Designing topographic heterogeneity for tidal wetland restoration

Ecological Engineering 123 (2018) 212–225

Contents lists available at ScienceDirect

Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng

Designing topographic heterogeneity for tidal wetland restoration a,⁎

b

a

a

Heida L. Diefenderfer , Ian A. Sinks , Shon A. Zimmerman , Valerie I. Cullinan , Amy B. Borde a b

T a

Pacific Northwest National Laboratory, Coastal Sciences Division, Marine Sciences Laboratory, 1529 West Sequim Bay Road, Sequim, WA 98382, United States Columbia Land Trust, 850 Officers’ Row, Vancouver, WA 98661, United States

A R T I C LE I N FO

A B S T R A C T

Keywords: Constructed mound Engineering design Hummock Levee Marsh Microtopography Mound Restoration Ridge Soil moisture Tidal freshwater Tidal marsh Tidal wetland Topographic heterogeneity Terrace Wetland

Topographic heterogeneity affects abiotic and biotic components of vegetated ecosystems and has the potential to provide functions and services such as promoting floral and faunal biodiversity. Yet, few studies have documented the design, implementation, and outcomes of mound-type features in estuarine and tidal freshwater wetland restoration projects. Here we report data from a synoptic survey of soil temperature and moisture on mounds at tidal wetland restoration sites in the Pacific Northwest, together with the results of a literature review, and the insights of regional restoration practitioners regarding ecological and practical considerations for mound construction. Few papers reviewed addressed conditions for plant establishment on mounds, such as soil moisture, which is important in climates with annual dry seasons, like the Pacific Northwest. We report 2015 data on soil moisture and temperature for the tops, sides, and toes of 15 mounds constructed between 2001 and 2013 of fill dirt, or dirt and logs, at 5 tidal wetland restoration sites. Two of the restoration projects studied are located in the lower Columbia River and estuary, one is in the Puget Sound, and two are in outer Pacific coastal estuaries in Oregon. The heights of mounds ranged from 0.5 m to 1.5 m above the marsh plain, and the area of individual mounds ranged from 21 m2 to 5185 m2, while most published information treated somewhat smaller topographic features (termed mounds, hummocks, tussocks, ridges, levees, fans, or terraces). Differences in the minimum and maximum values of moisture between the toe and the top of mound in summer ranged from 2.9% to 40%. Statistical analysis strongly suggested stratification of soil moisture by elevation. Analysis of soil temperature was less conclusive, but temperature appeared to be positively correlated with elevation. Soil moisture relative to mound aspect (cardinal direction) was significant in some cases, but inconsistent between mounds. We also report our observations from a third recent restoration project on the tidal Columbia River, and observations made by other practitioners who have designed and installed mounds from San Francisco Bay to northern Puget Sound. Our findings will support design of wetland restoration sites where topographic heterogeneity is an objective.

1. Introduction 1.1. Background Different groups of plant species have long been observed to occur along gradients in land elevation. Ecologists and agricultural scientists have elucidated how elevation gradients are related to the tolerance of plants for variation in fundamental environmental conditions such as temperature, moisture, light, soil nutrients, and the disturbance regime. Even topographic differences on the order of a few centimeters have been linked to micro-environments that affect seed germination and establishment (Harper et al., 1965). The term microtopography has been defined as variability on the scale of individual plants, or ∼1 cm to 1 m, and has components of both relief (vertical extent) and

⁎

Corresponding author. E-mail address: [email protected] (H.L. Diefenderfer).

https://doi.org/10.1016/j.ecoleng.2018.07.027 Received 25 May 2018; Accepted 24 July 2018 0925-8574/ © 2018 Published by Elsevier B.V.

roughness (topographic variability) (Moser et al., 2007). Microtopography is particularly important to plant distribution in floodplain wetlands because of its direct association with surface and groundwater dynamics and edaphic gradients (Huenneke and Sharitz, 1986; Titus, 1990). Topographic heterogeneity has been shown to affect numerous aspects of ecosystems in addition to the distribution of organisms and patterns of abiotic factors. It affects ecosystem processes, animal behavior and habitat use, herbivory and other trophic interactions, competitive exclusion, development, and genetic and reproductive attributes (Larkin et al., 2006). In intertidal areas, it is clear that elevation strongly controls wetland plant distribution, but the proximity to tidal creeks is also a controlling factor (Zedler et al., 1999), as are soil permeability and surface and groundwater dynamics. Microtopography and flood disturbance

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

1.2. Study purpose

interact on marsh plant species survival and distribution (Pollock et al., 1998; Tessier et al., 2002). In swale and dune habitats, plants with better drainage respired aerobically in contrast to those in lower, poorly drained marsh areas (Burdick and Mendelssohn, 1987). Bubier et al. (1993) showed that seasonal mean water table position can explain the variability in methane emissions at the hummock and hollow scale. Plant species richness on hillocks in salt marshes of southeastern Denmark is higher (Vestergaard, 1998) and the surface area of tussocks is positively correlated with species richness (Werner and Zedler, 2002). Seed banks in sheltered micro-habitats in salt marshes are larger than areas that have greater sediment mobility (Inglis, 2000). Through these types of mechanisms, topographic heterogeneity can increase plant biodiversity in marshes. Microtopography (e.g., mounds, hummocks) has been used as a wetland restoration and management technique for this purpose, for example, increasing roughness with tilling at non-tidal wetland creation sites (Moser et al., 2007), or retaining and protecting mounds with shrubby vegetation at a coastal wetland restoration site (Gallego Fernández and García Novo, 2007). For coastal restoration, the germination and establishment of plants is one of the most important criteria for any technique under consideration, and it is generally understood that the design specifications of mounds, hummocks, ridges, peninsulas, and berms have the potential to affect environmental conditions and therefore the success or failure of plantings and the degree of natural colonization. For instance, smoothly mounded higher-elevation islands made from dredged material at a San Diego Bay salt-marsh restoration site caused salts to be wicked to the surface preventing plant establishment because of soil salinity (Larkin et al., 2006). In addition to plant community effects, other potential benefits of using mounds in tidal wetland restoration include diverse edaphic conditions (Cantelmo and Ehrenfeld, 1999; Bruland and Richardson, 2005; Courtwright and Findlay, 2011), and edge effects on aquatic species productivity (Rozas et al., 2005). They may also provide ecosystem services, such as resilience to variable environmental conditions (Doherty and Zedler, 2015), and slowing the rate of wetland loss (Zedler and Kercher, 2005). Elevation is a central element of coastal wetland restoration designs because it affects the inundation regime, which together with sediment supply and wetland plant productivity control the rate at which coastal marshes sequester sediments, a fundamental process in these depositional ecosystems (Allen, 2000). The rate of sediment accretion is often very important in regions with subsidence behind dikes, which may take decades for elevation to equilibrate with surrounding areas once they are reconnected with adjacent waterways, and thus retard the ability of desired plant species to establish (Diefenderfer et al., 2008). The timing of excavation and grading that expose soil, and of plantings to enhance retention, thus are important in coastal wetland restoration planning because of the positive feedback on increased elevation. With inundation, the soil environment can rapidly change from oxidized to reduced, and previously thriving plants such as pasture grasses are likely to die back and be replaced by more hydrophytic species (Frenkel and Morlan, 1991; Blackwell et al., 2004). However, oftentimes there is excess mineral sediment available in coastal wetland projects in comparison with more nitrogen-rich organic soils, and depositing it on sites can adversely affect ecosystem development, retarding root development and reducing above ground biomass (French, 2006). The construction of mounds has been proposed to coastal restoration program reviewers on the basis of providing topographic diversity with the potential to reduce the impacts of subsidence behind dikes, accelerate the development of woody plant communities, control an invasive non-native plant (reed canary grass (Phalaris arundinacea)), produce a plant-community mosaic, and generally increase habitat complexity at the restoration site (Krueger et al., 2017).

