Macro-Tidal Salt Marsh Ecosystem Response to Culvert Expansion

RESEARCH ARTICLE

Macro-Tidal Salt Marsh Ecosystem Response to Culvert Expansion Tony Bowron,1 Nancy Neatt,1 Danika van Proosdij,2,3 Jeremy Lundholm4 and Jennie Graham1 Abstract The purpose of this paper was to examine the vegetative, sedimentary, nekton and hydrologic conditions prerestoration and the initial 2 years post-restoration at a partially restricted macro-tidal salt marsh site. Replacement of the culvert increased tidal flow by 88%. This was instrumental in altering the geomorphology of the site, facilitating the creation of new salt marsh pannes, expansion of existing pannes in the mid and high marsh zones, and expansion of the tidal creek network by incorporating relict agricultural ditches. In addition, the increase in area flooded resulted in a significant increase in nekton use, fulfilling the mandate of a federal habitat compensation program to increase and improve the overall availability and accessibility of fish habitat. The restoration of a more natural hydrological regime also resulted in the die-off of freshwater and terrestrial vegetation along the upland edge

of the marsh. Two years post-restoration, Salicornia europea (glasswort) and Atriplex glabriuscula (marsh orache), were observed growing in these die-back areas. Similar changes in the vegetation community structure were not observed at the reference site; however, the latter did contain higher species richness. This study represents the first comprehensive, quantitative analysis of ecological response to culvert replacement in a hypertidal ecosystem. These data will contribute to the development of long-term data sets of pre- and post-restoration, and reference marsh conditions to determine if a marsh is proceeding as expected, and to help with models that are aimed at predicting the response of marshes to tidal restoration at the upper end of the tidal spectrum.

Introduction

ecological implications, which are preventable and repairable (Sinicrope et al. 1990; Burdick et al. 1997). Salt marsh ecosystem function may be restored passively when a dike is breached during a storm with little to no human interference (Crooks et al. 2002) or through active means by planned removal or modification of a barrier to restore hydrology (Kl¨otzli & Grootjans 2001). Examination of passive restoration sites can provide long-term records about vegetation and geomorphic recovery (Crooks et al. 2002; French 2006; Byers & Chmura 2007); however, this provides limited information about driving and/or limiting variables within the first few years after the breach event that may influence the ultimate recovery of the system (e.g., Kl¨otzli & Grootjans 2001; Able et al. 2008). Both the European and North American experiences have shown that salt marsh reestablishment is not inevitable at all sites following a breach (e.g., Haltiner et al. 1997; French et al. 2000; Kl¨otzli & Grootjans 2001; Williams & Orr 2002). Properly monitored restoration projects can sometimes provide additional information on constraints that may have caused a restoration project to proceed along successional pathways not initially anticipated (Kl¨otzli & Grootjans 2001). Constraints include inadequate sedimentation or hydrology, high groundwater salinity, low nutrient levels, erosion, and lack of appropriate seed source

Over approximately the last 1,000 years, human activities have reduced the amount of salt marsh and free-flowing tidal river habitats in many areas, including Europe (e.g., Tonis et al. 2002; Elias & van der Spek 2006), North America (e.g., Wells 1999; Neckles et al. 2002), and Australia (e.g., Wolanski et al. 2001). Conservative estimates for the Bay of Fundy put the loss of salt marsh habitat between 80 and 85% (Gordon 1989), mostly associated with diking and conversion to agricultural land. Although these activities are of historical and social significance, it is now recognized that the significant loss of habitat, species, and primary productivity that has resulted from the construction of dikes, modern tidal barriers (causeways), and coastal development has had significant

1 CB Wetlands and Environmental Specialists Inc., 37 Royal Masts Way, Bedford,

Nova Scotia B4A 4B1, Canada

2 Department of Geography, Saint Mary’s University, 923 Robie St., Halifax, NS

B3H 3C3, Canada 3 Address correspondence to D. van Proosdij, email [email protected] 4 Department of Biology, Saint Mary’s University, 923 Robie St., Halifax, NS B3H 3C3, Canada

© 2009 Society for Ecological Restoration International doi: 10.1111/j.1526-100X.2009.00602.x

Restoration Ecology

Key words: Bay of Fundy, GPAC protocol, LiDAR, nekton,

restoration monitoring, RSET.

