Sea-level rise (SLR) and biological invasions have the potential to drive significant changes in tidal marsh ecosystems, impacting habitat availability and species interactions. However, we understand little about the trophic links between aquatic and terrestrial habitats in tidal marsh ecosystems. To better understand SLR impacts, we experimentally increased inundation in situ using experimental weirs to simulate SLR and increased flooding in areas of California cordgrass (Spartina foliosa) at two tidal marsh sites for two consecutive growing seasons in China Camp State Park, San Rafael, CA, USA. To quantify potential interactions with non-native invasive species, we also enclosed invasive European green crabs (Carcinus maenas) within the inundation treatments at one of the sites. We examined changes in microalgal biomass and responses of insects and benthic invertebrates to increased inundation time and crabs, and use of habitats by birds to better understand the potential impacts on higher trophic levels. Our results showed benthic macrofaunal grazers, consumed by shorebirds, responded favorably to inundation at one site, likely in response to increased microalgal biomass in inundated treatments. In contrast, we found a negative effect of invasive crabs on benthic microalgae. Marsh insects, songbird prey, showed contrasting responses to increased inundation, with chironomids responding positively, and planthoppers (Prokelisia spp.) responding negatively. Insects were more abundant in cordgrass, where we also found more Song Sparrow (Melospiza melodia), Marsh Wren (Cistothorus palustris), and Common Yellowthroat (Geothlypis trichas) individuals, relative to mid elevation perennial pickleweed (Sarcocornia pacifica) areas. Overall, we found fewer Song Sparrows in 2022 relative to the prior year, potentially due to reduced cordgrass integrity coupled with lower chironomid abundance in 2022. Together, these results suggest cordgrass may well be a ‘leverage species’ whose decline may produce effects extending even beyond salt marshes. These findings underscore the importance of incorporating aquatic-terrestrial linkages into predictive models for effects of multiple stressors on coastal ecosystems and strategies for natural resource management, providing valuable insights for proactive conservation and sustainable coastal planning.

Site Description

This study took place in two sites within China Camp State Park (SP), located in San Rafael, CA, USA: Buckeye Point and Bullhead Flat (Fig. 1 in related manuscript). China Camp SP contains a large, remnant historic marsh. Buckeye Point is characterized by a low elevation band of California cordgrass (Spartina foliosa), that leads into an expansive middle/higher elevation area dominated primarily by pickleweed (Sarcocornia pacifica) with fleshy jaumea (Jaumea carnosa) and western marsh rosemary (Limonium californicum) scattered throughout. Channel edges within pickleweed habitat also contain marsh gumplant (Grindelia stricta), as well as cordgrass at the base of the channel interior. Higher elevation pickleweed habitat can also be intermixed with saltgrass (Distichlis spicata). In Buckeye Point, the study site primarily encompasses pickleweed dominated mid/high marsh, mudflat, and the low elevation band of cordgrass that bisects those two habitats. Bullhead Flat is characterized by a long, gently sloping mudflat leading to a lower elevation cordgrass band, followed by a thin band dominated by pickleweed that is cut off by a low-lying road. However, it is important to acknowledge the presence of transitional areas between these adjacent habitats at each site composed of varying densities of each plant species, and the interconnected nature of this system.

Experimental Design

We used three inundation treatments in both sites. In spring of 2021 at Buckeye Point, we installed eight weirs or ‘marsh boxes’ (1m x 0.6m x 0.6m), submerged 10cm into the sediment in areas of native cordgrass. Boxes were constructed of clear acrylic sheeting and intended to simulate increased inundation with SLR by allowing inflow during flood tides and then holding water by restricting outflow as the surrounding marsh drains during ebb tide. These plots were accompanied by eight unaltered areas of cordgrass for comparison (“ambient control”). As boxes may impact sediment characteristics and flow around the plots, we also installed eight marsh box controls, featuring shorter walls (0.2 m) with corner openings for free water flow, to account for any artifactual effects of the boxes. Surface elevation of plots ranged from -0.98m to -0.74m in 2021, and -0.93m to -0.73m in 2022 (heights all relative to MHHW). For control plots, we dug a thin trough around the plots to simulate disturbance from box or partial box installation and cut through any belowground root masses connected to the plot. Each set of three inundation treatments were randomly ordered within either six (2022) or eight (2021) blocks that were distributed at least 2 m apart across the experimental area parallel to the shore to address marsh habitat heterogeneity and account for factors like elevation and channel proximity. We also randomly distributed crab treatments within each inundation treatment (Fig. 2). The efficacy of the inundation treatments and response of soil parameters and cordgrass are described by Gonzalez et al. (2024).

Over winter 2021, several boxes filled with sediment and/or were broken due to waves or heavy tidal flows, affecting the cordgrass within and likely disrupting natural processes occurring in boxes. Therefore, in summer 2022, we reduced the number of replicates to six for all treatments and moved all boxes approximately one meter away from the previous box’s location into an area of previously unaffected cordgrass.

In summer of 2022, we replicated this experiment at nearby Bullhead Flat (Fig.1) and added an additional treatment where plots of existing cordgrass were enclosed with the European green crab Carcinus maenas. The experimental setup included enclosures (0.5m x 0.5m, made of 7mm Vexar mesh), with two green crabs added based on local densities. We also included control cages with no crabs, using the same cage structure, and cageless controls. Crabs were captured locally and two additional crabs were placed in each cage in July to account for potential losses. Even with crab additions, cage densities remained consistent with natural levels in San Francisco Bay, reaching up to four crabs per trap.

