Study Site and Experimental Design
Our study was conducted on Harbor Island, near Port Aransas, TX (27.86N, 97.08W), in the Mission-Aransas National Estuarine Research Reserve. The vegetation at the site was initially dominated (90-100% cover) by A. germinans with patches (~10% cover) of salt marsh vegetation (mostly the succulents Batis maritima and Sarcocornia spp. and the grass Spartina alterniflora). In 2012, we established ten experimental plots (42 m 24 m) along the edge of the Lydia Ann Channel. We removed mangroves from the plots by cutting them at the soil surface to create a gradient of mangrove cover (nominally 0, 11, 22, 33, 44, 55, 66, 77, 88 and 100% mangrove cover). The plots were arranged in three blocks, with each block containing at least one low mangrove cover, one intermediate mangrove cover, and one high mangrove cover plot (FIG. S1, Guo et al. 2017). To facilitate maintenance and to simulate the natural patchiness of the vegetation, mangroves were removed or left in place within 3 3 m cells in a stratified random checkerboard pattern. Marsh vegetation naturally recolonized most cleared patches within 2 years (Guo et al. 2017). In each plot, we collected samples from four 3 × 3 m patches along the coastal fringe (fringe patches in the front third of each plot) and four 3 × 3 m patches within the plot interior (interior patches). Replicate cells (n = 8 per plot) were all marsh (in the 0% mangrove plot), all mangrove (in the 100% mangrove plot) or half marsh and half mangrove (in mixed-species plots) for a total of n = 80 replicates.
On 25 August 2017, experimental plots were directly affected by Hurricane Harvey, a Category 4 storm that made landfall nearby (FIG. S1). Plots were exposed to hurricane-force winds exceeding 119 kph for approximately 6 h, with gusts up to 225 kph (NOAA 2019). A tide gauge at Port Aransas, near the experimental plots, recorded a storm surge of 1.6 m above MLLW (NOAA 2019), and estimates of storm surge based on debris deposition and other flood evidence indicated a storm surge of up to 2.4 m (USGS 2019). Storm surge flooding (0.8 m above MLLW) persisted for approximately 6 h.
Soil Temperature
Throughout the experiment, we recorded soil temperature continuously (30-minute intervals) using HOBO temperature sensors (Onset Computer Corporation, Bourne, Massachusetts, USA) placed at 5 cm depth. Temperature sensors were inserted into interior patches of the 0% mangrove (100% marsh) and 100% mangrove plots (n = 2 total).
Subsurface Soil Sulfide Concentrations
In 2017 and 2018, we collected soil cores (5 cm diameter 30 cm depth) in each of the same randomized cells where feldspar marker horizons were established (n = 8 per plot) to measure chemistry in subsurface soils. We removed roots from cores and homogenized soils prior to chemical analysis. Subsamples were dried at 60C to a constant dry mass. We ground and homogenized portions with an 8000-D ball mill (Spex SamplePrep, Metuchen, New Jersey, USA). We measured stable isotopes of sulfur from subsurface soils (0-30 cm). The ratios of heavy to light stable isotopes are expressed as to indicate relative depletion (-) or the enrichment (+) of the heavy isotope compared to the lighter isotope relative to a standard, according to the formula: δX (‰) = ([R sample / R standard] - 1) 103 where X is 34S and R is 34S:32S. Results were presented as deviations from a standard (Canyon Diablo triolites for S). Repeatability was δ34S ± 0.3‰.
Surface and Subsurface Organic Matter Breakdown Rates
On 23 June 2017, we deployed rooibos (red) and green tea standard substrate litter at 15-cm depth of soil in marsh and mangrove patches at the front (fringe) and back corners (interior) of each of the 10 plots (see Keuskamp et al. 2013). Tea litterbags were retrieved on 15 June 2018 after 357 d of incubation. We calculated green and red tea breakdown rates (k) using the exponential decay equation Mt = M0*e-kt, whereby Mt is the final dry mass, M0 is the initial dry mass, and t is time in days (k, d-1), as well as per degree-day (k, dd-1).
We harvested S. alterniflora and A. germinans green leaves (hereafter litter) from the field on 07 November 2018 and allowed them to air dry for 24 h. Approximately 5 g of individual and mixed-species litter were placed into 1-mm nylon mesh bags (5 15 cm) that were divided in half. Individual species litter was placed in one half, and mixed-species litter was placed in the other half. On 09 November 2018, litterbags were deployed into replicate marsh and mangrove patches in fringe and interior zones in each of the 10 plots. Replicate patches (n = 8 per plot) were all marsh (in the 0% mangrove plot), all mangrove (in the 100% mangrove plot) or half marsh and half mangrove (in mixed plots) for a total of n = 80 replicates. A total of n = 160 litterbags (n = 80 individual species, n = 80 mixed-species) were deployed. Litterbags containing individual S. alterniflora litter and mixtures of S. alterniflora and A. germinans litter were deployed in marsh patches. Litterbags containing individual A. germinans litter and mixtures of A. germinans and S. alterniflora litter were deployed in mangrove patches. On 24 June 2019, litter samples were retrieved after 227 d incubation and returned to the lab on ice. We used the exponential decay equation (above) to quantify breakdown rates per day (k, d-1), as well as per degree-day (k, dd-1).
Microbial Respiration Rates
We measured microbial respiration rates (R) on decomposing S. alterniflora and A. germinans litter retrieved from individual and mixed-species litterbags after field incubation. Subsamples of field incubated litter were placed in glass vials (40 mL) in the laboratory. Vials were completely filled with filtered site seawater to remove any gas head space, sealed with caps, and placed in the dark for 1 h. Dissolved oxygen concentrations were measured at the beginning and end of lab incubations using YSI ProODO Dissolved Oxygen meters (Yellow Springs, Ohio, USA). Additional vials (n = 2) containing only site water served as controls. Oxygen consumption rates were determined as the slope of the regression of dissolved oxygen concentration over time minus the slope of the control, and respiration rates were expressed per gram dry mass of litter per hour.