1

Marsh Response to Culvert Expansion

(e.g., Burdick et al. 1997; Haltiner et al. 1997; Williams & Orr 2002; French 2006). Tidal wetlands are valuable ecosystems and their restoration has become a common practice in the Gulf of Maine over the past two decades (Short et al. 2000; Neckles et al. 2002; Konisky et al. 2006). In Atlantic Canada, salt marsh restoration is in its infancy and although decommissioning projects have been conducted by non-government organizations such as Ducks Unlimited Canada (DUC) and government departments such as Nova Scotia Department of Agriculture over the last decade, no formal salt marsh restoration projects with associated monitoring programs were undertaken until this study. The Cheverie Creek (CHV) restoration project, undertaken by the Nova Scotia Department of Transportation and Infrastructure Renewal (NSTIR) as a Habitat Compensation Project (”Harmful Alteration Disruption or Destruction of Fish Habitat” [HADD]), represents the first fully monitored, active salt marsh restoration project undertaken in Atlantic Canada. As a HADD it must include a long-term (6-year) monitoring program to ensure project success. We examined the hydrologic, sedimentary, vegetative, and nekton conditions pre-restoration and the initial 2 years postrestoration at a macro-tidal salt marsh site. A modified form of the Global Program of Action Coalition for the Gulf of Maine Regional Monitoring Protocol (GPAC protocol) was employed (Neckles et al. 2002). Although culvert replacement projects are a popular restoration practice in the Gulf of Maine Region (Konisky et al. 2006), this study was the first comprehensive, quantitative data set of ecological response to culvert replacement in a high macro-tidal ecosystem. The goals of the study were to evaluate the rate of vegetation recovery after culvert replacement and to assess the influence of marsh morphology, sediment dynamics, and hydrology in re-establishing conditions and processes critical for selfmaintenance of tidal marshes and fish habitat. We hypothesized that increasing the size of the culvert would permit natural marsh functions to be enhanced. In addition, we determined if and how monitoring protocols need to be adapted in environments where abiotic factors are at the extreme range (e.g., tides and ice). Given the paucity of reference salt marshes in the region, we also evaluated the adequacy of a single reference site as a control.

Methods

and their minerogenic nature. The CHV (45◦ 09 31.44 N, 64◦ 10 09.84 W) marsh is located on the southern shore of the Minas Basin in the upper reaches of the Bay of Fundy. The water depth over the marsh during high tides can be upwards of 1–1.5 m. In the winter, rafted ice floes can become trapped on the marsh surface, depositing sediment and root rhizome material in the spring (van Proosdij et al. 2006b). The monitoring program was conducted on the front 29.5 ha of CHV, with the majority of monitoring activities focused on the north side of the creek. Highway 215 crosses the mouth of the CHV by way of a two-lane causeway-culvert structure. Historically, this system was diked (approximately 200 years ago) and used for agricultural purposes. There are remnants of the abandoned dike on both sides of the main creek, running perpendicular to the creek. The causeway was originally constructed approximately 50 m seaward of the dike with a small bridge section spanning the main creek channel. The bridge was replaced by a wooden box culvert with flap gate in 1960, completely restricting tidal flow. Encroachment of tree, shrub, and freshwater species occurred over approximately 25 years until tidal flow was partially restored when the flap gate was removed unintentionally in the early 1980s. Tidal flow was partially restricted at this time, causing a significant (1 m) tidal restriction for spring tides, allowing tidal flow to 4–5 ha of the marsh surface. Restoration activities consisted of the replacement of the remnants of the wooden box culvert (Fig. 1a) with a 9.2 × 5.5–m elliptical aluminum culvert (Fig. 1b), increasing the culvert opening approximately 7-fold from 4.7 to 32.6 m2 . This restored tidal flow to 43 ha (430,000 m2 ) of former marsh surface on spring tide events. This site is directly behind the causeway, which protects the marsh edges from wave action (Fig. 2). The zonation of the CHV vegetation community resembles that of other Fundy salt marshes such as Allen Creek (45◦ 52 N, 64◦ 21 W; van Proosdij et al. 1999) and Dipper Harbour (45◦ 06 N, 66◦ 26 W; Byers & Chmura 2007). The majority of the marsh surface is of the high marsh variety, with low marsh fringing the main creek and some marsh surface pannes and drainage channels (Fig. 2). The reference site for this compensation site is Bass Creek (CHV-R), located in Bramber, 5 km North of CHV (45◦ 11 55.24 N, 64◦ 07 58.11 W). Highway 215 also crosses the mouth of this creek with a causeway-bridge combination. The bridge spans the width of the creek and does not restrict tidal flow. The study area at CHV-R is 5.9 ha.