Both restoration practitioners and reviewers of the Columbia Estuary Ecosystem Restoration Program (CEERP) on the West Coast of the continental United States (Diefenderfer et al., 2012, 2016), have requested guidance regarding the right balance between practical concerns and ecological function. Specifically, science-based construction specifications for mounds (e.g., height, width, aspect [cardinal direction], and slope) used to increase topographic heterogeneity had not been established for the program. On some projects, grading is required to produce a land-surface elevation suitable for wetland restoration given the prevailing hydrologic regime, and the formation of mounds may help defray the costs of moving excavated material offsite. However, data on the effectiveness of mounds for tidal wetland restoration in the Pacific Northwest were not available in the peer-reviewed literature. In fact, information about the ecological and practical considerations for designing mounds for tidal wetland restoration was needed to reduce uncertainties in program planning. The purpose of this study was to provide the missing science-based information to practitioners and managers of restoration projects in the CEERP. It was specifically intended to improve the effectiveness of restoration actions, through the CEERP’s adaptive management program (Ebberts et al., 2017). The objectives of this research were to compile published literature that is highly relevant to mound design for tidal wetland restoration, obtain insights from regional restoration practitioners regarding ecological and practical considerations for mound construction, and generate new data on environmental conditions and plant establishment from mounds previously constructed at tidal wetland restoration projects in the Pacific Northwest. 2. Methodology 2.1. Study area The climate of the coastal Pacific Northwest (PNW) is characterized by high annual rainfall but summer drought. Due to the proximity of coastal mountain ranges to the Pacific coast and ubiquitous rocky headlands, tidal wetlands are generally smaller and more isolated than those on the Gulf and Atlantic coasts. Two of the largest complexes of wetlands on the Pacific coast are the lower Columbia River and estuary (LCRE) and Puget Sound (Callaway et al., 2012). Field data were collected from six tidal wetland restoration sites in these two areas and the outer Pacific coast in the U.S. states of Washington and Oregon (Fig. 1). The wetlands are subject to a mixed, semi-diurnal mesotidal regime. In the 234 km tidal portion of the Columbia River, floodplain wetlands are also strongly affected by river flows from the 668,000 km2 basin, and during low to moderate flows salinity extends only 20–40 km inland from the river mouth, so there is a large expanse of tidal freshwater area (Jay et al., 2016). The six tidal wetland restoration sites were identified through outreach to practitioners involved with current and past restoration projects in the PNW to learn where mounds had been incorporated in project designs (Appendix A, Table A.1). The criteria for selecting study sites were that the object of ecological restoration was a wetland with a tidally influenced hydrologic regime, that one or more mounds had been constructed for the purpose of restoration, and that access to the site was feasible and permissions to conduct research could be obtained through land owners. On this basis, three sites were selected on the LCRE (Colewort Creek, Ruby Lake, and Seal Slough), one on the Puget Sound (Marietta Slough), and two on the outer PNW coast (Anderson Creek and Drift Creek) (Fig. 2). The six sites are all located on the tidally influenced sections of river floodplains. All sites, with the exception of Marietta Slough and

213

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

underwent restoration construction since 2012. Colewort Creek, located ∼3 km up the Lewis and Clark River from Youngs Bay at Columbia River km 16 on the Oregon side, is a 18-ha tidal marsh wetland that experiences salinity effects during low Columbia River flow (National Park Service (NPS) 2012). Restoration construction on this site on the South Clatsop Slough within Lewis and Clark National Historical Park was completed in 2012 by the NPS and the Columbia River Estuary Study Taskforce (CREST), and distinct mounds were constructed and planted with woody vegetation (C. Cole, NPS, and M. Van Ess, CREST, pers. comm, 4 August 2015). Ruby Lake is a 50 ha freshwater tidal floodplain wetland located along an interior slough of Sauvie Island, Oregon (river km 143), hydrologically driven by seasonal river stage and diurnal tides along the mainstem LCRE (CREST, 2013). Restoration at Ruby Lake was completed in 2013 by CREST with the goal of restoring emergent wetlands and four peninsula type mounds were constructed along the forested slope of the wetlands (Tom Josephson, CREST, pers. comm., 1 July 2015). Seal Slough is a freshwater tidal floodplain located ∼5 km up the Grays River from the CRE at river km 32, site of the former Kandoll Farm. Final phase of restoration construction at Seal Slough was completed by Columbia Land Trust in 2013 with the goal of restoring the historical Sitka spruce swamp community, and a variety of mounds and simulated natural river levees were constructed to restore forested and scrub-shrub vegetation communities (Sinks, 2014; I. Sinks, Columbia Land Trust, pers. comm., 5 August 2015).

Fig. 1. Study sites on the Pacific Coast (triangles). The City of Seattle is on the Puget Sound and the City of Portland is on the Columbia River.

Colewort Creek, are freshwater. At four out of six of the sites, the intent of restoration activities was to restore tidal marshes, which are emergent wetlands mostly composed of erect, rooted, herbaceous hydrophytes and may have small scattered areas of woody plants (Cowardin et al., 1979). At Marietta Slough riparian woodlands were another restoration objective (Washington Department of Fish and Wildlife (WDFW), 2008), and at Seal Slough the long-term project goal was Sitka spruce forested wetland (Cardno ENTRIX, 2013). All of the mounds at all of the sites had been planted with woody vegetation during restoration at various years in the past, with the exception of Drift Creek where a native grass seed mix was applied (W. Hoffman, MidCoast Watersheds Council, per. comm.12 August 2015). The three non-Columbia River sites were restored between 2001 and 2007. Upstream from Coos Bay, Oregon (Fig. 1), Anderson Creek drains a 100-ha watershed, where restoration construction of a 1160-m meandering channel was completed in 2001 by the South Slough National Estuarine Research Reserve (NERR) (Cornu, 2005). Construction of this section of Anderson Creek, now a freshwater marsh wetland located at the head of tide, included the installation of nurse log features composed of wood and soil, to mimic historical floodplain features (Cornu, C., South Slough NERR, pers. comm. 2 July 2015) (Fig. 2a). The Drift Creek site is 518 ha of tidal freshwater floodplain acquired by the U.S. Forest Service (USFS) located along Drift Creek ∼5 km upstream from Alsea Bay, Oregon (USFS and MidCoast Watersheds Council, 2004) (Fig. 1). Restoration at Drift Creek was completed in 2005 by the MidCoast Watersheds Council/USFS, and one objective was to create alluvial fans composed of scrub-shrub and forested vegetation communities within the restored emergent wetland (W. Hoffman, MidCoast Watersheds Council, per. comm.12 August 2015) (Fig. 2c). Marietta Slough is a tidal forested wetland ∼3 km up the Nooksack River from Bellingham Bay in the Puget Sound in Washington State. Restoration construction at Marietta Slough was completed in 2002, with planting completed by 2007 by WDFW with the goal of restoring a more complex forested swamp community (WDFW 2008; R. Kessler, WDFW, pers. comm, 25 June 2015). A large number of mounds in a variety of shapes and sizes were constructed at Marietta Slough to facilitate the restoration of forested wetland communities (Fig. 2d). The three sites that are part of the CEERP on the Columbia River

2.2. Outreach to restoration practitioners and literature review Developing a more productive and effective relationship between restoration science and practice has long been a goal. Authors of a recent summary of a related effort by the Society for Ecological Restoration International wrote: “Ideally, restoration ecologists provide ideas, guidance, and rigorous data that benefit restoration practitioners, whereas practitioners put the science into practice, exchange insights with the scientists, and make their project sites available for them to develop and test their theories” (Cabin et al., 2010). Accordingly, we reached out to the CEERP restoration sponsors and other restoration practitioners in Puget Sound and the outer PNW coast estuaries to explain the purpose of the study, discuss key environmental and design considerations, sharpen the focus of the research to directly support current needs, and identify potential restoration project sites for field examination of mounds. Our purpose was to seek insights and unpublished reports, and generally to gain information from other systems that could be relevant to LCRE sites, while recognizing caveats related to differing physical processes in the tidal-fluvial gradient. (See Appendix A). Based on our experience and the initial literature review, we developed a list of salient discussion points for outreach to restoration managers regarding the mound design challenge. In effect, this exercise generated elements of a conceptual model for mounds: 1) features, e.g., height, slope, and soils material; 2) environmental effects, e.g., soil temperature and time to plant establishment; 3) relevant site conditions, e.g., historical and existing topography, sediment regime, and plant community; and 4) practical considerations, e.g., regulatory constraints, cost, and constructability (Fig. 3). Systematic review of the literature on mounds was conducted using Web of Science and Endnote tools (Appendix A). First, we developed primary keyword search strings that produced the results most relevant to mounds, hummocks, or microtopography in tidal wetland restoration. We used a variety of secondary search terms, e.g., soil/sediment, height/elevation, function, nutrient, retention and root. We assessed the resulting abstracts to ensure their relevance and reviewed the full text of relevant papers for findings about mounds that are of interest to restoration practitioners. We archived information from the literature review in annotated bibliographies for key references.

