduous vegetation and those dominated by conifers.

Upland Vegetation and ESDs
Species richness was similar between Merrimac and Enfield soils (p = 0.72; Appendix II) and between sites with coniferous and deciduous cover (p = 0.81).
Upland sites of both soil types contained mostly facultative and facultative upland species (Figure 1.4). A total of 9 tree species were observed in the upland canopy stratum, which for this study, was defined as woody vegetation greater than 10 m. The deciduous sites were classified as oak woodlands in the Rhode Island Ecological Communities Classification (RIGIS, 2014). The majority of canopy species were oaks and pines, with the exception of Acer rubrum (red maple), which was a major canopy species at several sites on both upland soil types. A. rubrum is known to be a generalist species that grows on a variety of soil types that is also tolerant of drought and shade (Fergus and Hansen, 2005). Since it can tolerate a vast majority of environmental conditions, the presence of A. rubrum alone, does not provide any insight for differentiating silty versus sandy upland glaciofluvial sites.
The coniferous sites used in this study were identified as plantation and ruderal forest in the Rhode Island Ecological Communities Classification (RIGIS, 2014).
Coniferous Merrimac sites were dominated by P. strobus, but also contained several hardwood species analogous to the deciduous sites. P. strobus has been known to invade disturbed sites, such as abandoned fields or pasture, and mature to old growth forest (Hibbs, 1982;Abrams, 2001). Pinus rigida (pitch pine) was a major constituent of the canopy at one Merrimac site (BZM), likely the result of the large fire which took place in much of western Rhode Island in the early 1930s (Kivela, 2009;Dupree, 2012;personal communication). P. rigida has thick bark, serotinous cones, and is capable of stump-sprouting making it highly fire-adapted (Fergus and Hanson, 2005 where at the third, Quercus coccinea accounted for the most cover. Both of these species fall within the red oak category of oak species in New England (Fergus and Hanson, 2005). Red oaks are defined by having pointed tipped leaves and can thrive under a variety of soil types (Fergus and Hansen, 2005). These species are intolerant of shade and also hybridize regularly making field identification difficult (Fergus and Hansen, 2005). Therefore using the presence of one red oak species over another to support ecological site concepts is limiting. The subcanopy tree and shrub strata for the Merrimac sites was similar between cover types and was mainly composed of young hardwood trees and shrub species in the heath family Ericaceae. The species Ericaceae are known for tolerating highly acidic and nutrient poor soils, such as those observed in this study (FEIS, 2015

Riparian Vegetation and ESDs
Average total species richness did not differ between upland and riparian soils  Overall, the response of the vegetation to selective logging seems site specific.
The two Merrimac harvested sites both had higher herbaceous cover, but at FPH the higher cover was the result of several species colonizing canopy gaps, and at PTH it was exclusively D. punctilobula. Reader and Bricker (1992) found that selective harvesting had no short or long term effects on herbaceous species loss following selective cutting. Similarity in species richness between the selectively harvested sites and the controls (p=0.286) supports this finding. Based on the data collected from P.
strobus tree cores, harvesting also has no effect on tree productivity. Although harvesting removes competition, more time may be required to observe a response in tree production.
Significantly less litter deposition and more emergent vegetation were observed at all three harvested sites when compared to their paired controls ( at PTH and FPH than their controls, but deadfall at YWH was almost twice the amount measured at YWC. The high amount of deadfall at YWH was due to a small tree that fell within one of the plots which greatly influenced the data. When this data point was removed from the dataset, it was determined that deadfall was slightly higher at control sites but statistically similar between treatments (p=0.175). Although I concluded that selective harvesting decreases litter deposition and increases emergent vegetation production, harvested sites did not differ in organic horizon thickness, SOC within 50 cm, nitrogen, or pH. Therefore, it can be concluded that harvesting reduces carbon additions, but the soil-carbon dynamics of these particular soils show resilience to this disturbance 15 years after selective harvest.

Riparian Nutrient Enrichment
In-growth cores removed after two years of N additions indicated higher fine root biomass in the high treatment plots (p=0.006), but no difference was detected between the low treatment plots when compared to the control (p=0.135). Other studies have also recorded short term increases in fine root production following nitrogen additions in upland soils (Safford, 1974;Pregitzer et al., 1993;Hendricks et al., 2000). Although these results support the findings of these studies, Yuan and Chen (2012) determined that, in wetlands, N enrichment was not important in influencing root production. Yuan and Chen (2012) attributed the negative response of root production to N-additions to high nitrogen content in the wetland systems (greater than 1%). Total fine-root biomass generally decreases with increasing nitrogen availability (Nadelhoffer, 2000). If root biomass decreases with increased nitrogen, but production increases as the data from the riparian plot enrichment suggests, then root turnover must also increase (Nadelhoffer, 2000). Nadelhoffer (2000) found that higher turnover is due to higher N concentrations in fine roots (Hendricks et al., 2000), which increases root metabolism and thus N cycling rates.
Although it seemed likely that faster turnover of fine roots would increase soil respiration, no response in respiration was detected between the three treatments in 2012 (p=0.460) or 2013 (p=0.283; Table 1.5). Soil respiration was highest during the months of July through August, but was similar between treatments (Appendix 3). No significant difference was observed between the amount of emergent vegetation between the control plots and the two treatments (Table 1.5). In 2013, the VRS site had significantly less emergent vegetation than both BZS (p=0.004), and the HLS site (p=0.010). In 2014, the same difference was observed between sites (p=0.003). As noted earlier, the VRS pedon did not meet the classification requirements for the Scarboro soil series as it was mapped. This site also had significantly lower respiration rates in 2012 when compared to HLS (p=0.013) and BZS (p=0.044). VRS also had the lowest root biomass, but the difference was not statistically significant from the other riparian sites (p=0.484). The different morphology found at VRS could be the cause of lower above and belowground biomass production, which may have reduced respiration from the soil.