Study Site

The Bay of Fundy is an extension of the Gulf of Maine, on the east coast of Canada. It is a high macro-tidal estuary with semidiurnal tides that may exceed 15 m, exposing extensive intertidal mud flats at low tide. Suspended sediment concentrations within the Minas Basin in the upper Bay are high, ranging from 150 mg/L over the marsh surface (van Proosdij et al. 2006a) to 4,000 mg/L (Amos & Tee 1989) in the Basin. This inorganic material contributes to the high sedimentation rates recorded on marshes within the region (e.g., Chmura et al. 2001; van Proosdij et al. 2006a, 2006b)

2

Field Methods

A pre- and post-restoration monitoring program was developed for the CHV Salt Marsh Restoration Project in 2005 (Bowron & Chiasson 2006), which was based on the GPAC protocol with elements drawn from other programs/studies. The program involved a minimum of 1 year pre- and 5 years post-restoration data collection for a series of physical and biological parameters within the ecological indicator categories of hydrology, soils and sediments, vegetation, fish, and invertebrates. The indicator categories, physical and biological

Restoration Ecology

Marsh Response to Culvert Expansion

(a)

(b)

Figure 1. (a) Old box culvert at CHV in 2002 and (b) new culvert installed in December 2005.

Figure 2. Sampling design for first 8 of 26 lines at CHV superimposed over LiDAR elevations. LiDAR provided by T. Webster of the Advanced Geomatics Research Group at COGS, 2007. Elevations expressed relative to CGVD28 vertical datum.

Restoration Ecology

3

4

Soils and sediments

Hydrology

Category

Sediment plates*; marker horizons (three per RSET) sampled using a cryogenic corer (Cahoon et al. 1996)

Sediment cores (soil samples) paired samples: (30 mL syringe with base cut and 5 × 15–cm core)

Sediment accretion

Sediment characteristics (bulk density, organic matter content, sediment type)

Groundwater wells; sipper

Pore water salinity

Rod Sediment Elevation Tables (RSETs)

Groundwater wells (0.02 m × 1 m sampled at depth 0.9 m)

Depth to groundwater

Sediment elevation

Daily maximum water level (manual water level recorder), continuous (1- or 5-minute intervals) water level recorders (Solinst Levelogger Model 3001)

Sampling Method

Hydrology signal

Parameters

Daily maximums: CHV: 18 September–2 October 2003 Continuous water level: CHV (5 minutes): 3 November–24 November 2006; CHV (1 minute): 19 November–19 December 2007; and CHV-R (5 minutes): 19 December 2006–9 January 2007 CHV and CHV-R: biweekly July–November 2003; July–November 2004; monthly from July to September 2006 and July to September 2007 CHV and CHV-R: biweekly July–November 2003; July–November 2004; monthly from July to September 2006 and July to September 2007 CHV: four stations, installed June 2005, measured September 2005, 2006, and 2007 CHV-R: two stations installed September 2006, one station installed October 2006, measured October 2006 and 2007 Sediment plates— CHV: 25 locations; monthly from July to December 2002; July to October 2003; January to November 2004; annually July 2005, 2006 and 2007 Marker horizons— CHV: installed June 2005, measured September 2006, 2007 CHV-R: installed October 2006; measured October 2007 CHV: 25 paired samples, July 2002; 23 paired samples, October 2006 CHV-R: 19 paired samples, October 2006

Annual Sampling Frequency

X

X

X

X

X

X

Pre

Post

X

X

X

X

X

—

Yr 1

—

X

X

X

X

X

Yr 2

Table 1. The Cheverie Creek Salt Marsh Restoration monitoring program, including core and additional ecological indicators, methodologies, and the site-specific application (X: all sites; C–CHV-R).