214

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

Fig. 2. Six study sites where data and observations were collected.

moisture data collection locations and sufficient to calculate the height of the mound and the angle or slope of its sides; 2) a set of 10 Durac analog soil thermometers (H-B Instrument Company, Trappe, Pennsylvania) to simultaneously measure soil temperature at 5 cm and 15 cm depths along transects (Fig. 2), as well as air temperature; and 3) a HydroSense II soil moisture sensor (Campbell Scientific, Logan, Utah) to measure moisture at the 12 cm depth. These elevation, temperature, and moisture data were collected during one day at each site. To evaluate small-scale spatial variability, we also collected a total of 20 moisture measurements, and 19 5-cm and 19 15-cm temperature measurements, within the diameter of a 30 cm circle centered on the sampling location (termed “field replicates”). To evaluate precision, we also collected a total of 9 moisture measurements, and 10 5-cm and 10 15-cm temperature measurements in the same hole used for the original primary measurement (termed “field duplicates”). The height of the mounds was calculated by first determining the “toe” and “top” of each mound. Elevation data were collected at these points, which we defined by a pronounced change in relative slope. Specifically, the toe was identified by an increase in slope at the border of the marsh surface surrounding the mound, and the top was identified by a decrease in slope near the top of slope on each side of the center of the mound. Toe elevation was subtracted from top elevation to derive the height. In all, 143 locations were measured for all four data points: elevation (cm), soil temperature

2.3. Field research 2.3.1. Data collection In the PNW, summer drought tends to limit plant establishment, and restoration projects are infrequently able to provide irrigation, so we focused research on summer soil moisture and temperature and the plant community on constructed mounds. Data were collected at or near mid-day during the summer of 2015, and ambient air temperatures were high relative to historical averages (the range of the daily maximum air temperature measured during data collection was 19 °C at Anderson Creek to 42 °C at Ruby Lake). In total, soil moisture and temperature data were collected at eleven mounds—two at Colewort Creek (Mounds CC-1 and CC-2), two at Drift Creek (Mounds DC-1 and DC-2), five at Ruby Lake (Mounds RL-0, RL-1, RL-2, RL-3, and RL-4), and two at Anderson Creek (Mounds AC-1 and AC-2) (Fig. 2). Mound perimeters were delineated using a Trimble GeoXH handheld GPS and mound area was calculated from these data in GIS. Quantification of topographic heterogeneity generally must include both vertical and horizontal measurements (Larkin et al., 2006). We measured physical features using a synoptic survey approach. For each face of each mound, i.e., the north, south, east, and west faces, we used 1) a Real-Time Kinematic Global Positioning System (RTK-GPS) to measure elevations on the mound corresponding to temperature and

215

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

Fig. 3. Site conditions affect the structural elements of design and the consequent environmental effects.

2.3.2. Data analysis For each site with sufficient sample size (n > 8), Pearson pairwise correlation and scatter plots were used to assess the linear relationship between elevation, soil temperature at 5 and 15 cm depths, and soil moisture. For each mound and site, a Kruskal-Wallis multiple comparison test comparing the median moisture was conducted between the relative vertical location on the mound (toe, lower side, upper side, top of slope, and top of mound), both with and without the toe and top of mound, and between aspects (N, S, E, and W). If only two categories were being tested then the overall Kruskal-Wallis result was the only result reported; if more than two categories were tested then the Kruskal-Wallis pairwise comparison was also reported. A general linear model was fit for each class type using elevation as a covariate. In reporting statistical results, significance was detected at α = 0.05. To assess measurement precision and spatial variability, the percent difference (PD) between the original measurement and either duplicate measurements obtained from the same hole or replicate measurements obtained from within a 30 cm diameter circle around the transect was calculated and then characterized by site and type. The descriptive statistics including the sample size, mean, standard error of the mean (SE Mean), 1st and 3rd quartiles (Q1 and Q3), median, minimum, maximum, and a 95% confidence interval (lower and upper CL) were calculated.

(°F) at 5 cm and 15 cm depths, and soil moisture (%). We summarized the data for each variable by mound and based on the relative location (toe, lower side, upper side, top of slope, and top of mound; see Appendix B, Table B.1). Survey and temperature data collected and analyzed in feet and degrees Fahrenheit were converted to meters and degrees Celsius for reporting. At each site, we documented visual observations of vegetation establishment and herbivory, took photographs to document site conditions and findings, and compared our observations with descriptions of the original restoration plantings from reports and interviews with practitioners. Only observations, not quantitative measurements, were made at Seal Slough for logistical reasons. Mounds were not delineated by relative vertical location (i.e., top and toe, and cardinal direction) at Anderson Creek because of the low height and linear shape of the logs, or at Marietta Slough because logistical problems prevented data collection on all four sides of the mounds; therefore, for these two sites, the height above marsh plain was calculated as the difference between the maximum and minimum elevation on the mound. The small sample sizes at these two sites were insufficient for statistical analysis because 1) the Anderson Creek mounds were composed of downed logs, decayed wood, duff, and loose soil, which made measurements difficult, and 2) at Marietta Slough, faulty equipment prevented us from measuring soil moisture, though temperature data were collected at four mounds (Mounds MS-1, MS-2, MS-3 and MS-4).

216

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

3. Results

naturally formed from stumps, logs, and roots at the forested wetland, which had a mean height of 18.6 cm and length of 40 cm, the depth and duration of flooding were lower than in hollows, herbaceous plant species richness was higher, and soluble phosphate and ammonium concentrations in porewater were lower. Bruland and Richardson (2005) found that created hummocks ∼1.0 m high and 1.5 m in diameter at the restoration sites were always above the water table in contrast to flats and hollows, plant species richness and levels of aboveground biomass on hummocks were significantly lower, and soil nitrate and ammonium levels were significantly higher. Creating surface roughness of ± 15 cm in the form of mounds and hollows tilled into smoothed, crowned agricultural land at another North Carolinian wet hardwood forest restoration site in 2003 increased surface water storage and produced hydrologic conditions similar to the reference site when monitored 3–6 years later (Jarzemsky et al., 2013). At a third floodplain forested swamp restoration site in North Carolina, where fill was removed and the buried soils were contoured into low ridges and depressions differing by ∼50 cm in elevation, the ridges had a lower water table (20–40 cm lower on average), lower soil organic carbon level, lower cover of obligate wetland species, higher cover of facultative upland species and upland species, more forb species, more woody species, and higher overall plant species richness (Rossell et al., 2009). In contrast, Drouin et al. (2011) found no difference in floodplain soil organic carbon of riparian forests including natural levee, depression, and terrace features. At a leveed floodplain reclamation site in Texas, ridges 1.9 m higher than flats were created to resemble meander scrolls, and mounds ∼30–40 cm above the flats and pools ∼30–40 cm below flats were created to replicate natural hummocks in bottomland forests; on the created ridges and mounds less flooding occurred than in pools, survival of most late-successional plant species was higher (ridges > mounds and pools > flats), but there was no difference in the survival of pioneer species, and soil nitrate, total nitrogen, and sulfur levels were lower, despite the fact that soil texture was the same on mound tops, side slopes, and pools (Simmons et al., 2011, 2012). The only effect of surface ridges and furrows on plant communities detected in an extensive study of 20 created depressional wetlands in Delaware was an increase in the percent cover of obligate wetland species (Alsfeld et al., 2009). Insect species richness and biomass showed no effect of microtopography but a positive correlation with coarse woody debris volume (Alsfeld et al., 2009). Two PNW sites were identified in the literature review, but the Willamette Valley wet prairie restoration publication only briefly discussed the microtopography at the reference sites, which was not central to the study (Pfeifer-Meister et al., 2012). For an experimental study of mounding during prairie restoration on a landfill in the Puget Sound lowlands, two studies were published a decade apart. The first showed that although 20 cm mounds have neutral or negative effects on some plant species, and mounding produces positive effects that far outstrip the effects of fertilizer and mulch on others, thereby significantly increasing plant growth, survival, and the number and height of inflorescences (Ewing, 2002). Ten years after creation, mounds now averaging 17 cm in height above “inter-mound” areas on what had become a grassland/wetland ecotone restoration site, had 1.3% lower soil moisture than inter-mound areas, significantly fewer obligate and facultative wetland species, and a higher proportion of native plant species; native indicator species for mound tops were adapted to drought stress and non-native indicator species were forbs with large taproots, while in between the mounds native indicator species were flood-tolerant wetland plants and non-native indicators were graminoids tolerant of both flood and drought (Hough-Snee et al., 2011). Three studies addressed faunal components of Texas coastal saltmarsh ecosystems. Rozas et al. (2005) conducted spatial analysis of five marsh restoration projects in Galveston Bay to compare proportions of vegetated marsh edge within 1 m of the shoreline, which was associated with densities of fish and crab in prior research. They showed that