SUMMARY AND CONCLUSIONS
Merrimac or Enfield soil components with either deciduous or coniferous cover were compared to determine whether or not these soils are different ecological sites. The presence of deciduous, coniferous, and mixed forests on both sandy and silty upland glaciofluvial soils indicates that the forest cover type cannot be explained by the soil type alone. The presence of deciduous and coniferous stands on both soil types is likely the result of different disturbance regimes. Since oaks are drought tolerant, adapted to fire, and can colonize sites with poor nutrient conditions, oak dominated stands represent a state in which one of these disturbances occurs in high frequency (Abrams, 1992). Where drought and fire are absent and nutrients are plentiful, a coniferous state will be more likely. It is also apparent that many of the coniferous stands were planted. Either way, I believe the coniferous and deciduous communities observed in this study represent two different states or community phases within one upland forest ecological site.
The sites used in this study showed that even though the Merrimac soils are sandier and better drained than Enfield, the similarity in vegetation composition and tree productivity indicate that these soils have similar ecological potential. The slight differences in species composition that were observed between these soils was due to variability in species distribution or competition, not because the site conditions were limiting. The similarity in vegetation composition could be due to the high amount of precipitation in southern New England relative to the range of precipitation these species thrive under. Similarity in the tree production data also supported similar ecological potential between these soils. Typically, herbaceous and shrub production are also measured to differentiate ecological sites. Since these variables were not measured in this study, there is still a chance that they should be different ecological sites.
Following the selective harvest of glaciofluvial upland sites, soil dynamic properties related to carbon were determined to be resilient to change. I concluded that the 50% removal of overstory trees decreases carbon additions from litter, but that this reduction does not significantly impact the distribution of soil carbon within the soil profile over the 15 years since selective harvest. The vegetation response to selective logging seems to be site specific. Canopy openings can lead to species such as Dennastaedtia punctilobula to outcompete other understory species, but this occurrence is haphazard, and the colonization of openings depends on a variety of factors including what species already occupy the site. Since the sites chosen were logged within the last 15 years, it may be that not enough time has passed to affect the properties recorded in this study.
I also investigated differences in ecological sites and dynamic soil properties in wetland forests. Both the Walpole and Scarboro riparian sites had stands of Acer rubrum, but there were observable differences in the understory species composition that support separate ecological sites for these soil systems. Carex stricta and Symplocarpus foetidus were the two species that seemed to indicate the very poorly drained conditions of the Scarboro soils. In contrast, tree production did not support different ecological sites, but as mentioned earlier, herbaceous and shrub production may help differentiating these sites. The higher water table found at the Scarboro sites is also the likely cause of increased organic matter accumulation and thus the higher SOC pool that was observed in comparison to the other soils used in this study. Better drainage in the Walpole soils increases aerobic decomposition which explains why these soils lack a thick organic surface horizon and have lower SOC pools.
In riparian zones, I tested whether nitrogen additions alter dynamic soil properties in Scarboro soils and found that N was a limiting nutrient for plant growth.
Although the aboveground biomass measurements did not support this conclusion, the increase in root growth showed that N could increase plant production. No conclusions could be made on how nitrogen additions influence short term riparian soil respiration.
Average nitrogen in the upper 50 cm was similar between the Walpole and Scarboro soil types. Since my findings suggest they are different ecological sites, it is possible that the response of soil-vegetation dynamics to nitrogen enrichment in Walpole could differ from Scarboro.  Table 1.5: Multiple pairwise comparison (Tukey's) results on biomass and soil respiration compared between plots used in nitrogen enrichment experiment. Root biomass was measured using in-growth cores at 5-20 cm below the soil surface (Ricker et al., 2014). P values in bold indicate a significant difference between the treatment and the control. N=9.    (12) and riparian (6) soils. A percentage is reported for each indicator class that was calculated from the total number of species. A significant difference was detected between upland and riparian sites for all indicator classes. Indicator classes determined from the National Wetland Plant List (Lichvar, 2012 where the community in the Pishagqua soils mostly consisted of deposit feeders (78%). The differences in soils and geomorphic setting likely influenced the carbon pools and resulted in the observed differences in macroinvertebrate assemblages of the two soil types. In both subaqueous soil types, invertebrate density was reduced in the dredged soils, with a 97% difference observed in Massapog and a 71% decrease in Pishagqua. In the Massapog soils, eelgrass colonization following dredging induced a shift from dominantly filter feeding organisms to deposit feeders. I found that water depth influences the presence of eelgrass. I believe that in most cases dredging lagoon bottom soils will inhibit their ability to support eelgrass because depth will be too great. In contrast, dredging in the flood-tidal delta could inhibit or induce eelgrass presence. For both Massapog and Pishagqua dredging increased depth which resulted in finer textures and greater SOC accumulation.