Marsh Response to Culvert Expansion

Restoration Ecology

Restoration Ecology

Composition Abundance Height

Composition Species richness Density Length

Larval mosquito (abundance)

Nekton

Invertebrates

Parameters

Vegetation

Category

Table 1. Continued.

Six dip samples per panne (dipper)

Minnow traps in pannes, tidal creeks, and main channel (small fish); beach seine (30 m × 1 m; 6 mm mesh size) on marsh surface (all sizes)

Point intercept method (1-m2 plots)

Sampling Method

CHV: 129 plots, annually August 2002, August 2004, August 2006, August 2007 CHV-R: 27 plots, annually August 2003, August 2006, August 2007 All sites: spring tide. Four minnow traps and three pulls with beach seine per sampling date CHV: 18 October 2005; 17 August 2006, 8 September 2006, 6 October 2006; 2007 and CHVR: 19 October 2005; 11 September 2006, 7 October 2006; 2007 CHV: 10 pannes; monthly from June to September 2004; monthly from July to September 2005; 6 July 2006, 20 July 2006, 14 August 2006; monthly from June to September 2007 CHV-R: monthly from June to September 2004; monthly from July to September 2005; 6 July 2006, 20 July 2006, 14 August 2006; monthly from June to September 2007

Annual Sampling Frequency

X

X

X

Pre

Post

X

X

X

Yr 1

X

X

X

Yr 2

Marsh Response to Culvert Expansion

5

Marsh Response to Culvert Expansion

parameters, data collection methods, and sampling frequency are described in Table 1. Sampling was conducted at all sites using transects established in a non-biased, systematic sampling design tied into surveyed benchmarks. Twenty-six were established at CHV (Fig. 2) and eight at CHV-R. Data collection was conducted at sampling stations established at equal intervals along each transect, and surveyed with the Total Station.

Hydrology

A digital elevation model (DEM) was produced for each site using a 2004 1:10,000 aerial photograph, and elevation survey data collected using a combination of differential global position system (DGPS), Leica TCR-705 Total Station, and 2007 LiDAR data (Fig. 2). All topographic data were referenced to the Canadian Geodetic Vertical Datum 1928 (CGVD28). The hydroperiod and summary hypsometric curve for each site were modeled using the tidal signal and DEM for the marsh surface (Table 1). Depth to groundwater was sampled using seven groundwater wells installed adjacent to vegetation stations at CHV (Fig. 2), and six at CHV-R. A single measurement (centimeters below marsh surface) was taken at each station during low tide (Table 1).

Soils and Sediments

The location and frequency of pore water salinity sampling at all sites were matched with depth to groundwater sampling (Fig. 2). Salinity was measured at shallow (0.15 m) and deep (0.45 m) wells using a handheld refractometer (2003–2004). In 2005, the salinity well method was replaced by the use of a soil probe (sipper) (Roman et al. 2001). Due to the high sediment concentration of water samples collected by the sipper, samples were bottled and returned to the lab and allowed time for the sediment to settle out before a refractometer reading was taken. Marsh surface elevation change (+/ − 0.001 m) was measured once per year using four rod sediment elevation tables (RSETs; Cahoon et al. 2002) at CHV and three at CHV-R. All measurements were taken at low tide and at the same time in the growing season (September) to minimize the influence of evapotranspiration (Paquette et al. 2004). Vertical accretion at all sites was measured using feldspar marker horizons and a cryogenic corer as described by Cahoon et al. (1996). Three 0.5-m2 marker horizons per RSET station were established at each site (Table 1). In 2002, twenty-five 10 × 20–cm aluminum plates were inserted 10 cm below the marsh at CHV (Fig. 2) according to the design, installation, and sampling method described in van Proosdij et al. (2006a). Sediment cores (bulk density, organic matter [OM], and grain size) were collected using a stratified random sampling procedure paired with vegetation sampling plots (Fig. 2; Table 1). OM content was determined by loss on ignition (LOI), while grain size was analyzed using a Coulter LS200.