3.1. Outreach and literature review The full synopsis of discussions with restoration practitioners including ecological and practical considerations is provided in supplementary material (Appendix A) and summarized here. In our outreach to regional restoration practitioners, we heard that practical considerations (i.e., material handling and hauling in a wet environment), and therefore cost is the primary driver for the inclusion of mounds in restoration designs when upland placement of excavated material is not an option within the acquisition boundary. Ecological components of mound restoration design such as size, location, configuration and composition are not well established by LCRE practitioners and each has approached design relatively independently from each effort and the available literature. The use of mounds within restoration project designs in the LCRE is rapidly increasing because of regrading requirements associated with other elements of project design (i.e., channel excavation and levee removal). When designing restoration projects, practitioners are thinking about mimicking the natural topography of the landscape to the extent possible, though size and configuration are also driven by the quantity of material requiring relocation. Practitioners are thinking about providing habitats with trees and shrubs in addition to emergent vegetation, and the possibility of shading marsh and channel areas. The primary design guideline used concerns elevation; designers keep the mounds below the 2-year flood elevation (ERTG, 2013) and/or regulatory limits on jurisdictional wetlands to avoid converting wetlands to uplands. Biological components such as the effects of aspect and slope on moisture and photosynthetically active radiation, soil type, and organic matter content, are not currently considered in mound design, though in one case an effort was made to steepen slopes to minimize mid-elevation habitat suitable for reed canary grass. Mound stability and slope considerations relative to the potential for erosion are elements of the engineering process. Planting success on mounds has been variable and often requires multiple years of planting for plants to become established. The primary literature review search string yielded 75 records (Appendix A). Key questions or uncertainties expressed by LCRE practitioners are not well represented in the published literature. Although many of the papers focused on microtopography, the microtopography studied was typically much finer than the mounds currently being designed on the LCRE, i.e., on the order of tussocks (Werner and Zedler, 2002) or furrows of a smaller size that added roughness to the land surface. Other papers concerned depressions, pools, or tidal creeks below the marsh plain elevation, not mounds or other features above the elevation of the marsh (e.g., Larkin et al., 2008), and therefore are not reported here. Only four records directly relevant to the objectives of this study were found for estuarine and tidal freshwater wetland systems. Therefore, the 17 papers most relevant to mound construction in the LCRE (Table A.2) also include findings from mounds in other wetland ecosystems concerning differences in environmental controls (e.g., moisture, temperature) related to aspect or elevation, and the associated effects on plantings. While the intent of the review was to focus on restoration, four of the 17 papers offered highly relevant results of experimental, controlled observational, or modeling efforts, though they did not directly involve restoration sites. Predictably, flooding is consistently lower at higher elevations, and microtopography can help to retain water in previously graded agricultural land being restored, but the responses of soil and plant-community metrics across studies are inconsistent. For example, two studies of microtopography at a natural deciduous forested wetland on the Hudson River floodplain (Courtwright and Findlay, 2011) and a 3-yearold deciduous forested restoration site on the North Carolina coastal plain (Bruland and Richardson, 2005) documented similar effects on hydrology but contrasting characteristics of soils and plant communities. Courtwright and Findlay (2011) found that on hummocks 217

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

top of the mound, the top of slope, and the upper side of the mound (Ruby Lake n = 49, p < 0.0041 for the three analyses) (Fig. S2). Differences in the minimum and maximum values of moisture between the toe and the top of mound ranged from 2.9% to 40% with a median of 15.2% for minimum values and 27.6% for maximum values (Table B.1). There were some indications of similar trends along the sides of some other mounds. Soil moisture at Anderson Creek Mound 1 (n = 2) ranged from 43% to 51% at the top of the mound and was too wet to measure at the toe. At Anderson Creek Mound 2 (n = 4), soil moisture was 43% near the channel and ranged from 15% in very loose soil to 40% in decayed wood substrate mid mound. As expected PD (%) from duplicates was generally smaller than those from replicates for all measurements (Table B.2). Further, the PD for soil moisture (%) was greater than the 5 cm depth soil temperature PD and both were greater than the 15 cm depth soil temperature PD. The 95% CI for 5 cm depth soil temperature PD (%) for duplicates and replicates was 0 to 5% and 4 to 9% respectively. The 95% CI for soil moisture PD (%) for duplicates was 7 to 30%, which may reflect disturbance of the soil on the second sample. The 95% CI for soil moisture PD (%) for replicates was 15 to 42%, reflecting the high local spatial variability of soil structure and permeability. The 95% CI for 15 cm depth soil temperature PD (%) for duplicates and replicates was 1 to 3% and 2 to 5% respectively, indicating relatively high precision and low spatial variability at this depth. Several analyses of soil temperature also produced significant and interesting results suggesting a positive correlation with elevation (Table 1). On 11 of 14 transects at Colewort Creek, Drift Creek, and Ruby Lake (those with more than 5 observations) soil temperature at 5 cm was positively correlated with elevation. At 10 of the 14 transects soil temperature at 15 cm was positively correlated with elevation. Temperature was significantly different between the toe and top of the mound at Colewort Creek Mound 2, and between the lower side and top of slope when tops and toes were removed. For all transects at Colewort Creek but not at Drift Creek, median soil temperature at the 5 cm depth was significantly greater at the top of the slope than at the lower side (n = 20, p = 0.0035). At Colewort Creek, the temperature at the 5 cm depth was significantly cooler on the lower sides of mounds than at the top of the slope. Soil temperatures at Anderson Creek Mound 1 were more variable at 15 cm depths, where they ranged from 14.4 to 17.8 °C, than those measured at 5 cm depths, where they ranged from 15 to 16.4 °C. At Anderson Creek Mound 2, soil temperatures at 5 cm depths (ranging from 16.1 to 17.5 °C) were greater than or equal to temperatures at 15 cm depths (ranging from 15.8 to 16.7 °C). Soil temperatures at Marietta Slough Mound 1 ranged from 16.4 to 20.6 °C at 5 cm depths and from 14.4 to 16.7 °C at 15 cm depths (Fig. B.3). At Mound 2, soil temperature from two locations ranged from 18.9 to 19.4 °C at 5 cm depths and 16.4 to 17.8 °C at 15 cm depths. At Mound 3, soil temperature ranged from 18.9 to 20.8 °C at 5 cm depths and 16.7 to 18.1 °C at 15 cm depths. Although examination of the data for temperature and moisture relative to aspect indicated that there could be substantial differences between the sides of a particular mound, the differences were not consistent between mounds. For example, the soils on the north and west sides of Colewort Creek Mound 2 were moister than those on the south and east sides, while soils on the east side of Ruby Lake Mound RL-0 (the reference mound) were significantly moister than those on the west side (n = 7, p = 0.034). Examination of residuals on Colewort Creek Mound 2 supported parametric analysis with ANOVA and Tukey’s pairwise comparison, which showed a significant difference between soil moisture on the south and west sides (Table 1). Aspect was nearly significant (n = 11, p = 0.052) for the western mound at Drift Creek, where moisture is high on the eastern side along the channel. We plotted the elevation of mounds against soil temperature at 5 cm and 15 cm depths, and soil moisture at the 12 cm depth, for all transects at Drift Creek, Colewort Creek, and Ruby Lake (Fig. 5, Figs. B.1–B.2). The number of samples collected at Anderson Creek and Marietta

terraces and some island designs produced more marsh edge than mounds. Armitage et al. (2013) found that the difference in patch size between marsh mounds 0.5 m in diameter and terraces > 50 m long only affected the response to grazing of one of several salt-marsh arthropod guilds studied. Two years after construction of marsh mounds made from either onsite soil excavation or offsite dredged material, which were located in different water depths, a comparison of the construction methods by Armitage et al. (2014) suggested much overlap between the fish communities, no difference in aquatic invertebrates, and overall only subtle differences despite extensive data collection on the water, soils, and plants. A few non-restoration studies focused on wetland plant response to topographic heterogeneity nevertheless offered results relevant to tidal wetland restoration in the PNW. Most of the genera and a few species in the experimental wetlands created by Vivian-Smith (1997) are the same as those of PNW wetlands, so we report her findings that species richness and evenness were consistently increased by heterogeneous microtopography on the order of 1–3 cm. Yet the studies all show that at the site and ecosystem scale, the presence of hummocks increases complexity in hydrology, soils, and plant communities. Similarly, the sedges Carex obnupta and C. lyngbyi are common in marshes of the PNW, so we include the findings of Peach and Zedler (2006) regarding the ecosystem engineering role of 15–25 cm tall C. stricta tussocks in enhancing species richness through the increase surface area by 40%, provision of micro-sites, and changes in composition by season. A restoration study of C. stricta by Doherty and Zedler (2015) also showed that mounds 32 cm in height were dryer than shorter ones, and that a topographically heterogeneous restoration site, by providing a range of dry to wet sites, can help ensure survival and biomass production of some species every year despite inter-annual environmental variability in the hydrologic regime. 3.2. Field research 3.2.1. Soil moisture and temperature In general, moisture was negatively correlated with elevation at all mounds, and this was significant for certain transects on certain mounds (Fig. 4). Analysis by relative vertical location on the mound (toe, lower side, upper side, top of slope, and top of mound) showed that median moisture was significantly greater at the toe of the mound than at the top of the mound for all mounds at two sites (Drift Creek n = 12, p = 0.0011; Colewort Creek n = 12, p = 0.0038) (Fig. 5, Fig. B.1) (Table 1). At the site with the most intensive sampling, median moisture was significantly greater at the toe of the mound than at the