INTRODUCTION
Estuarine intertidal and subtidal wetlands are important components of coastal ecosystems as they provide services such as habitat for benthos, sinks for carbon and pollutants, and sites for recreational and commercial fisheries (Bradley and Stolt, 2006;O'Higgins et al., 2010;Wieski et al., 2010;Sousa et al., 2012). Although shoreline counties of the U.S. only account for 10% of the nation's total land area, the population of these counties has increased by 40% since 1970 and currently accounts for 39% of the total population (NOAA, 2013). Due to their close proximity to developed areas and their resource value, estuarine ecosystems are subject to a variety of anthropogenic disturbances. Coupled with their limited areal extent, these wetlands may be recognized as threatened in respect to their soils and associated ecosystem services (Drohan and Farnham, 2006). Inventorying and monitoring these systems is essential in order to understand and preserve the ecosystem services provided by tidal wetlands.

Subaqueous Soils and Dredging
Over the last two decades soil scientists have been studying shallow subtidal estuarine substrates as soil (Demas, 1993;Demas and Rabenhorst, 1999;Bradley and Stolt, 2003;Stolt and Rabenhorst, 2011). These substrates are recognized as subaqueous soils because they undergo pedogenesis (Demas and Rabenhorst, 1999) and support aquatic vegetation (Bradley and Stolt, 2003). Estuarine subaqueous soils occur in the subtidal zone of protected coves, bays, inlets, and lagoons (Bradley and Stolt, 2003). In a manner similar to subaerial soils, soil-landscape relationships exist in subaqueous environments (Demas and Rabenhorst, 1999;Bradley and Stolt, 2003).
These relationships have been used to classify subaqueous soils and map soil units within selected estuaries along the eastern U.S. (Demas, 1993;Bradley and Stolt, 2003;Payne, 2007;. Subaqueous soils provide valued ecological and economic services, and therefore soil interpretations have recently been developed such as suitability for shellfish aquaculture, eelgrass restoration, and upland placement of dredge materials (Pruett, 2010;Salisbury, 2010).
In estuarine subaqueous soils, anthropogenic alterations including dredging activities may influence ecosystem processes by altering soil dynamics. Studies have shown that dynamic soil properties such as organic matter content, pH, and particle size influence shellfish production, eelgrass distribution, and water quality (Bradley and Stolt, 2006;Payne, 2007;Salisbury, 2010). For example, Salisbury (2010) found a positive relationship between eastern oyster (Crassostrea virginica) growth rates and sand content of soils and a negative relationship between growth and organic carbon content. Likewise, a study on the relationship between particle size and flounder distribution revealed that small juvenile flounder (<40 mm) are selective of finegrained habitats, while larger juveniles (>40 mm) preferred coarser grained soils (Phelan et al., 2001). Subaqueous soil dynamics are highly dependent on the amount of energy present in the system, which is often depth dependent. Low-energy depositional environments, such as lagoon bottom and bay-floor soils, tend to have a finer particle size distribution, whereas high-energy features, such as washover fans and flood tidal deltas, tend to have more sand and a coarser particle size distribution (Bradley and Stolt, 2003). Currently, no research has been done to quantify the resistance and resilience of soil properties such as carbon content to dredging activities.

Estuarine Salt Marshes, Ditching, and Climate Change
Salt marsh soils are intertidal and often form on the fringe of brackish or saltwater estuaries at the land-water interface. Competition and plant physiological tolerances create distinct zones of plant cover in New England salt marshes (Bertness and Ellison, 1987). The low marsh is inundated by daily high tides and is typically dominated by the salt tolerant Spartina alterniflora (Bertness and Ellison, 1987;Bertness et al., 1992). The portion of the marsh that is only flooded during the highest tides, the high marsh, is typically covered by the less salt tolerant Spartina patens (Bertness and Ellison, 1987;Bertness et al., 1992). Tidal salt marshes are the nursery grounds for a range of estuarine fish and wildlife while providing ecosystem functions such as groundwater filtration, carbon sequestration, and upland storm protection (Nixon and Oviatt, 1973;Boesch and Turner, 1984;Valiela et al., 2000;Wieski et al., 2010;Sousa et al., 2012). Sea level rise, ditching, nutrient loading, and other human induced disturbances, however, have altered salt marsh plant community dynamics and ecosystem services . Therefore, it is critical to monitor the impact of anthropogenic effects and disturbances to salt marsh soil ecosystems.
Humans have been ditching New England salt marshes since the early 17 th century to increase yields of S. patens and to mark property boundaries (Rozsa, 1995).
During the early 18 th century land managers increased ditching practices with the intention to drain pools at the soil surface, potentially eliminating mosquito larval habitat (Resh and Balling, 1983;Rozsa, 1995). Ditches have also been constructed to increase tidal flooding to the marsh providing access for predatory fish (Resh and Balling, 1983). Previous ditching practices have led to changes in tidal inundation patterns which has resulted in changes in soil properties which influence salt marsh plant composition (Resh and Balling, 1983;Vincent et al., 2013a;2013b). In a study on the Pacific coast, Resh and Balling (1983) found that only soils within 4 m of the ditch were drained and recharged by daily high tides. The change in the hydrology at ditched marshes resulted in a salinity gradient which increases with distance from ditch (Resh and Balling, 1983). Vincent et al. (2013a) noted that these changes in soil conditions influence the distribution of salt marsh vegetation. For example Salicornia europea and short-form S. alterniflora occupy zones outward from the ditch margin where sulfide accumulation and highly reduced conditions prohibit the colonization of high salinity intolerable species such as S. patens (Vincent et al., 2013a;2013b). How ditching affects other dynamic soil properties that influence important ecological processes is presently unknown.
Climate warming is another factor which may influence salt marsh soil and vegetation dynamics. Global climate models project global surface temperatures to increase 1.5 to 4 °C by 2100 (IPCC, 2007). Projected surface temperature increases have been simulated using passive open-top warming chambers (Marion 1996;Marion et al., 1997;Gedan and Bertness, 2010). These chambers trap air near the soil surface, stimulating the greenhouse effect by increasing soil temperature as much as 3 °C (Marion 1996;Marion et al., 1997). Surface temperatures have been shown to influence vegetation community assemblages and biogeochemical cycles in salt marshes Gedan and Bertness, 2010). Total soil respiration is the result of the production of CO2 from microbial decomposition, diffusion through culms, and root and rhizome respiration (Howes et al., 1985;Wigand et al., 2009). Richardson (2006) and Davis et al. (2010) found a positive correlation between temperature and soil respiration in New England forested uplands and palustrine wetlands. How such an increase in soil temperature will affect soil carbon dynamics in salt marshes is in question. Soil temperature increase may stimulate respiration from microbial decomposition, but this increase may be surpassed by CO2 uptake from increased aboveground biomass production (Chumura et al., 2003;Davidson and Janssens, 2006).