6

Vegetation

The marsh vegetation community was surveyed using permanent 1-m2 plots. A detailed vegetation survey (species composition, percent cover, and height) was conducted at CHV in 2002 on the 92 plots on the first eight transects (Fig. 2). This method was replicated in 2004 on the remaining 27 plots at CHV as well as on the 27 at CHV-R. Plots sampled in 2002 were combined with plots sampled in 2004 to assess the pre-restoration condition. Post-restoration species composition and abundance were determined within each plot using a point intercept method (abundance = frequency of contact of each species with 25 points within plot) (Roman et al. 2001). Mean plot species richness and cover of bare ground were compared between sites and years using repeated measures analysis of variance (ANOVA). Differences in vegetation composition were assessed using nonparametric multivariate ANOVA (NPMANOVA), with species abundances as dependent variables (including only species occurring in five or more plots; 100 permutations) (Anderson 2001).

Nekton

A beach seine and a set of four minnow traps were used to sample nekton on the marsh surface, pannes, and tidal creeks during spring tide events. Sampling with the beach seine was conducted according to the methodology developed and used by the Community Aquatic Monitoring Project (CAMP) (Weldon et al. 2005). Smaller species in the salt pannes and tidal creeks were sampled using the minnow traps. The traps were deployed before high tide and retrieved once the tide level had dropped (approximately 3.5 hours). Nekton density measures were limited to samples of known area (i.e., minnow trap excluded) (Konisky et al. 2006). All captured specimens were held in buckets, identified to species level, counted (to a maximum of 300 per species), and measured for length (15 individuals per species).

Results Hydrology

Based on the 2007 tidal signal data, LiDAR, and on-site measurements including observed tree die-off, the total area of recovering marsh at CHV has increased approximately 8fold from 5 to 43 ha (Fig. 3). A comparison of the upstream tidal signal to the downstream tidal signal (Table 1) indicates that the new culvert has eliminated the tidal restriction on all tides under 7.1 m (CGVD28) (85% of tides). There was no significant difference in depth to water table between pre- and post-measures at CHV (t test, t = 1.20, df = 40, p = 0.235); however, depth to water table at CHV-R was significantly deeper than observed at CHV for all years (t test, t = −3.16, df = 52, p = 0.003). The detectable difference was therefore spatial in nature and there was no temporal change. Both sites showed strong seasonal trends, with the greatest depths to water table recorded in July.

Restoration Ecology

Marsh Response to Culvert Expansion

the first 0.5 m of the tidal creek were lost due to erosion, likely due to ice scour. There was no significant difference between pre- and post-sediment accretion rates using plates (paired t test: t = 1.2108, df = 12, p = 0.249). The mean prerestoration accretion rate was 0.7 ± 0.5 cm/year compared with a 0.4 ± 0.5 cm/year rate from 2005 to 2007 (Table 2). Table 2. Mean rate of change in marsh surface elevation measured at sediment plates at CHV pre- (2002–2005) and post-restoration (2005–2007).

Mean Rate of Change (cm/yr± SD) Station

Figure 3. Hypsometric curves pre- and post-restoration at CHV and base conditions at CHV-R. Original culvert restricted tides to less than 6 m above geodetic datum, new culvert permits tides up to 7.1 m. Soils and Sediments

Average soil salinity ranged between 1 and 31 ppt at CHV-R and 0–36 ppt at CHV and were generally lower in the summer and increasing in the fall (Fig. 4). Mean pore water salinity levels at CHV post-restoration were significantly higher (t test, t = −6.78, df = 72, p = 2.75E−09) than pre-restoration levels. The largest changes in salinity were observed at the high marsh stations and those further upstream. In addition, mean pore water salinities at CHV were found to be significantly higher than those of CHV-R (t test, t = 14.15, df = 77, p = 4.00E−23) in the same year. Pre-restoration, at CHV, no significant differences were detected between shallow and deep wells (t test, t = −0.388, df = 134, p = 0.698). When the shallow wells at CHV and CHV-R were compared, no significant difference was observed (t test, p < 0.174); however, there was a significant difference (t test, t = 11.02, df = 45, p = 2.23E−14) measured between the deep wells at both sites (CHV > CHV-R). A significant difference was detected post-restoration between the shallow and deep samples at CHV (t test, t = 2.11, df = 26, p = 0.041). A trend of decreasing bulk density and increasing organic matter content with increasing geodetic elevation and distance from the main creek was observed both pre- and postrestoration at CHV (Fig. 5a & 5b). One year post-restoration there was a minor increase in bulk density across the marsh (up to CHV-R levels) and a decrease in organic matter when compared to pre-restoration conditions. Pre-restoration, CHV recorded an increase in mean grain size with increasing elevation and distance from the main tidal creek (Fig. 5c). Post-restoration, mean grain size decreased with increasing elevation, but exhibited no discernable pattern at CHV-R. The mean net change in surface elevation at CHV was highly variable both at a station between years and between different sites across the marsh within the same year (Table 2). After the first year, all of the sediment plates installed within