Fig. 4. The moisture content at five relative vertical locations on mounds at Colewort Creek, Drift Creek, and Ruby Lake. The horizontal lines are percentiles 10, 25, 50, 75, and 90. The black dots are observations that lie outside of the extreme percentiles. 218

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

Fig. 5. Topography of two mounds at the Drift Creek study site with temperature and moisture data at relative locations along transects.

restoration-construction activities. The mounds at Anderson Creek were unique in that they had been formed by placing a dead tree, including the root wad, at an angle to the creek, pushing dirt up to it on one side and planting trees in the dirt. Based on examination of the mounds at Colewort Creek, the source of mound material, i.e., whether it is from the bottom of a slough or the topmost layer of a floodplain, appears to result in visibly different soil color, texture, and organic content. At most sites, mounds were stand-alone features surrounded by floodplain. At Seal Slough heterogeneity was achieved by constructing mounds and features mimicking natural river levees, and several practitioners interviewed had constructed or were planning on building similar mound features. Another exception was Drift Creek, where mounds had been designed to replicate natural alluvial fan structures in the coastal mountain region, and extended from the roadbed out onto the floodplain. Also, one mound at Colewort Creek was immediately adjacent to a forested hillslope from which it extended, and some mounds at Ruby Lake extended similarly from higher-elevation treed areas. It was evident from outreach that avoiding compaction or smoothing of the surfaces of mounds is understood to be beneficial for water penetration and plant growth, but we observed that this was not always possible given the environmental conditions at the time of construction (e.g., rain, relative consolidation of the sediments), and very hard consolidated mounds can result. At least one CEERP practitioner observed this issue on mound construction where the equipment operator smoothed the mound during construction resulting in a roughly 5 cm hardened ‘skin’ surface.

Slough was insufficient for statistical analysis (Fig. B.3). The variability in elevation within each of the five relative locations within a mound was small (CV ≤ 9.2%). Except for mounds with a maximum elevation range less than 76 cm (e.g., DC-1 and DC-2), the variability in average elevation between relative locations was 1 to 3 times greater than the variability within relative locations. Except for Ruby Lake Mound RL-4, the maximum CV for temperature measured at the 5 cm depth within mounds and relative locations was similar to the variability in elevation, CV = 10%. (Mound RL-4 had only 2 observations within the toe and lower side and had a CVs of 31% and 22%, respectively.) Except for Ruby Lake Mound RL-3, the variability in temperature measured at 15 cm depths within mounds and relative locations tended to be less than or similar to the variability in temperature measured at 5 cm depths. (The variability in temperature measured at 15 cm depths at Mound RL-3 within relative locations was slightly more variable (2%) than the measurements at 5 cm depths.) The variability in soil moisture within mounds and relative locations ranged from CV = 5% to 71.5% with an average of 30%. 3.2.2. General observations The heights above the marsh plain of mounds examined ranged from 0.5 m to 1.5 m (Table 2). At Anderson Creek, mounds were about 0.5–0.6 m above the marsh plain—i.e., the diameter of the log in addition to some material below and alongside it. Mounds bore distinctively different appearances at the restoration sites in terms of height, shape, and vegetation (Table 2) (Fig. 6). At four of the five sites, the mounds had been formed from fill material available from

219

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

Table 1 Statistical analysis results. Summary of statistical analysis of soil moisture and temperature and driving variables at mounds. For each pair of driving and response variables, the first row is the p-value for the Kruskal-Wallis test of equal medians (α = 0.05) and the second row is the significant pairwise comparison and p-value (individual α = 0.008; outlined with a black box) or the observed trend between medians when the overall test p-value was p < 0.1 but pairwise comparisons were not significant. (NS = not significant and NA = no analysis because of too few observations.)

(a)

In one case, the overall Kruskal-Wallis test was not significant but the pairwise comparison was significant, which can happen based on the structure of the variability.

220

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

4. Discussion

Table 2 The general morphology of fifteen mounds at five tidal wetland restoration sites in the Pacific Northwest. Year Constructed

Mound Code

Height (m)

Area (m2)

Anderson Creek

2001

Colewort Creek (CEERP)

2012

Drift Creek

2005

Marietta Slough

2002

Ruby Lake (CEERP)

2013

AC-1 AC-2 CC-1 CC-2 DC-1 DC-2 MS-1 MS-2 MS-3 MS-4 Refa RL-1 RL-2 RL-3 RL-4

0.6 0.5 1.5 1.5 0.6 0.7 0.6 1.5 1.0 0.9 0.6 1.2 1.1 1.0 0.9

21.2 21.1 730.7 977.5 849.6 544.3 ND ND ND ND ND 2670.1 5184.8 3889.8 3606.7

The elevation of coastal and river-floodplain wetlands is receiving considerable attention because of coupled changes in basin-scale processes and sea level caused by climate change, which alter historical patterns of flooding. The potential functions of tidal wetlands include flood protection (Möller et al., 1999; De Jonge and De Jong, 2002). Coastal ecosystem restoration is a management technique that frequently changes the topography of the site, thereby deliberately altering its relationship to hydrologic processes of surrounding waterways, and in turn the surface and groundwater inundation regime of the wetland and adjacent terrestrial areas (Frenkel and Morlan, 1991). The use of topographic heterogeneity is an important design element being implemented in the LCRE primarily for practical reasons associated with material management, but ecological justifications have also been identified. Practitioners developed design approaches for size, location, configuration and composition that are related to site-specific hydrology, historical landform and features, and vegetation composition and diversity (Fig. 3). This literature review (Table A.2) supported the concept that topographic heterogeneity can be a “bet-hedging strategy,” helping to ensure that individuals from some species survive and reproduce at the site despite the variability in environmental factors such as hydrology, as described at the tussock scale by Doherty and Zedler (2015). Some papers described in the review found species richness or evenness was higher on mounds (Vivian-Smith, 1997; Peach and Zedler, 2006; Courtwright and Findlay, 2011), while others found that it was lower (Bruland and Richardson, 2005), though all of the mounds in these studies were generally smaller in area and height above marsh plain than the mounds being constructed in the LCRE. Nevertheless, the identification of plant associations on mounds that are different from other areas of the wetland supports the proposition that topographic heterogeneity functions to hedge bets for plant species survival at the site scale. Restoration practitioners in the CEERP interviewed in this study identified bet-hedging relative to sea-level rise as one rationale for the construction of mounds. The idea that increasing the number of micro-sites with differing environmental conditions makes habitats available for more species is consistent with fundamental concepts in ecology (Harper et al., 1965; Huenneke and Sharitz, 1986). Though it is well-established in the literature that features above the marsh or floodplain are less frequently flooded (Table A.2), only a few studies measured the results for soil moisture on mounds at restoration sites, a crucial condition for plant establishment in the study region, and the mounds in all but one study were substantially smaller than those under construction in the LCRE (Ewing, 2002; Bruland and Richardson, 2005; Hough-Snee et al., 2011; Doherty and Zedler, 2015). Data collected using methods very similar to ours (summer, 12 cm depth, a similar instrument), 10 years after construction at another PNW restoration site, showed 1.3% lower soil moisture on the mound top relative to intermound areas (an average of 4.1% moisture on mound tops and 5.4% within the inter-mound area; Hough-Snee et al., 2011). This is consistent with the statistically significant difference between mound tops and toes that we showed at Colewort Creek, Drift Creek, and Ruby Lake, which strongly suggested that the mounds can stratify in terms of soil moisture (Tables 1 and 2). However, soil moisture was typically greater in our study, on average 33% at the toe and 16% at the top of mound. HoughSnee et al. (2011) also analyzed aspect, but it was not a significant explanatory variable for soil moisture or plant-community characteristics. We found significant but inconsistent relationships between aspect and soil moisture at two and possibly three out of the nine mounds. An explanation for the difference was not clear from the variables we studied. Temperature appears to be positively correlated with elevation at the mounds we studied, which suggests that temperature-driven evaporation drives moisture stratification. Longer studies have shown a microtopography by time interaction for soil temperature and moisture (Bruland and Richardson, 2005).

ND means the area of the mound was not delineated. CEERP means the restoration project is part of the Columbia Estuary Ecosystem Restoration Program. a Ref (RL-0) is a mound used for reference during restoration design and monitoring.

