Changes in plant communities of low-salinity tidal marshes in response to sea-level riselow-salinity tidal marshes in response to sea-level rise

As sea-level rises, low-salinity tidal marshes experience greater flooding with more saline water. In the Chesapeake Bay estuary, we compared the 1980 and 2014 tidal marsh inventories (TMIs) of plant communities from James City County, Virginia, USA, with respect to the spatial distribution of two species—the invasive reed Phragmites australis and native salt marsh grass Spartina alterniflora–plus overall species richness. Since the 1980 TMI, the total area of low-salinity tidal marshes in which P. australis occurred increased from 0.46 km to 6.30 km in 2014. Between TMIs, however, the total area of lowsalinity marshes occupied by S. alterniflora increased by only 0.02 km. Species richness in low-salinity tidal marshes decreased from 41 to 25 between TMIs. To assess seedling emergence under increased flooding and salinity, we completed two seed bank germination experiments using soil samples collected from six low-salinity marshes containing established P. australis stands. In the first experiment, more seedlings emerged in the two low-salinity (0 vs. 5 ppt) treatments after seven weeks, irrespective of flooding (water 3.75 cm below vs. at soil surface), but no P. australis or S. alterniflora germinated. For the second experiment, we added seeds of P. australis and S. alterniflora to soils exposed to the same flooding and salinity treatments to test the impact of these plant competitors on seedling emergence. No difference in number of seedlings was detected among treatments, but the number of species and their relative abundance was significantly affected by flooding (ANOSIM, R = 0.138, P < 0.001). The presence of P. australis and S. alterniflora seedlings appeared to shift the physical factor more influential on seedling emergence from salinity to flooding. For both seed bank experiments, more seedlings but not more species emerged from soils collected from marshes where P. australis coverage was <50%. High diversity plant communities of low-salinity tidal marshes along the upper reaches of this estuary are gradually being replaced by those dominated by P. australis and S. alterniflora—a trend expected to continue here and in other riverine estuaries of the Atlantic and Gulf Coasts.


INTRODUCTION
Low-salinity tidal marshes (0-5 ppt) are extensive along the upper reaches of large-river estuaries of many parts of the world, particularly along the Atlantic and Gulf Coasts of the United States (Odum et al. 1984). While specific plant composition may vary, general community types are similar among low-salinity tidal marshes ). Unlike higher salinity, Spartina alterniflora dominated marshes located in mesohaline to polyhaline reaches, lowsalinity marshes are characterized by a diverse plant community of scores of species living in freshwater to oligohaline tidal water (Odum 1988, Elsey-Quirk andLeck 2015). In addition to botanical diversity, low-salinity tidal marshes sequester carbon, are proficient at nutrient and sediment assimilation, and provide valuable nurseries for many fish species (Odum et al. 1984, Whigham et al. 2019. Vegetated, low-salinity tidal wetlands face numerous environmental challenges that largely are associated with sea-level rise (Noe et al. 2013, Beckett et al. 2016, Palinkas and Engelhardt 2016. With sea-level rise, the influx of more water increases wetland hydroperiod, and wetlands that cannot accrete sediment to match rapidly rising waters may drown (Kirwan andMegonigal 2013, Weston 2014). Alternately, wetlands may migrate farther upstream or into adjacent undeveloped uplands if the elevation is gradual, but those opportunities may be limited adjacent to forests (Field et al. 2016) and along incised coastal plain river systems (Torio andChmura 2013, Mitchell et al. 2020). The encroaching tidal water also brings additional salt, which is a stress to many of the plant species living in low-salinity tidal marshes (Spalding andHester 2007, Neubauer 2013). The salty water may carry propagules of salt marsh species like S. alterniflora, thereby introducing both a physical stress (salt) and a potential plant competitor. Finally, many wetlands in the United States are experiencing expansion of non-native Phragmites australis-a competitive dominant species capable of rapid establishment and spread in tidal wetland environments experiencing saltwater intrusion (Chambers et al. 1999, Meyerson et al. 2000, Saltonstall 2002, Vasquez et al. 2005. Collectively, the plant diversity of low-salinity tidal wetlands is expected to change most dramatically with shifts in gradients of hydrology, salinity, and species distributions associated with rapid sea-level rise Atkinson 2009, Sharpe andBaldwin 2012).