Restoration Ecology

Elevation (m)

L1S2 L1S6 L1S8 L2S6 L3S2 L3S4 L5S2 L5S4 L5S6 L5S2(S) L7S2 L7S5 L7S8 L7S14 L8S3 L8S5 Overall mean

4.4 4.8 5.1 6.2 4.2 5.2 4.6 5.6 6.0 6.1 5.6 6.4 6.2 7.1 4.8 5.1

Pre

0.7 0.3 0.4 1.1 0.7 1.3 0.7 2.0 1.2 0.5 0.4 0.2 0.1 0.6 0.7 0.7 0.7

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

Post

0.1 0.3 0.2 0.6 0.2 0.3 0.6 0.4 0.2 0.2 0.2 0.3 0.2 0.2 0.0 0.2 0.5

NA 10.9 ± 4.9 0.5 ± 0.3 NA 1.2 ± 0.5 1.0 ± 0.9 0.4 ± 0.4 –1.3 ± 1.5 0.3 ± 1.1 0.2 ± 0.1 –0.4 ± 0.1 0.2 ± 0.2 NA 1.1 ± 0.5 0.4 ± 0.3 0.0 ± 0.5 0.4 ± 0.7

All elevations relative to CGVD28 datum. NA = plate was not relocated postrestoration. Only stations that survived at least 1 year were included in the analysis. See Figure 2 for location of stations.

Since the RSET stations were not installed until 2005, no RSET data are available pre-restoration. One year postrestoration, the change in surface elevation was highest at RSET-02 and RSET-03, both with a net rate of change of 2.2 cm/year (Fig. 6). The lowest recorded change in surface elevation was in the sedge community at RSET-04 with 0.6 ± 0.1 cm/year (± SE). Both marker horizons and buried plates (within a 30 m radius of RSETs) were within the same range of accretion values (Fig. 6). Subtracting the marker horizon data from the RSET measurement leaves 0.35 cm to subsurface processes. The lowest value (0.63 cm/year) was at RSET-04 indicating minimal subsurface processes at work. RSET-02 had the highest change in surface elevation attributable to subsurface processes (1.23 cm/year), which could be due to pore water flux given its proximity to the tidal creek (Fig. 6). Two years post-restoration, the 2005–2006 elevation trends were repeated with slightly lower values (Fig. 5). There was no significant difference between years (2006 vs. 2007) at CHV (t test: t = 1.6646, df = 3, p = 0.1946). However, statistically significant differences in changes in surface elevation were recorded between CHV (mean ± SE = 1.22 ± 0.08) and CHV-R (mean ± SE = 0.26 ± 0.03) (t test: t = −3.62, df = 3.7, p = 0.025) in 2007.

7

Marsh Response to Culvert Expansion

Frequency (%)

(a)

45

CHV-pre

DEEP

40

CHV-post

35

CHV-R

30 25 20 15 10 5 0 81

01

21 41

61

82 02

22 42

62 83

03

23 43

6

Salinity class (ppt) Mean (ppt)

Max (ppt)

Min (ppt)

CHV-pre

24

30

15

CHV-post

24

35

8

CHV-R

19

31

2

(b) 45

SHALLOW

40

Frequency (%)

35 30 25 20 15 10 5 0 81

01

21

41

61

82

02

22

42

62

83

03

23

43

6

Salinity class (ppt) Mean (ppt)

Max (ppt)

Min (ppt)

CHV-pre

24

31

15

CHV-post

25

38

4

CHV-R

20

32

1

Figure 4. Change in mean pore water salinity at (a) deep and (b) shallow wells for CHV and CHV-R from 2002 to 2007.