3.2.3. Vegetation and hydrologic regime Observed differences in vegetation appeared to be associated with the position of mounds relative to bodies of water and the original planting plans (Fig. 6). Both Anderson Creek and Drift Creek were near the head of tide. The mounds at Drift Creek, intended to imitate alluvial fans and among the lowest in height above marsh plain that we measured, were not planted and natural recruitment resulted in red alder (Alnus rubra) as the only tree, invasive non-native Himalayan blackberry (Rubus armeniacus) dominating the shrub layer, and a variable herb layer dominated by reed canary grass. In contrast, the mounds at Anderson Creek, which also had relatively low height above the marsh plain, intended to imitate nurse logs intersecting with slough channels, supported successful plantings of willow (Salix spp.) and Sitka spruce (Picea sitchensis) (aside from the deleterious effects of spruce budworm), creating a shady environment immediately over and around slough channels. The mounds highest above the marsh plain that we measured, at Marietta Slough and Colewort Creek, were planted with willows, shrubs, and Sitka spruce with variable success. Those at Marietta Slough that were planted with Scouler’s willow (Salix scouleriana) and Sitka spruce produced a low-canopied, shaded environment and some development of understory plant species richness, while Pacific willow (Salix lucida) resulted in a relatively higher-canopied and brighter environment with a reed canary grass understory. Observed differences in plant mortality and the vigor of plantings appeared to correspond to differences in soil organic matter (e.g., at Colewort Creek). Seal Slough mounds were densely planted with a variety of woody species with variable success, due in large part to herbivory by beaver (Castor canadensis) and elk (Cervus elaphus). Long term survival and vigor related to soil and moisture conditions has not been assessed at Seal Slough and the earlier restoration sites we studied. Mounds at Ruby Lake were seeded with tufted hairgrass (Deschampsia cespitosa), which was outcompeting the non-native surrounding reed canary grass. The mounds were subsequently planted with willows, with variable success, and extensive early natural recruitment of cottonwood was evident. The plants that successfully established evinced an elevation gradient from the top of mound to the water’s edge in many cases, such as Ruby Lake, where native species wapato (Sagittaria latifolia) is at the water’s edge with tufted hairgrass above. The relationship to water also controlled herbivory at Seal Slough, where willows planted on the side of mounds adjacent to the channel were eaten by beaver and those on the top and far side of the mounds were not. 221

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

Fig. 6. Photographs of mound features observed in this study. At Anderson Creek, nurse-log based mounds at water’s edge were planted with Sitka spruce (left) and willows (right). At Colewort Creek, mounds were fenced for protection from herbivory, and substrate was floodplain topsoil with relatively high plant survival and vigor (left) compared to material from channel excavation (right). Mounds at Drift Creek, shaped and positioned as alluvial fans, were not planted and invasive species are now dominant. Mounds at Marietta Slough had contrasting understory light availability and species richness associated with mixed taller-canopy deciduous (left) and shorter coniferous (right) plantings. The gradation of plant species from top of mound to water’s edge is evident at Ruby Lake. Mounds at Seal Slough showed more evidence of herbivory near the channel (left) than at other locations (right).

(Diefenderfer and Montgomery, 2009). With the exception of the hummocks formed from large stumps and root wads, however, the hummocks in Sitka spruce forested wetlands generally have less vertical relief than those we examined at restoration sites on the Columbia River. Some of the original published work on wetland microtopography in fact involved studies of woody species in floodplain swamps (Titus, 1990; Huenneke and Sharitz, 1986), extended by later studies (Bledsoe and Shear, 2000). Mounds being constructed on the Columbia River floodplain are below the 2-year flood elevation and/or regulatory limits on jurisdictional wetlands, but many are in excess of

Much of the research on restoration sites identified in this literature review was from the southeastern U.S. coast where forested wetlands or “swamps” are naturally hummocky features like we have previously documented for the native Sitka spruce forested wetlands of the tidal freshwater Columbia River floodplain (Diefenderfer et al., 2008). Like the mounds created around logs at the Anderson Lake restoration site on the outer Pacific coast (Fig. 6), and those on the Hudson River floodplain described by Courtwright and Findlay (2011), we have observed that hummocks in forested wetlands of the Columbia River floodplain occur around wood that is widespread in these ecosystems 222

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

more moisture from tides during summer drought months, affecting plant establishment. However, in fact the sizes of mounds observed in restoration designs in the LCRE are often in between these two poles. Third, the relative utility of mounds in different ecosystem settings (e.g., restored marsh, shrub-dominated wetland, and surge plain forested wetland) and river or estuarine reaches is unresolved. Identifying and characterizing the types of features that occur naturally in the LCRE (e.g., bar and scroll, natural levee, alluvial fan, and tree fall), and their association with types of hydrology, geomorphology, and plant communities in reference conditions would provide templates for restoration design (cf. Cannon, 2015).

the 1 m upper bound on microtopography described by Moser et al. (2007) (although given the heights attained by shrubs west of the Cascade Mountains in the PNW, the mounds conform to the within plant height component of the definition). Practitioners also raised the history of subsidence as a rationale for mound construction. Subsidence behind dikes due to ecosystem processes including compaction, drying, and diminishing organic matter content can be severe and limiting to restoration potential because of the disequilibrium between land-surface elevation and the hydrologic regime in adjacent waters (e.g., Beauchard et al., 2011). At one Sitka spruce swamp restoration site on the Columbia River floodplain, which had been agricultural grazing land prior to dike breaching, we estimated that it would take > 50 years until elevations became comparable to a reference site and suitable for the establishment of woody plants, based on the post-breaching sediment accretion rate (Diefenderfer et al., 2008). It was clear from our interviews (Appendix A) that planting shrubs and trees on mound features is the typical approach taken and it is understood that, as the literature review indicated, obligate wetland plants will not fare as well (Rossell et al., 2009). Given documented rates of subsidence and sediment accretion in the system, this has the potential to accelerate the introduction of woody species to restoration sites. However, not all sites restored through the CEERP are intended to restore forested wetlands; some are intended to become marshes, and some practitioners expressed wariness about introducing atypical topographic features onto historically marsh plains. The definition of a marsh does permit patches of shrubs (Cowardin et al., 1979), and it is known that large wood can collect on PNW marshes (Eilers, 1975), but we observe that topographic mounds of the scale currently being constructed are outside the range of reference marshes in the region (Borde et al., 2011) and above current 50-year relative sea-level rise scenarios for the region. Practitioners mentioned two other geomorphic features being replicated by the largest mounds: natural levees and bar and scroll. Floodplain ridges were also mentioned by Simmons et al. (2011, 2012) where 1.9 m ridges were constructed, and by Rossell et al. (2009) where 0.5 m ridges were constructed.

4.2. Conclusions and implications for restoration practice In conclusion, only 13 published records meeting the criteria for restored sites with constructed mounds in tidal areas were identified by the literature review though additional experimental studies were useful. Responses of soil and plant-community metrics were inconsistent in the literature reviewed, except that flooding was consistently lower at higher elevations. The stratification of soil moisture by elevation was strongly suggested by our statistical analysis of 2015 data for the tops, sides, and toes of 15 mounds constructed at 5 tidal wetland restoration sites in the PNW region between 2001 and 2013 of fill dirt, or dirt and logs. Study sites represented the Columbia River, Puget Sound, and outer Pacific coastal estuaries. Soil temperature also appeared to be positively correlated with elevation, though statistical analysis was less conclusive. Soil moisture relative to mound aspect was significant in some cases, however, it was inconsistent between mounds. Outreach to PNW restoration practitioners resulted in numerous insights gained from the design, construction, and monitoring of 21 individual restoration projects including mounds, 19 of which were tidal. Our findings will aid engineering design of wetland restoration sites where topographic heterogeneity is an objective. Several implications for restoration practice in mound design emerge from these findings:

• Consider project goals during mound design; e.g., for sites with

4.1. Remaining uncertainties and future research On balance, considering the results of literature review, outreach, and data analysis, three remaining uncertainties stand out in regard to mound design in the CEERP. First, typical planting success is unknown, as are the most efficacious methods for establishing a viable native plant community under variable tidal-fluvial hydrologic regimes including summer-dry soils on mounds. Rigorous experimental investigations such as those by Doherty and Zedler (2015) are promising approaches to addressing these questions. In the Pacific Northwest, intensive research on shrub and tree species establishment on mounds is being undertaken on non-tidal river floodplains (Latterell et al., 2014; Hartema and Latterell, 2015a,b; Appendix A), and its utility in the reduction of plant mortality and associated costs in tidal regions should be evaluated. Additionally, a systematic approach to assessing the relative tolerance of locally important native plants and plant associations could provide the basis of a list of general planting recommendations for the relative vertical positions on mounds as a tool for practitioners. Secondly, although we had the opportunity to compare mounds of different sizes and shapes (Fig. 5, Figs. B.1–B.2), they were not replicated or randomly selected, so much remains to be learned about the pros and cons of differences in the size, shape, and configuration of mounds that can be investigated through spatial analysis techniques (Rozas et al., 2005). In regard to mound size, potential advantages of larger mounds include less edge and more canopy cover, i.e., environments more like interior woody plant communities, which may help protect the interior area from herbivory. On the other hand, a larger number of smaller mounds may better mimic the microtopography or hummocky environment typical of forested wetlands and may receive