Seed banks play a critical role in the make-up of plant communities, as seeds persist longer under stress than the standing vegetation (Wang et al. 2008). Recruitment from the wetland seed bank is a primary determinant of plant community composition in low-salinity tidal marshes exposed to sea-level rise (DeBerry and Perry 2000, Sharpe andBaldwin 2012). Both seedling emergence and density are reduced with elevated salinity and prolonged inundation of tidal marsh soils (Baldwin et al. 1996, Baldwin et al. 2001, Peterson and Baldwin 2004. Of course, species first must reach the soil seed bank to germinate, and the pathways to arrival in lowsalinity tidal marshes can vary. For example, water-borne S. alterniflora seeds may be introduced via saltwater intrusion (hydrochory; Elsey-Quirk et al. 2009), whereas P. australis seeds may be carried in by water or by the wind (anemochory; Soomers et al. 2013). The physical vectors of wind and water create opportunities for these species to spread into new wetland environments.
Given the ongoing pressures from sea-level rise, salt water intrusion, and invasive species (Whigham et al. 2019), we completed a study to assess potential changes in plant community structure associated with these pressures in selected, low-salinity tidal marshes of two largeriver sub-estuaries of the Chesapeake Bay in southeastern Virginia (VA). Environmental pressures are exacerbated here, as the rate of relative sea-level rise in the Chesapeake Bay (recently measured at 4-6 mm/yr; Boon andMitchell 2015, Ezer andAtkinson, 2015) is greater than the global average rate (~3.2 mm/yr; Church andWhite 2011, Ezer 2013). Models of sea-level rise in the region (Hilton et al. 2008, Rice et al. 2012) and qualitative observations (Sutter et al. 2015) suggest ongoing salinity intrusion. We used GIS to compare plant inventories in the 1970s and the 2010s, specifically noting changes in the spatial distribution of S. alterniflora and P. australis and overall species richness. Additionally, we used low-salinity marsh soils to conduct two seed bank germination experiments in a greenhouse setting. The first experiment examined seed germination as affected by the anticipated physical impacts of sea-level rise, that is, higher water table and higher salinity. The second experiment examined whether seed germination patterns would change in the presence of S. alterniflora and P. australis as novel plant competitors in lowsalinity marshes. Collectively, we hypothesized that both S. alterniflora and P. australis coverage have increased with sea-level rise and that v www.esajournals.org 2 July 2021 v Volume 12(7) v Article e03630 associated physical factors (increased flooding and salinity) and biological factors (increased novel competitors) negatively affect plant community structure.