The three crab treatments in 2022 underwent a tidal inundation simulation to mimic the effects of SLR. Similar to Buckeye Point, we employed 'marsh boxes' (2m x 1m x 0.4m) parallel to the shore, increasing inundation by delaying water drainage. The boxes were partially submerged in mud, limiting lateral drainage, and had in-flow check valves for water entry, with no exit valve, ensuring slower drainage relative to controls. We also installed marsh box controls, as in Buckeye Point, and unmanipulated ambient controls that had no boxes. The inundation treatments were randomly assigned position in blocks as above and arranged as 2 x 1 m plots, spaced at least 2 m apart. Surface elevations of plots ranged from -1.00m to -0.823m (relative to MHHW). The experimental design comprised six blocks to account for marsh heterogeneity, and within each block we randomly assigned inundation treatments, and randomly distributed crab treatments within each inundation treatment.

In each of the 18 inundation treatment areas, we measured orthometric heights of the marsh surface using an RTK GPS GS07 GNSS receiver and CS20 LTE controller (Leica Geosystems). Elevation data were received via the Leica California SmartNet Real-Time Kinematic network. We measured geodetic benchmarks near each site pre- and post- sampling. For each sampling event, the difference between average recorded and the known benchmark value never exceeded 0.03m. Vertical and horizontal precision values never exceeded the standard deviation, and the Position Dilution of Precision (PDOP) was below 2 for all points.

We calculated inundation from water level data gathered using HOBO U20-L water level loggers (Onset Data Loggers, Cape Cod, MA) deployed in each inundation treatment with the sensor positioned on the sediment surface. Water level in each plot was calculated from temperature and pressure data using the Barometric Compensation Assistant in HOBOware Pro (version 3.7.23) and corrected using local barometric pressure data. From those water level data, we calculated average inundation time per inundation plot by summing the number of minutes where the water level was 2-4 cm above the marsh surface to account for any baseline noise in the data (2 cm in July 2021, 4 cm in September 2022). These values were then converted to inundation hours per day. See Gonzalez et al. (2024) for a full description of inundation calculation methodology.

Microalgal biomass, benthic macrofaunal invertebrates, and insects

We gathered data on several trophic levels that could respond to our experimental treatments. To quantify soil infauna and surface feeding invertebrates, we took a sediment core in each treatment plot at the end of each growing season in 2021 and 2022 using a 5cm PVC corer to a depth of 10cm. The sample was sieved through a 500μm mesh sieve, and invertebrates were fixed in 10% formalin, stored in 70% ethanol and sorted into broad taxonomic categories (Neira et al. 2006). In 2022, we observed an abundance of eastern mudsnails (Ilyanassa obsoleta) inside and outside of plots while sampling and conducted surface counts of individuals in plots in the final month of sampling that year at each site. We also quantified microalgal biomass, which is an important food source for many benthic invertebrates including larval Chironomidae (Diptera). Microalgal biomass was quantified with a second soil core (1.23cm² diameter x 5mm deep) for chlorophyll a concentration, which is a common proxy for microalgal biomass, in the top layer of the sediment according to Plante-Cuny (1973). We measured flying insect abundance by deploying sticky traps (Gideal brand, Sabo and Power 2002) over a 24-hour period in treatment plots approximately monthly throughout the experiment. Simultaneously, we deployed 6-8 sticky traps in the mid marsh area dominated primarily by pickleweed to compare with control treatments in low marsh cordgrass. Sticky traps were made of 310 cm² plastic sheets covered with Tanglefoot insect trap coating (Tanglefoot, Grand Rapids, Michigan, USA) and rolled in a cylinder around PVC posts just above the vegetation to capture insects flying from all directions. After field collection, each sticky trap was scanned to generate a digital image suitable for analysis in FIJI/ImageJ. The image was then overlaid with a grid, allowing for the counting and identification of insects in each cell, down to the lowest practical taxonomic level.

Bird abundance and use of habitat

We assessed potential impacts on higher trophic levels by quantifying bird presence and foraging in (1) fringing cordgrass meadow (low marsh), and two additional distinct habitats, 2) pickleweed plain (mid/high marsh), and 3) intertidal mudflat (low elevation, just below cordgrass). The fringing cordgrass meadow treatment included but was not constrained to the plots in the inundation/crab enclosure experiment. Bird species were identified and enumerated in low and mid/high marsh habitats using a standardized Area Search Census protocol (Point Blue 1999, Patten and O’Casey 2007). Each of the three habitats comprised one area, and we surveyed three replicate areas. We conducted points counts using a spotting scope (27-60x magnification, Vortex Optics, Barnveld, WI) to identify birds in mudflat/subtidal habitat. Surveys occurred during low tide and were timed to coincide with high bird densities in the fall, winter and spring migration periods for various groups. We completed four surveys in between September 2021 and January 2022 and four surveys between October 2022 and April 2023. In this study, we focus primarily on three species that were found consistently in the vegetated marsh habitats : Song Sparrow (Melospiza melodia), Marsh Wren (Cistothorus palustris) and Common Yellowthroat (Geothlypis trichas).