• • • •

• • 223

goals to restore historically forested wetlands, reed canary grass may be shaded out by designing many small mounds in very close proximity to mimic forested wetland microtopography and using spruce and woody plants to achieve shading. Think in terms of relative vertical position when designing planting plans, because volumetric water content (moisture) is negatively correlated with elevation. Evaluate the importance of soil moisture relative to locally important native plants and plant associations, using the hydrologic regime and elevation data as the design basis, to produce a list of general planting recommendations for the different vertical positions on mounds. Evaluate the potential effects of available light and moisture sources relative to aspect site-specifically, because findings related to moisture and aspect differed by site. Consider the source of mound material, whether it is from the bottom of a slough or the topmost layer of a floodplain, especially regarding organic matter content. If possible, place topsoil on the top layer of mounds to enhance plant vigor and success. (In some cases, qualitatively observed differences in plant mortality and the vigor of plantings appeared to correspond to differences in soil organic matter and/or soil moisture.) Consider the potential for a weedy seed bed, and perhaps implementing an intervening year (or likely more) of control to eradicate weed seeds before topsoil is moved to the top of the mound and hydrology is reconnected. Additional weed control may be needed in subsequent years. Consider the presence and type(s) of topographic variability when examining reference site and historical conditions to inform

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

restoration project design. For instance, quantitative assessment of the size (elevation, width, and length), shape, and density of higherrelief features on the landscape, and their collective distribution across the landscape particularly relative to water, will help to determine the suitability of such features on the restoration site.

Cardno ENTRIX, 2013. Design Report: Kandoll Farm Wetland Restoration Project Phase 2. June 2013. Prepared for Columbia Land Trust, Vancouver, Washington by Cardno ENTRIX, Seattle, Washington. CREST (Columbia River Estuary Study Taskforce), 2013. Habitat Restoration Design and Implementation Project Application: Sauvie Island North Unit Wetland Restoration and Enhancement Project. Lower Columbia Estuary Partnership, Portland, Oregon. Cornu, C.E., 2005. Restoring Anderson Creek Marsh. South Slough NERR Coastal Resource Management Series. CRMS-2005-3. Coos Bay, Oregon. Cowardin, L.M., Carter, V., Golet, F.C., LaRoe, E.T., 1979. Classification of Wetlands and Deepwater Habitats of the United States. U.S. Department of the Interior Fish and Wildlife Service, FWS/OBS-79/31. Office of Biological Services, Washington. Courtwright, J., Findlay, S.E.G., 2011. Effects of microtopography on hydrology, physicochemistry, and vegetation in a tidal swamp of the Hudson River. Wetlands 31 (2), 239–249. De Jonge, V.N., de Jong, D.J., 2002. Ecological restoration in coastal areas in the Netherlands: concepts, dilemmas and some examples. Hydrobiologia 478, 7–28. Diefenderfer, H.L., Coleman, A.M., Borde, A.B., Sinks, I.A., 2008. Hydraulic geometry and microtopography of tidal freshwater forested wetlands and implications for restoration, Columbia River, U.S.A. Ecohydrol. Hydrobiol. 8, 339–361. Diefenderfer, H.L., Johnson, G.E., Skalski, J.R., Breithaupt, S.A., Coleman, A.M., 2012. Application of the diminishing returns concept in the hydroecologic restoration of riverscapes. Landscape Ecol. 27, 671–682. Diefenderfer, H.L., Johnson, G.E., Thom, R.M., Buenau, K.E., Weitkamp, L.A., Woodley, C.M., Borde, A.B., Kropp, R.K., 2016. Evidence-based evaluation of the cumulative effects of ecosystem restoration. Ecosphere 7 (3), e01242. https://doi.org/10.1002/ ecs2.1242. Diefenderfer, H.L., Montgomery, D.R., 2009. Pool spacing, channel morphology, and the restoration of tidal forested wetlands of the Columbia River, U.S.A. Restor. Ecol. 17, 158–168. Doherty, J.M., Zedler, J.B., 2015. Increasing substrate heterogeneity as a bet-hedging strategy for restoring wetland vegetation. Restor. Ecol. 23 (2), 15–25. Drouin, A., Saint-Laurent, D., Lavoie, L., Ouellet, C., 2011. High-precision elevation model to evaluate the spatial distribution of soil organic carbon in active floodplains. Wetlands 31 (6), 1151–1164. Ebberts, B.D., Zelinsky, B.D., Karnezis, J.P., Studebaker, C.A., Lopez-Johnston, S., Creason, A.M., Krasnow, L., Johnson, G.E., Thom, R.M., 2017. Estuary ecosystem restoration: Implementing and institutionalizing adaptive management. Restor. Ecol. 26, 360–369. https://doi.org/10.1111/rec.12562. Eilers, H.P., 1975. Plants, Plant Communities, Net Production and Tide Levels: the Ecological Biogeography of the Nehalem Salt Marshes Tillamook County, Oregon (Ph.D. dissertation). Oregon State University, Corvallis. ERTG (Expert Regional Technical Group), 2013. ERTG Analysis of Water Levels for Site Delineation in Tidal-Dominated Regions. ERTG 2012-01, Rev 1, prepared for the Bonneville Power Administration, U.S. Army Corps of Engineers, and NOAA Fisheries. Portland, Oregon. Available from http://cbfish.org/ERTG. Ewing, K., 2002. Mounding as a technique for restoration of prairie on a capped landfill in the Puget Sound lowlands. Restor. Ecol. 10, 289–296. French, P.W., 2006. Managed realignment – the developing story of a comparatively new approach to soft engineering. Estuar. Coast. Shelf Sci. 67, 409–423. Frenkel, R.E., Morlan, J.C., 1991. Can we restore our salt marshes? Lessons from the Salmon River, Oregon. Northwest Environ. J. 7, 119–135. Gallego Fernández, J.B., García Novo, F., 2007. High-intensity versus low-intensity restoration alternatives of a tidal marsh in Guadalquivir estuary, SW Spain. Ecol. Eng. 30, 112–121. Harper, J.L., Williams, J.T., Sagar, G.R., 1965. The behavior of seeds in soil I. The heterogeneity of soil surfaces and its role in determining the establishment of plants from seed. J. Ecol. 53, 273–286. Hartema, L.P., Latterell, J.J., 2015a. Patterson Creek Trib 383 Wetland Mitigation Final Monitoring Report: Year 10. King County Water and Land Resources Division, Seattle, WA. Hartema, L.P., Latterell, J.J., 2015b. Pautzke levee Setback and Floodplain Restoration Project, 2014 Monitoring Report: Year 5. King County Water and Land Resources Division, Seattle, WA. Hough-Snee, N., Long, A.L., Jeroue, L., Ewing, K., 2011. Mounding alters environmental filters that drive plant community development in a novel grassland. Ecol. Eng. 37, 1932–1936. Huenneke, L.F., Sharitz, R.R., 1986. Microsite abundance and distribution of woody seedlings in a South Carolina cypress-tupelo swamp. Am. Midl. Nat. 115, 328–335. Inglis, G.J., 2000. Disturbance-related heterogeneity in the seed banks of a marine angiosperm. J. Ecol. 88, 88–99. Jarzemsky, R.D., Burchell, M.R., Evans, R.O., 2013. The impact of manipulating surface topography on the hydrologic restoration of a forested coastal wetland. Ecol. Eng. 58, 35–43. Jay, D.A., Borde, A.B., Diefenderfer, H.L., 2016. Tidal-fluvial and estuarine processes in the lower Columbia River: II. Water level models, floodplain wetland inundation, and system zones. Estuaries Coasts 39 (5), 1299–1324. Krueger, K., Bottom, D., Hood, G., Johnson, G., Jones, K., Thom, R., 2017. An expert panel process to evaluate habitat restoration actions in the Columbia River estuary. J. Environ. Manage. 188, 337–350. Larkin, D.J., Madon, S.P., West, J.M., Zedler, J.B., 2008. Topographic heterogeneity influences fish use of an experimentally restored tidal marsh. Ecol. Appl. 18, 483–496. Larkin, D., Vivian-Smith, G., Zedler, J.B., 2006. Topographic heterogeneity theory and ecological restoration. In: Falk, D.A., Palmer, M.A., Zedler, J.B. (Eds.), Foundations of Restoration Ecology. Island Press, Washington, D.C, pp. 142–164. Latterell, J.J., Hartema, L., Minella, H., Julvezan, K., Lantz, D., 2014. McElhoe-Pearson Project, 2013 Monitoring Report: Year 1 of 10. King County Water and Land