Historical changes in plant distribution
This study was conducted in low-salinity tidal marshes in James City County, VA (465 km 2 ), bounded to the northeast by the York River and to the southwest by the James River subestuaries of Chesapeake Bay (Fig. 1). Tidal marsh inventories (TMIs) for the county were completed first between 1974and 1980(1980CCRM 1992) and then more recently between 2010(2014CCRM 2014, Mitchell et al. 2017. For both TMIs, we excluded marshes listed as dominated by saltmarsh cordgrass (S. alterniflora), saltmeadow hay (S. patens), black needlerush (Juncus roemerianus), and brackish/mixed communities, assuming the salinity of these marshes was mesohaline. With this operational definition, we designated all other marsh communities identified with freshwater species present as low-salinity tidal marshes, including Arrow Arum-Pickerelweed (Peltandra virginica-Pontederia cordata), Cattail (Typha spp.), Yellow Pond Lilly (Nuphar spp.), Big Cordgrass (S. cynouroides), Reedgrass (P. australis), and freshwater/mixed communities. ArcGIS 10 was used to determine changes in the areal extent and  Table 2 for the area of coverage in each cover class. Arrows point to smaller marshes in which P. australis was observed. v www.esajournals.org 3 July 2021 v Volume 12(7) v Article e03630 distribution of low-salinity marshes in which the invasive P. australis and native S. alterniflora occurred. The occurrence data for the invasive P. australis and native S. alterniflora were collected as cover classes (trace <1%, present 1-50% or dominant >50%); therefore, we reported the total area of low-salinity tidal marshes in which each cover class for P. australis occurred and for marshes in which the trace and presence cover classes for S. alterniflora occurred (by our definition, S. alterniflora-dominant marshes were not considered low-salinity marshes). For the 1980 TMI, species occurrence data were extracted from a hard copy report (Moore and Silberhorn 1980) and added to the GIS shapefile depicting historic marsh extent (CCRM 1992). The historical distribution of P. australis and S. alterniflora was then mapped by the cover class identified in the 1980 TMI. Similar maps were created for the 2014 TMI to determine changes in relative area of low-salinity marshes in which P. australis and S. alterniflora occurred.

Physical impacts of sea-level rise on seed germination
Soil samples were collected from six lowsalinity tidal marsh sites in James City County. Based on the 2014 TMI, three sites were randomly selected from all low-salinity marshes in JCC with <50% P. australis coverage (present) and three sites were randomly selected with >50% P. australis coverage (dominant; Table 1). From each site, the top 2-5 cm of marsh soil, that is, the recent seed bank, was collected immediately adjacent to a P. australis stand during early March 2016, prior to seed bank germination. Before experimental setup, the soil samples were stored in the dark at 3°C to inhibit premature germination and maintain seed viability. For the experiment, individual 15-cm plastic nursery pots were filled with soil from the field samples and placed inside larger, 27 cm diameter tanks. We employed a 2 9 2 factorial design, with two flooding and two salinity treatments in triplicate for each of the six marsh soils: (1) no salt, no flood (NSNF) treatment: control pots with freshwater to within 3.75 cm of the soil surface; (2) no salt, flood (NSF) treatment: pots with freshwater to the soil surface; (3) salt, no flood (SNF) treatment: pots with brackish water to within 3.75 cm of the soil surface; (4) salt, flood (SF) treatment: pots with brackish water to the soil surface. Seed banks in the no flood treatments were unsaturated at the soil surface, whereas seed banks in the flooded treatments were fully inundated, with the water table at the soil surface. Brackish water for salt treatments (5 ppt salinity-a realistic elevation of salinity during spring when germination occurs) was made from Instant Ocean; freshwater was used for no-salt treatments. Salinity levels were monitored using a YSI handheld salinity meter, and freshwater was added to individual treatment tanks approximately weekly to replace water lost to evapotranspiration. Artificial lighting was set on 12-h periods in the greenhouse, and temperatures were maintained at 23°-24°C throughout the experiment. Each week for seven weeks during spring 2016, sprouted seeds in each pot were identified, counted, and monitored as part of the data collection protocol. For some seedlings, growth had to be extended for longer than seven weeks to allow for taxonomic identification.

Physical and biological impacts of sea-level rise on seed germination
For the second germination experiment, the same setup for testing physical impacts was used. The experimental pots were filled with soil from the same marsh sites, and the same 2 9 2 factorial design was employed, with four replicates per soil and salinity and flooding treatment group. To assess biological impacts, we scattered 20 of the larger S. alterniflora seeds and a sprinkling of approximately 50 of the smaller P. australis seeds across the surface of each pot of soil. The seeds of both species had been cold-stratified at 3°C prior to use but, neither seed fullness nor viability was tested before the experiment. After setup, the germination of S. alterniflora, P. australis, and seed bank species was monitored weekly for seven weeks in each of the 96 individual tanks (four replicates of six soils, with two salinity and two flooding treatments).