Ecological Site Descriptions
The Natural Resource Conservation Service (NRCS) ecological site inventory is a framework developed for inventorying soil, vegetation, and abiotic features; delineating landscape scale units that share similar responses to management activities or disturbance processes; and estimating ecosystem services that can be expected from particular soil/vegetation combinations (Townsend, 2010). An ecological site is a defined as a distinctive kind of land having recurring soil, landform, geological, and climate characteristics that produces distinctive kinds and amounts of vegetation, and responds similarly to management actions and natural disturbances (NRCS, 2013).
Soil-landscape units can be used for distinguishing ecological sites in these systems because they provide a mechanism for grouping soils that occur in a similar landscape setting. Within each ecological site, management or disturbance is the mechanism which changes soil-vegetation dynamics away from a referenced community resulting in different "states" (Briske et al., 2005). Thus, an ecological site is composed of a reference community and a series of states that have transitioned from the reference state. Currently, no ESDs exist for subaqueous systems or salt marshes.
The objectives of this study were: i) to identify concepts for distinguishing ecological sites in selected salt marsh and subaqueous soils, ii) to quantify the effects of dredging, ditching, and warming on soil and vegetation dynamics of these ecosystems, and iii) to elucidate the effect of soil-vegetation community relationships relative to dynamic soil properties.

Subaqueous Sites
The Ninigret Pond, also referred to as Charlestown Pond, is coastal lagoon isolated from the Block Island Sound by a barrier spit. In the 1950s, a breachway was constructed in order to maintain boat traffic into the pond, which increased tidal force and thus sedimentation (Conover, 1961). Soil materials from the breachway channel were removed to maintain navigable waters in 2008 along with material from the sedimentation basin for eelgrass restoration (Figure 2.1). Both the control and dredged areas are part of the flood-tidal delta flat (Bradley and Stolt, 2003). Dense eelgrass (Zostera marina) beds (>95% cover) occupy the dredged site whereas the control site is totally barren. Soils in both of these areas were mapped as Massapog (fine-sandy, mixed, mesic Fluventic Psammowassents. Subaqueous soils of Point Judith Pond were mapped by Mapcoast and the Natural Resource Conservation Service in 2010 (Mapcoast, 2010;Pruett, 2010). This pond also has a barrier spit, but differs from other estuaries in this study because it was formed from individual ice-block basins, which were flooded by sea level rise (Pruett, 2010). Point Judith is subject to daily boat traffic through a permanent inlet, which was created in 1909, to allow large vessels into the pond through the southern end Histic Sulfaquents). The difference in the morphology of these two soils is likely the result of their environmental setting (Wood et al., 1989). The Narrow River has small fluvial marshes occupying the upper margins of the estuary, whereas at Winnapaug the marshes occur behind the back-barrier component of the spit (Wood et al., 1989).
Each site was investigated in an initial field reconnaissance to confirm soil types and to choose an area of the marsh for study.

Experimental Warming Site
Four pentagonal open-top warming chambers (OTCs) and four control plots, all 1 m 2 , were installed at Fox Hill Salt Marsh on Conanicut Island located in lower Narragansett Bay (Figure 2.3). This marsh is a 10 ha transitional marsh and is relatively pristine in terms of nutrient loading (Wood et al., 1989;Wigand et al., 2009  analysis. Benthic macroinvertebrates samples were sorted and identified to the species level when possible using basic dichotomous keys (Smith 1964;Weiss, 1995).