In closing, we emphasize the importance of developing clear restoration goals and matching mound designs to achieve them. While beneficial use of excavated material is a practical consideration for many projects, it is important that mounds be used in an ecologically appropriate manner that increases habitat function and long-term resilience. Acknowledgements This study, funded by Bonneville Power Administration, originated under the direction of B. Zelinsky and was subsequently directed by J. Karnezis. G. Johnson served as project manager and peer reviewer. We sincerely thank the restoration scientists, managers, and practitioners named herein for their generosity in discussions, site visits, and examination of project files. We appreciate the assistance in literature search provided by N. Sather, graphics by N. Johnson, and editing by S. Ennor. Supplementary Materials: Appendix A and Appendix B Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.ecoleng.2018.07.027. References Allen, J.R.L., 2000. Morphodynamics of Holocene salt marshes: a review sketch from the Atlantic and Southern North Sea coasts of Europe. Quat. Sci. Rev. 19, 1155–1231. Alsfeld, A.J., Bowman, J.L., Deller-Jacobs, A., 2009. Effects of woody debris, microtopography, and organic matter amendments on the biotic community of constructed depressional wetlands. Biol. Conserv. 142 (2), 247–255. Armitage, A.R., Ho, C.K., Madrid, E.N., Bell, M.T., Quigg, A., 2014. The influence of habitat construction technique on the ecological characteristics of a restored brackish marsh. Ecol. Eng. 62, 33–42. Armitage, A.R., Ho, C.K., Quigg, A., 2013. The interactive effects of pulsed grazing disturbance and patch size vary among wetland arthropod guilds. PLoS One 8 (10), e76672. Beauchard, O., Jacobs, S., Cox, T.J.S., Maris, T., Vrebos, D., Van Braeckel, A., Meire, P., 2011. A new technique for tidal habitat restoration: evaluation of its hydrological potentials. Ecol. Eng. 37, 1849–1858. Blackwell, M.S.A., Hogan, D.V., Maltby, E., 2004. The short-term impact of managed realignment on soil environmental variables and hydrology. Estuar. Coast. Shelf Sci. 59, 687–701. Bledsoe, B.P., Shear, T.H., 2000. Vegetation along hydrologic and edaphic gradients in a North Carolina coastal plain creek bottom and implications for restoration. Wetlands 20, 126–147. Borde A.B., Zimmerman, S.A., Kaufmann, R.M., Diefenderfer, H.L., Sather, N.K., Thom, R. M., 2011. Lower Columbia River and Estuary Restoration Reference Site Study: 2010 Final Report and Site Summaries. PNWD-4262, prepared for the Lower Columbia River Estuary Partnership and the Bonneville Power Administration by Pacific Northwest National Laboratory, Marine Sciences Laboratory, Sequim, WA. Bruland, G.L., Richardson, C.J., 2005. Hydrologic, edaphic, and vegetative responses to microtopographic reestablishment in a restored wetland. Restor. Ecol. 13 (3), 515–523. Bubier, J.L., Moore, T.R., Roulet, N.T., 1993. Methane emissions from wetlands in the midboreal region of northern Ontario, Canada. Ecology 74, 2240–2254. Burdick, D.M., Mendelssohn, I.A., 1987. Waterlogging responses in dune, swale and marsh populations of Spartina patens under field conditions. Oecologia 74, 321–329. Cabin, R.J., Clewell, A., Ingram, M., McDonald, T., Temperton, V., 2010. Bridging restoration science and practice: results and analysis of a survey from the 2009 Society for Ecological Restoration International meeting. Restor. Ecol. 18, 783–788. Callaway, J.C., Borde, A.B., Diefenderfer, H.L., Parker, V.T., Rybczyk, J.M., Thom, R.M., 2012. Pacific Coast tidal wetlands. In: Batzer, D.P., Baldwin, A.H. (Eds.), Wetland Habitats of North America: Ecology and Conservation Concerns. University of California Press,, Berkeley, CA, pp. 103–116. Cannon, C.M., 2015. Landforms Along the Lower Columbia River and the Influence of Humans (Master of Science Thesis). Portland State University, Portland, Oregon Paper 2231. Cantelmo, A.J., Ehrenfeld, J.G., 1999. Effects of microtopography on mycorrhizal infection in Atlantic white cedar (Chamaecyparis thyoides (L.) Mills.). Mycorrhizae 8, 175–180.

224

Ecological Engineering 123 (2018) 212–225

H.L. Diefenderfer et al.

treatments in riparian forest restoration. Restor. Ecol. 20, 369–377. Sinks, I., 2014. Restoring process and function to a freshwater tidal floodplain constrained by partial levee removal. River Restoration Northwest, 4 February 2014, Stevenson, Washington. Available URL: http://www.rrnw.org/2014-symposium-program/. Tessier, M., Gloaguen, J.-C., Bouchard, V., 2002. The role of spatio-temporal heterogeneity in the establishment and maintenance of Suaeda maritima in salt marshes. J. Veg. Sci. 13, 115–122. Titus, J.H., 1990. Microtopography and woody plant regeneration in a hardwood floodplain swamp in Florida. Bull. Torrey Bot. Club 117, 429–437. U.S. Forest Service and Mid-Coast Watersheds Council. 2004. Lower Drift Creek Restoration Management Plan. Siuslaw National Forest, Corvallis, Oregon. Available URL: https://www.fs.usda.gov/detail/siuslaw/landmanagement/ resourcemanagement/. Vestergaard, P., 1998. Vegetation ecology of coastal meadows in southeastern Denmark. Opera Botanica 134, 1–69. Vivian-Smith, G., 1997. Microtopographic heterogeneity and floristic diversity in experimental wetland communities. J. Ecol. 85, 71–82. Washington Department of Fish and Wildlife, 2008. Whatcom Wildlife Area 2008 Management Plan Update. Olympia, Washington. Available URL: https://wdfw.wa. gov/publications/00544/whatcom_2008update.pdf. Werner, K.J., Zedler, J.B., 2002. How sedge meadow soils, microtopography, and vegetation respond to sedimentation. Wetlands 22, 451–466. Zedler, J.B., Callaway, J.C., Desmond, J.S., Vivian-Smith, G., Williams, G.D., Sullivan, G., Brewster, A.E., Bradshaw, B.K., 1999. Californian salt-marsh vegetation: an improved model of spatial pattern. Ecosystems 2, 19–35. Zedler, J.B., Kercher, S., 2005. Wetland resources: status, trends, ecosystem services, and restorability. Ann. Rev. Environ. Resour. 30, 39–74.

Resources Division, Seattle, WA. Möller, I., Spencer, T., French, J.R., Leggett, D.J., Dixon, M., 1999. Wave transformation over salt marshes: A field and numerical modeling study from North Norfolk, England. Estuar. Coast. Shelf Sci. 49, 411–426. Moser, K., Ahn, C., Noe, G., 2007. Characterization of microtopography and its influence on vegetation patterns in created wetlands. Wetlands 27, 1081–1097. NPS (National Park Service), 2012. Colewort Creek Wetland Restoration Environmental Assessment. U.S. Department of Interior, National Park Service, Pacific West Region, Lewis and Clark Historical Park, Astoria, Oregon. Available URL: https:// parkplanning.nps.gov/. Peach, M., Zedler, J.B., 2006. How tussocks structure sedge meadow vegetation. Wetlands 26 (2), 322–335. Pfeifer-Meister, L., Johnson, B.R., Roy, B.A., Carreño, S., Stewart, J.L., Bridgham, S.D., 2012. Restoring wetland prairies: tradeoffs among native plant cover, community composition, and ecosystem functioning. Ecosphere 3 (12), 121. https://doi.org/10. 1890/ES12-00261.1. Pollock, M.M., Naiman, R.J., Hanley, T.A., 1998. Plant species richness in riparian wetlands—a test of biodiversity theory. Ecology 79, 94–105. Rossell, I.M., Moorhead, K.K., Alvarado, H., Warren, R.J., 2009. Succession of a southern Appalachian mountain wetland six years following hydrologic and microtopographic restoration. Restor. Ecol. 17, 205–214. Rozas, L.P., Caldwell, P., Minello, T.J., 2005. The fishery value of salt marsh restoration projects. J. Coast. Res. (40), 37–50 (SI). Simmons, M.E., Wu, X.B., Whisenant, S.G., 2011. Plant and soil responses to created microtopography and soil treatments in bottom land hardwood forest restoration. Restor. Ecol. 19, 136–146. Simmons, M.E., Wu, X.B., Whisenant, S.G., 2012. Responses of pioneer and later-successional plant assemblages to created microtopographic variation and soil

225