Data analysis
For both seed germination experiments, the cumulative number of seedlings present within each of the four treatment groups was plotted over seven weeks. Analysis of variance and ttests compared the total number of seedlings present at week 7 by treatment group (salinity and flooding) and by soil source (marshes with <50% or >50% P. australis coverage). Analysis of variance compared the number of P. australis and S. alterniflora seedlings in each treatment group. The coefficient of community (Sorenson 1948) was used for qualitative comparisons of species occurrence between soil and flooding treatment pairs. Finally, an analysis of similarity (ANOSIM) determined the similarity of the seedling communities from each treatment group. Because soils from the York River site yielded no germination from any of the 12 pots in the first germination experiment, those results were not included in the ANOSIM. An additional 10 pots across all treatment types, however, also had no germination. To maintain a balanced sample design and reduce the analytical impact of pots with zero germination, 10 pots were randomly selected for removal from statistical analysis. This adjustment, which yielded the same number of samples in each treatment group, was repeated 10 times in ANOSIM for the first germination experiment. For both germination experiments, the final week's data distribution was normalized using a square root transformation; then, a similarity matrix was created in PRIMER 6.1. ANOSIM was then run on that similarity matrix, testing similarity among plant communities produced under each treatment. A two-way crossed ANOSIM with replicates was used with treatments (N = 4) and soils (N = 6) as factors.

Historical changes in plant distribution
From the 1980 TMI, P. australis was found in trace amounts (<1% coverage) in just a few lowsalinity tidal marshes in James City County (0.46 km 2 ). Since then, P. australis occurrence has increased dramatically throughout the county (Fig. 1). P. australis now is more widespread along the York River and lower portion of the county along the James River (dominant in 0.07 km 2 , present in 3.27 km 2 , trace in 2.96 km 2 ). Between TMIs, the total area of marshes in which P. australis occurred increased from 0.46 km 2 to 6.30 km 2 (Table 2).
Spartina alterniflora already was found in several low-salinity tidal marshes in the 1980 TMI ( Fig. 2; present in 1.93 km 2 and found in trace amounts in 0.16 km 2 ). In the 2014 TMI, marsh areas with S. alterniflora present had increased slightly to 2.10 km 2 and those with trace amounts had decreased to 0.01 km 2 . Additionally, the occurrence of S. alterniflora in lowsalinity tidal marshes along the James River moved farther upstream between 1980 and 2014 (Fig. 2). The total area of low-salinity tidal marshes in which S. alterniflora occurred, however, was only 0.02 km 2 higher in the 2014 TMI Note: Total areas of low-salinity and mesohaline salinity marshes are also shown. v www.esajournals.org 5 July 2021 v Volume 12(7) v Article e03630 relative to 1980 ( Table 2). The total area of lowsalinity tidal marsh in James City County in the 1980 TMI was 26.61 km 2 ; in the 2014 TMI, the area of marsh was 16.95 km 2 (Table 2). A total of 41 plant species were recorded from the 1980 TMI, including shrubs, forbs, grasses, rushes, sedges, and trees, with 25 recorded from the 2014 TMI (Table 3). Twenty-three species were common to both TMIs, 18 present in 1980 were absent in 2014, and two new species were listed in 2014.

Physical impacts of sea-level rise on seed germination
Germinating plants were observed in 45 of the 72 pots in the first germination experiment testing the effects of flooding and salinity. Over the seven weeks of the experiment, a total of 251 plants comprising 21 species germinated, of which 242 were alive at week 7, yielding an average of 3.4 AE 0.7 SE seedlings per pot, and 12-16 species per treatment group. Eleocharis acicularis was numerically dominant (30.5% of all seedlings) and germinated in 20 of the pots. Zizania aquatica was found in 15 pots, and Persicaria punctata was found in 10 pots. The rarest species were Sagittaria subulata, Rumex verticillatus, Eupatorium perfoliatum, Hypericum mutilum, Chenopodium album, and Amaranthus cannabinus, occurring separately in just one pot each. Five species were found only in the saline treatments (Amaranthus cannabinus, Hypericum mutilum, Bidens laevis, Rumex verticillatus, Sagittaria subulata), and five species were found only in fresh  Table 2 for the area of coverage in each cover class. Arrows point to smaller marsh areas in which S. alterniflora was recorded.