Vegetation

Increased tidal inundation had the almost immediate effect of causing visible die-off of non-salt tolerant vegetation at the upper edge of the marsh (Fig. 7). Since the area inundated exceeded initial estimates, no sampling stations were located within the upper limits of inundation at high spring tides. However, 2 years post-restoration, two primary salt marsh colonizers, Salicornia europea (glasswort) and Atriplex glabriuscula (marsh orache), were observed growing in these die-back areas.

8

Species density (SD) has been consistently greater at CHV-R (Fig. 8a) (t tests: pre: t = 2.75, p = 0.01; 2006 : t = 5.02, p = 0.0001; 2007 : t = 2.38, p = 0.02). Repeated measures analysis shows a significant site by year interaction (F[2] = 3.9473, p = 0.008) (Table 3), which is caused by the decline in SD at CHV immediately following reintroduction of tidal flow that recovered to pre-restoration levels by the second year post-restoration. The increase between year one and year two post-restoration is due to the colonization of CHV by more native halophytic species. The greater overall

Restoration Ecology

Marsh Response to Culvert Expansion

(b)

1.6

Bulk density (g cm–3)

1.4

CHV-pre

80

CHV-post

70 organic matter content (%)

(a)

CHV-R

1.2 1.0 0.8 0.6 0.4 0.2

60 50 40 30 20 10

0.0

0 05

0

100

150

200

05

Distance from main creek (m)

0

100

150

200

Distance from main creek (m)

(c) 180

mean grain size (microns)

160 140 120 100 80 60 40 20 0 05

01

00

150

200

Distance from main creek (m)

Figure 5. (a) Changes in bulk density, (b) organic matter content, and (b) mean grain size at CHV and CHV-R pre- and post-restoration relative to the main tidal creek. RSET-Yr1-post

Mean change in surface elevation (cm yr–1)

3.0

RSET-Yr2-post marker horizon-mean post

2.5

plate-pre plate-mean post

2.0

1.5

1.0

0.5

0.0 RSET-01 (6.5 m)

RSET-02 (6.35 m)

RSET-03 (6.45 m)

RSET-04 (6.66 m)

Figure 6. Mean annual change in surface elevation recorded at RSETs (± SE) and mean sediment accretion based on the marker horizons and sediment plates (within area of RSETs) for the first 2 years post-restoration at CHV. Elevations expressed relative to CGVD28 datum.

Restoration Ecology

9

Marsh Response to Culvert Expansion

Figure 7. Areas of change in vegetated community, panne expansion, and extent of die-off 2 years post-culvert replacement calculated from 2007 LiDAR survey and on-site tide record. LiDAR provided by T. Webster of the Advanced Geomatics Research Group at COGS, 2007. Elevations expressed relative to CGVD28 vertical datum.

SD at CHV-R can be attributed to a number of uncommon species such as Solidago sempervirens (seaside goldenrod) and Plantago maritima (seaside plantain) that are more frequent at CHV-R. These are typically found in the upper marsh/salt meadow areas where the dominants are Carex palacea (salt marsh sedge) and Spartina pectinata (prairie cordgrass). These species do occur at CHV but less frequently; thus, they are less likely to contribute to greater average SD. Overall, species richness (SR) (total number of species encountered in plots) was usually greater at CHV than CHV-R, but not consistent among years (pre: 24 vs. 18, year 1 post: 20 vs. 22, and year 2 post: 24 vs. 19). The amount of unvegetated area (bare ground) has been dynamic at both sites, with the formation of pannes being one of the main processes responsible for this. In both years post-restoration, there was a greater area of bare ground at CHV (site by year interaction: F[2] = 3.6579, p = 0.013) (Table 3; Fig. 8b). The two sites had similar plant communities pre-restoration and year one post (NPMANOVA for site differences: prerestoration F[1] = 1.592, p = 0.19; year 1 post F1 = 0.939,

10

Table 3. Repeated measures ANOVA for (a) species density and (b) unvegetated area. (a) Factor

df

Site 1 Residuals 128 Year 3 Site × year 3 Residuals 384

SS

148.97 767.46 15.12 13.83 448.55

MS

F

P

148.97 24.846