Salt Marsh Ditching
Within each estuary, three salt marsh ditches were chosen to capture the variability of soil properties and vegetation community attributes under study (Table   2.2). Two 15 m transects on each side and perpendicular to the ditch were established for sampling. Along the marsh ditch transects, sampling and field collection took place at distances of 0, 1, 5, and 15 meters from the ditch margin on either side of the ditch (Figure 2.4). At each sampling point, vegetation composition was recorded as a percentage within a 0.25 m 2 quadrat and the average height and density of each species was recorded. Soil was collected and described in the field using a Macaulay peat sampler to 1 meter when possible (Bradley and Stolt, 2003;Twohig and Stolt, 2011). Peat thickness measurements were estimated by probing the soil with a metal rod. Relative surface elevations were also recorded using a rod and level along ditch transects. Marsh soil samples were stored in a freezer until analysis.

Soil Classification and Analysis
All laboratory soil analyses were conducted following standard soil survey methodology outlined in the Soil Survey Laboratory Methods Manual (Soil Survey Laboratory Staff, 2004). For each genetic horizon bulk density, soil organic carbon content, soil organic matter, initial and incubation pH, soil salinity, and particle size was determined. For organic samples rubbed fiber content was calculated for subordinate distinction determination. Calcium carbonate was also measured for subaqueous samples.
Bulk density was determined from Macaulay and vibracore samples taken from each horizon. Samples of a known volume were oven-dried at 105 °C. Oven-dry soil weight was divided by the volume yielding bulk density (g cm -3 ). Soil organic matter content and calcium carbonate were determined via loss on ignition (LOI) (Heiri et al., 2001). Total organic carbon was calculated using organic matter LOI at 550 °C assuming an organic carbon-organic matter ratio of 0.5 (Nelson and Sommers, 1996;Pruett, 2010). Calcium carbonate was determined by subtracting the soil dry weight after combustion at 1000 °C from 550 °C dry weight and dividing the product by the percent (59.95) of CaCO3 that is lost as carbon dioxide through combustion (Heiri et al., 2001;Payne, 2007;Salisbury, 2010). Soil pH measurements were taken using an Accumet pH ATC combination electrode with silver/silver chloride reference. A 1:1 slurry of soil and water was mixed immediately after returning from the field or after stored sampled thawed for measurements. Moist conditions were maintained and incubation pH was recorded weekly for 16 weeks to determine potential acidity and identify sulfidic materials (Soil Survey Laboratory Staff, 2004;Payne, 2007). Soil salinity was measured in a 1:5 slurry of soil and water using an Oakton WD-35607 hand held conductivity meter (He et al., 2012). Particle size distribution was conducted using air dry soil from Macaulay and vibracore samples.
Soil was wet sieved through a No. 270 standard sieve to determined sand content.
Sands fractions were separated by dry sieving sand content samples on a sieve shaker for 5 minutes. Clay content was determined using the pipette method (Soil Survey Laboratory Staff, 2004;Payne, 2007). Silt content was calculated by subtracting the oven dry clay and sand weights from the total oven dry sample weight. Soil carbon pools were calculated for the upper 50 cm in Mg ha -1 using bulk density and carbon content parameters (Compton et al., 1998;Payne, 2007;Davis et al., 2010). Pedons were classified using Keys to Soil Taxonomy, 12th edition (Soil Survey Staff, 2014).

Experimental Warming of Salt Marsh Soils
Three core samples (98.2 cm 3 ) from the upper 15 cm and three 10 cm 2 vegetation samples were collected randomly in May 2013 (preseason) and at the peak of the growing season in August 2013 to determine biomass production (Windham, 2001;Gedan and Bertness, 2010). Vegetation samples were used to determine shoot density, average shoot height, and total aboveground biomass (g cm -2 ). Shoot density was calculated by counting the number of live shoots within each 10 cm 2 quadrate. A subsample of 30 shoots were measured to determine average height. All live shoots from each quadrate were dried in an oven at 60 °C for total aboveground biomass.
Below ground biomass was determined by separating roots and rhizomes with tweezers, which were then soaked in a 0.5 g L -1 calgon solution, rinsed, and dried at 60 °C.
Thermochron iButton 1921G loggers with ±0.5 °C accuracy (Maxim Integrated Products, Sunnyvale, CA), were set to record soil and air temperatures hourly during the study period. Two loggers were installed in each plot, one 10 cm below the soil surface and one 15 cm above.
One 25 cm diameter PVC collar was installed 2.5 cm into the soil surface to form a seal and left for the duration of the study. In-situ soil CO2 respiration losses were measured monthly at each plot using a Li-Cor 6262 infrared gas analyzer (Li-Cor, Lincoln, Nebraska), which was affixed to the PVC collars. Once sealed, CO2 concentration was recorded from the analyzer every 10 seconds for a minimum of 5 minutes. To determine CO2 efflux, a linear regression was fitted to the final 60seconds of the measured CO2 concentrations plotted as a function of time (Davis et al., 2010;Ricker et al., 2014). Pressure and temperature recorded from the analyzer along with the chamber volume were used to calculate moles of CO2 per mole of air using the Ideal Gas Law. The moles of CO2 per mole of air was multiplied by the rate of CO2 flux and divided by the chamber area to yield µmol CO2 m -2 sec -1 .
Three nylon litter bags containing 5 g of clipped and oven dried aboveground vegetation, taken from onsite, were installed at the soil-air interface within each plot to estimate decomposition. Each bag was affixed to the soil prior to the growing season and removed in November. Following removal, partly decomposed materials were removed from each bag, rinsed and dried at 60 °C. Decomposition rates were plotted as a function of time using the difference between the initial oven dry weight of litter bag biomass and the oven dry weight of biomass following removal.