Each week over the seven weeks of observation, the cumulative number of seedlings present was consistently higher for the two freshwater treatments than for the saltwater treatments (Fig. 3). By week 7, significantly more seedlings germinated in the freshwater treatments (NSNF + NSF), relative to the saltwater treatments (SNF + SF; ANOVA F (1,14) = 5.29, P = 0.03). The cumulative number of species present in each treatment group was similar (16, 14, 12, and 13 for NSNF, NSF, SNF, and SF, respectively). More seedlings were produced from soil sites with <50% P. australis coverage (13.6 AE 3.9 SE), relative to sites with >50% P. australis coverage (6.6 AE 1.9 SE; ANOVA F (1,16) = 6.32, P = 0.013; Fig. 4). The treatment 9 site interaction, however, was not significant, indicating that the observed patterns associated with salinity, flooding, and P. australis coverage were consistent among sample locations. Across the four treatment groups, the total number of species that emerged from sites with <50% P. australis coverage was significantly lower than from sites with >50% coverage (7.5 AE 1.5 SE vs. 9.5 AE 0.4 SE; t-test, P = 0.02).
The number of common species between flooding and salinity treatment groups ranged from four to nine, and total species observed ranged from 25 to 30 (Table 4). The treatments with the most common species and highest coefficient of community (CoC) were NSF and SNF (CoC = 0.62), followed by NSNF and NSF (CoC = 0.60). The SNF and SF treatments shared the fewest common species and had the lowest community coefficient (CoC = 0.32). Overall, no effect of salinity, flooding, or salinity 9 flooding was Note: Owing to lack of species-level information for some genera from the TMIs, species within the same genus are combined. Functional groups are abbreviated B, shrub; F, forb; G, grass; R, rush; S, sedge; T, tree.
v www.esajournals.org 7 July 2021 v Volume 12(7) v Article e03630 found on the number of species and their relative abundance in week 7 of the germination experiment (ANOSIM, P > 0.05).

Physical and biological impacts of sea-level rise on seed germination
Germinating plants were observed in all 96 pots in the second germination experiment, with 1972 germinating plants present at the end of the experiment, representing 16 species (Table 3). Of these seedlings, 513 were S. alterniflora and 1072 were P. australis. By week 7, P. australis was found in 78 out of 96 pots and had germinated at some point in all but five pots (average number of seedlings = 11.8 AE 9.5). Similarly, S. alterniflora was found in 79 out of 96 pots and had germinated at some point in all pots (average number of seedlings = 5.3 AE 4.9). The numbers of germinated P. australis and S. alterniflora seeds were not significantly different among flooding or salinity treatments (P. australis F (3,87) = 0.94, P = 0.43; S. alterniflora F (3,92) = 0.32, P = 0.81).
A total of 387 seedlings from other species were present after week 7, yielding an average of 4.0 AE 0.4 SE seedlings per pot. Eleocharis acicularis was overwhelmingly the numerical dominant (79.8% of all seedlings). Relative to the first germination experiment, fewer species v www.esajournals.org 8 July 2021 v Volume 12(7) v Article e03630 germinated in these soils seeded with P. australis and S. alterniflora (Table 3), with 6-12 additional species observed per treatment group. Persicaria punctata, Schoenoplectus robustus, Echinochloa walteri, and Juncus acuminatus were found in all the treatments. The forbs Peltandra virginica and Pluchea odorata were only found in the non-flooded brackish treatment. Amaranthus cannabinus was only found in the non-flooded freshwater treatment. Acer rubrum, Sagittaria subulata, and Carex albolutescens were only found in the flooded freshwater treatment. Over the first six weeks of observation, the cumulative number of these additional seedlings was higher for the two non-flooded treatments (NSNF + SNF) relative to the flooded treatments (NSF + SF; Fig. 3). The ANOVA comparing the average number of seedlings other than S. alterniflora and P. australis at week 7, however, was not significant between non-flooded and flooded treatments (ANOVA F (1,14) = 0.004, P = 0.948), but more species germinated from non-flooded vs. flooded treatments (12, 11, 6, and 7 for NSNF, SNF, NSF, and SF, respectively). Significantly more seedlings were produced among soil sites with low P. australis coverage (21 AE 4.7 SE), relative to sites with high P. australis coverage (3.2 AE 0.9 SE; ANOVA F (1,24) = 11.7, P = 0.002; Fig. 4). The treatment 9 site interaction, however, was not significant. Across treatments, the total number of species that emerged from sites with <50% P. australis coverage was not significantly different from sites with >50% coverage (4.0 AE 1.0 SE vs 5.8 AE 4.8 SE; t-test, P > 0.05).
With the addition of P. australis and S. alterniflora seed germination, the number of common species germinating between soil and flooding treatment groups ranged from three to six, and total species observed ranged from 13 to 19 (Table 5). The treatments with the highest coefficient of community were NSNF and NSF (CoC = 0.67). The SF and NSF treatments shared the fewest common species and lowest community coefficient (CC = 0.44). Among treatments, the community differences were significant with respect to flooding (ANOSIM, R = 0.138, P < 0.001) but not with respect to salinity or the flooding 9 salinity interaction.
Eleven species were common to both germination experiments, and six species were common to both tidal marsh inventories and both germination experiments, including five perennial species (Amaranthus cannabinus, Juncus spp., Leersia oryzoides, Peltandra virginica, and Sagittaria spp.) and one annual species (Persicaria punctate; Table 3).

DISCUSSION
Results comparing tidal marsh inventories in James City County from 1980 and 2014 are consistent with the ongoing invasion and spread of non-native P. australis into tidal wetlands of North America ( Fig. 1; Chambers et al. 1999). Mechanistically, disturbance caused by increased salinity and inundation as stressors to freshwater species (Hazelton et al. 2014) may contribute to the expansion of P. australis into low-salinity marshes. The observed decrease in species richness between inventories from 41 to 25 (Table 3) could be due to fewer species adapted to increased salinity and inundation associated with sea-level rise in these low-salinity marshes. P. australis is tolerant of both salinity and  flooding (Chambers et al. 2003), and results from our second germination experiment showed no difference in P. australis seedling germination with respect to salinity or flooding. P. australis therefore may be able to exploit windows of opportunity for expansion and spread created by the die-off of other less tolerant species (Sutter et al. 2015). Alternately, clonal growth of P. australis in dense stands may physically inhibit the growth of other species via shading by shoots, by litter, or by direct displacement (Meyerson et al. 2000, Minchintin et al. 2006, Holdredge and Bertness 2011. From both germination experiments, we observed significantly fewer seedlings of other species from marsh soils where P. australis was dominant (Fig. 4). More species germinated from those P. australis-dominant soils in the first experiment, but fewer total seedlings were observed in both experiments, perhaps due to smaller seed banks. Lowsalinity marshes in James City County are transitioning to increased dominance by P. australis, for which its invasive ability may be exacerbated by both ongoing recruitment of its seeds to the seed bank (Baldwin et al. 2010) and the inhibitory effects of increased salinity and flooding from sea-level rise on seed germination by native species.