Statistical Analysis
Species composition and richness were calculated for the macroinvertebrate samples and each species was grouped into a functional feeding group (Weiss, 1995).
The average density (individuals m -2 ) and the total number of species within each feeding group were compared between treatments and soil types using paired t-tests.
Soil data for subaqueous sites were weighted by horizon thickness and averaged for the upper 50 cm for comparison.
The upper 50 cm averages were also calculated for soils sampled for the ditching comparison. For each ditch, these attributes along with site and vegetation properties were averaged for each distance sampled from the ditch margin (0, 1, 5, and 15 meters). Analysis of variance (ANOVA) was used to detect differences between sampling locations within each soil type. For the warming experiment, paired t-tests were used to compare properties between warmed treatment and the control

Soil Characterization and Dynamic Soil Properties
At all three sites the particle size distribution was slightly finer in the dredged soil (Appendix 4). For a given landscape unit, increasing the water depth by dredging likely diminished flow rates relative to the adjacent natural soil and thus allowed for finer particles to settle out in the dredged areas. Although post dredging deposition was apparent, dredged sites always had greater water depth than the control (Table   2.1) suggesting that finer materials will continue to be deposited at the dredged areas.
The increase in depth may have promoted the growth of eelgrass in Ninigret Pond, and may have inhibited it in Point Judith Pond. At Ninigret Pond, the control site is quite shallow (0.5 m) and eelgrass is absent. Where dredging has increased depth to approximately 1.2 m, eelgrass is plentiful. The presence of eelgrass in the dredged site at Ninigret may in part also explain the slightly finer texture as eelgrass is known to trap sediment. The opposite trend between dredging and eelgrass was observed at Point Judith Pond (Table 2.1). Depth at the control site was 0.6 m and eelgrass cover was approximated at 90% cover. In the dredged channel, which was 3 m deep, there was no eelgrass. Bradley and Stolt (2006) found that eelgrass rarely occurred in southern New England subaqueous soils with water depths less than 50 cm or greater than 1.8 m.
Both Ninigret Pond sites were classified as Sulfic Psammowassents; having sulfidic materials within 100 cm of the mineral soil surface and a sandy family particle size class (Appendix 4). In both the control and dredged areas, buried horizons were described at approximately 55 cm (Appendix 4). The slightly finer textures combined with the presence of eelgrass is likely the cause of greater carbon accumulation in the dredged soils. Soils having Z. marina exhibit greater carbon contents from the greater abundance of marine organisms and plant debris near the soil surface (Bradley and Stolt, 2006;Millar et al., 2015). The dredged soils also showed a larger change in pH following the 16 week incubation, which may indicate a greater accumulation of sulfides within the profile (Table 2.3A). Organic matter is required for sulfidization (Fanning et al., 2010) and the higher organic matter in the dredge material may have promoted the accumulation of sulfides resulting in lower incubation pH values in the Ninigret dredged soil. In contrast, although Payne (2007) found a significant positive relationship between organic carbon and total inorganic sulfide contents, the relationship between inorganic sulfide content and incubation pH was not significant. Payne (2007) argued that buffering of the pH from carbonates, clay, and organic matter may have confounded this relationship.
Similarly to the Ninigret soils, the Point Judith control pedon was classified as Sulfic Psammowassents (Appendix 4). However, the Point Judith dredged soil was classified as Sulfic Fluviwassents. Unlike the control site, the dredged soil at Point Judith had horizons finer than loamy fine sand within the control section (25-100 cm; Appendix 4). These dredged soils contained less sand (3.4%) and more silt (3.5%) and clay (0.1%) than their adjacent controls. Although both soils were dominated by fine sands, the dredged soil unexpectedly had more very coarse, coarse, and medium sand sized particles within the sand fraction. Point Judith dredged soils also had a greater bulk density, electrical conductivity, SOC and CaCO3 pool than the control soils.
Greater pH change was observed in the control soils suggesting higher sulfide content or less pH buffering.
The Wickford Harbor control and dredged soils both averaged more than 18% clay within the control section and therefore classified as fine-loamy Typic Sulfiwassents (Appendix 4). Sulfiwassents, such as the Pishagqua series, are known to develop in low energy, soft-bottom landscape settings (Payne, 2007). The Pishagqua control soils at Wickford had almost twice the SOC compared to the Massapog control soils of the other two sites (Figure 2.5). Both the control and dredged soils at Wickford were dominated by fine sized particles although the dredged soils contained 30% more silt and 4% more clay than the adjacent control (Table 2.4). Sand content was much greater in the control soil (35%), but in both soils fine and very fine sands were most abundant. These soils exhibited dark colors and enough of a change in pH during the 16 week incubation to suggest sulfidic materials were present within the soil profile.
Within the Wickford dredged soils, incubation pH change was 0.45 pH units greater than the change observed in the control soils (Table 2.3). Electrical conductivity, and SOC and CaCO3, contents were also greater than the control. The Wickford dredged soils were also more fluid and had a lower bulk density than the control; likely the result of high SOM and low sand content in the dredged materials.

Subaqueous macroinvertebrate assemblages and ESDs
The Wickford and Ninigret dredged and control sites were chosen for ecological site comparison because they represent two common and recognizably different landforms in estuarine systems (Table 2 Total invertebrate density and species richness were greatest at the Ninigret control site (Table 2.5). Of the twelve species observed at the Ninigret control, the majority (50%) were deposit feeders (Table 2.5). Although the deposit feeders were the most diverse community at the Ninigret control site, the majority of individuals were filter feeders with Clymenella torquata (common bamboo worm) as the most abundant species (Appendix 5). This non-motile species lives in tubes composed of cemented sand grains (Weiss, 1995). C. torquata is common to intertidal sandy mud flats and is widely distributed along the western coast of the Atlantic with densities reaching up to 150,000 m -2 (Sanders et al., 1962;Mach et al., 2012). Sanders et al. (1962) found an inverse relationship between C. torquata and bivalve abundance, although this trend was not observed in this study. Mya arenaria (soft-shell clam), a commercially important bivalve, was the second most abundant filter feeder species at the Ninigret control site (409 m -2 ). M. arenaria distribution is limited in highly fluid, fine grain soils which may collapse against shell valves (Abraham and Dillon, 1986).
Unlike the control, no filter feeders were found at the Ninigret dredged site.
Although this was the case, Clymenella torquata tubes were present in the sample suggesting that C. torquata may be present at the dredged site as well. Deposit feeders were the most common community in the Ninigret dredged soil. Nephtys picta and species within the family Ampeliscidae were the most common individuals observed.
Amphipods of the family Ampeliscidae dwell in sediment constructed tubes and have been documented to be a major prey item for juvenile winter flounder diet (Stehlik and Meise, 2000).
At both the Wickford control and dredge sites, Ilyanassa obsolete and Gemma gemma dominated benthic community composition (Appendix 5). Ilyanassa obsolete is a deposit feeding gastropod that is common on intertidal and shallow subtidal mud and sand flats (Weiss, 1995). This species was the most abundant species observed at the Wickford control site averaging 574 m -2 . Next in abundance at the control site, was Gemma gemma averaging 104 m -2 . Unlike the control site, more Gemma gemma (87 m -2 ) were observed than Ilyanassa obsolete (61 m -2 ) at the dredged site. Gemma gemma is a filter feeder common throughout New England estuaries that is a major constituent in the diet of shore birds during winter months (Sanders et al., 1952). Well sorted fine soils are preferred habitat for Gemma gemma because these soils retain seawater in pore spaces throughout low tides (Sanders et al., 1952).
In both subaqueous soil types, invertebrate density was reduced in the dredged soils (Table 2.5). This was unexpected for the Massapog soils since species richness and abundance is typically greater in eelgrass habitats than in unvegetated soils (Heck et al., 1995;Lee et al., 2001). Van Houte- Howes et al. (2004) explained an edge effect that may occur adjacent to eelgrass bed, and that macroinvertebrates within the eelgrass bed may be limited by the dense mat of roots. The effect of dredging did not appear to have as much of an effect on the invertebrate community at Wickford. Both treatments at Wickford exhibited similar invertebrate composition, but differed in the distribution of individuals. The contrasting invertebrate communities observed between the Massapog and Pishagqua control sites is likely the result of their different geomorphic setting. Although these soils were dredged around the same time, the invertebrate assemblages of the Pishagqua soils observed at Wickford also seem to be more resilient to dredging. These two findings support the placement of these soils into different ecological sites.

Soil Characterization and Peat Thickness
The soils observed at Narrow River had organic soil materials greater than 130 cm (Figure 2.6) and therefore are more representative of the Ipswich series (euic, mesic Typic Sulfihemists), then the Pawcatuck series (sandy or sandy-skeletal, mixed, euic, mesic Terric Sulfihemists) that they were mapped as (Table 2.2). Peat thickness ranged from 120 to 185 cm and did not differ between sampling locations (Figure 2.6; p=0.35). At two sites peat thickness was greater further from the ditch, but at the third site (NR3) peat thickness initially increased and then decreased with distance ( Figure   2.6).
The soils at Winnapaug are representative of the Matunuck series (sandy, mixed, mesic Histic Sulfaquents; Table 2.2). The thickness of the organic surface layer for this soil series ranges from 20 to 40 cm (Figure 2.6). Although some observations were slightly greater, the majority of peat thickness measurements at the Winnapaug marshes fit within this range. As observed at Narrow River, no relationship was determined between peat thickness (p=0.99) and the distance from the ditch (Figure 2.6).