From the 1980 and 2014 TMIs, we did not detect a large difference in the areas of lowsalinity marshes where S. alterniflora occurred. On the James River, S. alterniflora appeared in marshes farther upstream in the 2014 TMI, consistent with the ongoing penetration with sealevel rise of both saltwater and S. alterniflora propagules into low-salinity marshes. S. alterniflora, however, did not always persist in marshes where it was found in 1980. Our results are similar to another recent study showing that S. alterniflora is becoming more prominent in some but not all freshwater tidal marshes regionally (Sutter et al. 2013). The ability of S. alterniflora to establish and expand into low-salinity tidal marshes lends little support for the competition stress theory which would predict competitive exclusion of S. alterniflora by a more diverse assemblage of plants (Sutter et al. 2013). Our second germination experiment indicated that, similar to P. australis, seed germination of S. alterniflora was not inhibited by salt, by flooding, or by the presence of seedlings of other species. Sporadic tidal flows of salt water during summer periods of low freshwater runoff may initially increase the salinity level in these low-salinity marshes and convey S. alterniflora propagules at the same time. With seed delivery, opportunities exist for germination and establishment of S. alterniflora at low marsh elevations, with ample space for S. alterniflora to thrive under increased inundation levels. We doubt that P. australis as a high marsh species is excluding establishment of the low marsh S. alterniflora (but see Medeiros et al. 2013).
The established presence of S. alterniflora in many York River and lower James River marshes from the 1980 TMI (Figs. 2 and 3) and its seed distribution via hydrochory would indicate that propagules had previously been carried into these marshes. Some low-salinity marshes experience cyclical, temporal shifts in abundance of S. alterniflora versus freshwater perennial species owing to interannual variation in freshwater river flows and/or local groundwater discharge (Davies 2004). We found that the distribution of S. alterniflora changed between 1980 and 2014 but did not increase in total marsh area, suggesting the effects of short-term variation may still be a larger influence on S. alterniflora occurrence than longer-term forcing by sea-level rise ). Because our second germination experiment showed that S. alterniflora seedling emergence was not impacted by salinity or flooding, we suspect that emergence of perennial freshwater species (i.e., not from the seed bank but from established root/rhizome stock) may inhibit emergence of perennial S. alterniflora during fresher conditions, but that salt may inhibit freshwater species during more salty conditions. Of six native species common to both TMIs and to both germination experiments (Table 3), five are perennial. These species may be more important than annuals in resisting the spread of S. alterniflora into low-salinity marshes, as the germination of annuals is much less consistent in space and time (Hopfensperger et al. 2009).
From the first germination experiment, the physical effect of salinity reduced the number of seedlings emerging from the seed bank more than flooding, with significantly more seeds, but not species, germinating in the two no-salt treatments relative to the two salt treatments (Fig. 3). Further, the coefficient of community was low for the two comparisons for which flooding was v www.esajournals.org held constant (NSNF vs. SNF = 0.43; NSF vs. SF = 0.41; Table 4). This result of a salinity effect greater than flooding has been observed in prior studies of seed germination in low-salinity tidal wetlands (Sharpe andBaldwin 2012, S anchez-Garc ıa et al. 2017). For germination studies that tested flooding separately from salinity, however, flooding was also shown to decrease seedling emergence (Peterson and Baldwin 2004, Delgado et al. 2018, Sloey and Hester 2019. Indeed, our measured coefficient of community was lowest between the SNF and SF treatments (0.32; Table 4), so both flooding and salinity are physical factors affecting seed germination and plant success (Middleton 2016).
When both S. alterniflora and P. australis seeds were added to the seed bank, a different effect was observed with respect to emergence of other species. First, fewer seedlings of other species emerged from flooded treatments over the first six weeks of observation (Fig. 3). During week 7, the effect was altered by the emergence of large numbers of Eleocharis acicularis seedlings in both flooded treatments, similar to germination results obtained by Baldwin et al. (1996). Fewer species were observed in the two flooded treatments relative to the two non-flooded treatments, and ANOSIM analysis of seedling species and relative abundance found a significant effect of flooding. The coefficients of community, however, were difficult to interpret, as the most similar treatments were NSNF and NSF (0.67; Table 5), which we thought would have the fewest common species. Also, the average number of other seedlings per pot was not significantly different in the presence or absence of P. australis and S. alterniflora seedlings (4.0 vs. 3.3 seedlings, respectively). Schile et al. (2017) suggested that the combined effects of physical stress and competitive stress can create unexpected synergies regarding plant performance. We documented small differences in seed bank germination with the presence of S. alterniflora and P. australis seedlings (Fig. 3, Table 5), but other factors (salinity, flooding, dominance of P. australis in the source marsh) appeared to be more important. The mechanisms behind the observed shift in relative impact of flooding and salinity on the germination and success of other species are unknown. Belowground competition for water and nutrients by S. alterniflora and P. australis roots, for example, could reduce tolerance of other species to flooding. Biotic interactions appear to alter how physical stresses are manifest in plant performance (Schile et al. 2017).
Over time in these low-salinity marshes, salinity, water level, and the number of introductions of S. alterniflora and P. australis are expected to rise. We expect P. australis will continue to expand clonally and with seed dispersion via anemochory and perhaps also hydrochory (Baldwin et al. 2010), as seeds in our second experiment germinated under all experimental salinity and flooding conditions. Although prior research suggests that S. alterniflora coverage in these lowsalinity marshes appears to expand and contract in the short term among sampling years (Perry and Hershner 1999, Davies 2004), conversion to mesohaline plant communities dominated by S. alterniflora is expected to occur eventually (Fig. 2). Plant communities in these low-salinity marshes are expected to become less diverse with sea-level rise, with both P. australis and S. alterniflora becoming more extensive in total area. The short-term persistence of salt-tolerant native species, including Eleocharis acicularis, Persicaria punctata, Juncus spp., Leersia oryzoides, Schoenoplectus americanus, and Zizania aquatica, is also expected.
To preserve diversity in low-salinity tidal marshes along the east coast of the United States, management actions should continue to focus on P. australis control. With sea-level rise, low marsh areas exposed to greater inundation, salinity, and S. alterniflora propagules will be difficult to manage. As a high marsh species, however, P. australis is in competition with native plant species that occur at or above mean high water and can maintain their populations above thresholds of increasing inundation (Delgado et al. 2018). Also, the diversity of seed banks in marshes invaded by P. australis remains high (Baldwin et al. 2010; this study), so control of the invader could promote the diversity of native species. Relatively effective methods of P. australis management include targeting small stands with herbicides and introducing herbivory (Hazelton et al. 2014). Also, marshes may be more or less susceptible to P. australis invasion owing to their relative exposure to wind and water that deliver propagules and increase genetic diversity and expansion v www.esajournals.org 11 July 2021 v Volume 12(7) v Article e03630 rates of invasive stands (Baldwin et al. 2010). Because P. australis management can be costly to maintain, simultaneous restoration of disturbed tidal marshes, maintenance of existing tidal marshes in the face of development, and creation of new tidal marshes where possible may aid in preserving the diversity of low-salinity tidal marshes and their ecosystem functions under the increasing threat of sea-level rise.

CONCLUSION
The physical and biological manifestations of sea-level rise negatively influence plant communities in low-salinity tidal marshes. Physically, the germination and establishment of other species are reduced by increased salinity and flooding regimes. Biologically, the introduction of P. australis and S. alterniflora reduces the number of other species and, additionally, shifts the relative importance of salinity and inundation on seedling germination and establishment. Further, soils from tidal wetlands dominated by P. australis germinate fewer total seedlings relative to soils from wetlands where P. australis is not dominant. Without management, low-salinity tidal marshes along the upper reaches of estuarine river systems from the Atlantic and Gulf coasts of the United States will have to migrate upstream in response to sea-level rise as their prior, high diversity plant communities are replaced by those dominated by P. australis and S. alterniflora.