Vegetation and ESDs
At the Narrow River marshes, no statistical differences were observed between species percent cover estimates, bare cover, or shoot density and the different distances from the ditch (Table 2.6). Although this was the case, several trends were observed. Spartina patens was a major occupier of zones 0, 1, and 5 meters from the ditch, but was never found dominating the 15 meter zone. S. patens height and stem density were also lowest 15 meters from the channel. At all three sites the highest percent cover and greatest average height of Spartina alterniflora was observed either at the edge of the ditch or in the 1 meter zone. Distichlis spicata and Salicornia europaea were also found at the three Narrow River marshes, but their abundance was minimal and no trends were observed with distance from the ditch.
Vegetation at the three Winnapaug sites was dominantly S. alterniflora (Table   2.6). Similarly to Narrow River, percent cover of this species was greatest near the edge of the ditch, but did not differ significantly between sampling locations. S. alterniflora shoot density was greater at the edge and 5 meters from the ditch (p=0.02). Likewise S. alterniflora height was greater at the edge and 1 meter zones (p<0.01). Bertness (1984) found that S. alterniflora production was greatest at the seaward edge of the marsh where mussel density and soil nitrogen levels are elevated.
Although neither mussel density nor nitrogen were measured, I believe a similar effect occurs along the ditch margins.
Overall, Matunuck soils at Winnapaug sites averaged 25 percent more bare cover than the Ipswich soils at Narrow River, but the difference was statistically insignificant (p=0.08). Percent bare soil at the Winnapaug marshes was lowest near the edge of the ditch and often highest further from the channel. Total shoot density at the Winnapaug marshes was over 50% lower than at the Narrow River sites (p=0.03).
The main contributor to the difference in total shoots was the density of S. patens, which had significantly lower average values at the Winnapaug marshes (p=0.01).
Cover of S. europaea species was greatest at the Winnapaug marshes, where bare cover was greater than 80%. The high amount of bare soil in the Winnapaug marshes and presence of S. europaea is indicative high soil salinity, which prohibits the colonization of low salt tolerant species such as S. patens (Bertness et al., 1992). This also explains why S. patens was rarely found in the 15 meter zone at the Narrow River marshes. Plant height was observed to be similar between the Winnapaug and Narrow River sites.
Based on the kind of vegetation present, the Ipswich soils (Histosols) studied at Narrow River are the same ecological site as the Matunuck soils (Entisols) at Winnapaug. The data collected suggests that productivity and distribution of individual salt marsh species is based on soil salinity, which is often a function of the distance of the pedon to the marsh-water interface. Within a given soil map unit the variability of soil salinity is too high to identify where salinity limits productivity. shift from one community to another over time due to slight alterations in tidal fluctuations, without management or disturbance. Therefore, one ecological site should encompass the salt marsh soils reported in this study.

Dynamic Soil Properties
As mentioned previously, the primary difference between the soils at Narrow Rivers (Ipswich) from those at Winnapaug (Matunuck) is the thickness of organic materials (Figure 2.6). Since the Narrow River soils contained a thick organic surface, average SOM was high (40.6%; Figure 2.7). I found that the organic materials at Narrow River were mainly composed of hemic soil materials averaging 22% rubbed fibers (Figure 2.7) with no significant differences between sampling points (p=0.22;  SOM did not follow the same trend with distance from the ditch as observed at Narrow River (p=0.17). Unlike Narrow River, the organic materials sampled from the Matunuck soils at Winnapaug were primarily sapric, averaging 11% rubbed fiber.
Since rubbed fiber volume can be an indicator of soil decomposition, the Matunuck soils observed in this study were more decomposed than the Ipswich soils. Less vegetation production at Winnapaug (

Carbon Losses
Soil temperature showed a positive relationship to soil respiration (p<0.01), but the relationship was not as strong (R 2 =0.40) as previously observed by Richardson (2006) and Davis et al. (2010)  In both years, percent loss was greater for the warmed treatment than the control, but was not statistically significant (p=0.114). In 2012, litter bags were in situ for 107 days, and percent loss averaged 50.4% for the warmed treatment and 47.8% for the control. Similarly, bags were installed for 110 days in 2013, but percent loss was greater in the both the control (58.4%) and warmed treatment (65.1%) than in the previous year. These values are similar to those found by Charles and Dukes (2009) who also found that warming increased salt marsh grass decomposition.

Carbon Additions
Pre-season aboveground and belowground biomass, recorded in May 2013, was similar between the warmed plots and the control (p=0.99;  Table 2.10) suggesting that projected temperature increases will possibly only increase S. patens aboveground production possibly due to stronger warming by OTCs above the soil surface. Projected temperature increases with climate change may also impact belowground processes related to carbon, but the ability of OTCs to increase soil temperature seems insufficient.
These findings are consistent with Gedan and Bertness (2010) who also found that warming increases S. patens aboveground production. Contrasting to these results, Charles and Dukes (2009) did not observed an increase in S. patens production with experimental warming. Although that was the case, they did find that warming increased S. alterniflora production and S. patens stem length (Charles and Dukes, 2009). I also discovered that S. patens stems were significantly longer in the warmed plots than in the control (p=0.03; Table 2.9). Stem density measured in 2013 was also used as an indicator of S. patens production. Peak season stem density averaged 11,425 stems per m 2 for the warmed treatment, which was 28% more than the control, but not statistically different (p=0.166; Table 2.9). Increased stem density could lead to greater marsh accretion by trapping more sediment from tides, which would help salt marshes respond to projected seas level rise (Leonard and Croft, 2006;Charles and Duke, 2009 in the chambers. This was likely the result of shading by increased biomass and increased stem length. OTC plots had higher aboveground production, but this increase in carbon additions to the soil may be offset by increased decomposition.
Since observed soil respiration rates were slightly correlated to temperature, increases in carbon additions to the soil could be offset by the higher decomposition rates, with a possible carbon transfer from soil to atmosphere. These effects should be further studied to understand the possible implications on salt marsh carbon budgets.           - - -  -    -   - -  -      - -  - -    -  -