The study area is located in Liao River Delta, northeastern China (121°10′-122°30′E, 40°30′-41°30′N), where sediments are deposited by Liao River. The Liao River Delta is adjacent to Bohai Sea with the wetland area of over 1200 km². The stand of Phragmites australis in Liao River Delta covers over 800 km² and is the world’s largest reed wetland (Brix et al., 2014; Wang et al., 2016). The famous Suaeda heteroptera salt marsh referred to as “red beach”, is another main vegetation type in the delta. The wetlands are very important breeding ground for migration waterfowls including the red-crowned crane (Ciconia boyciana) and the white-naped crane (Grus vipio). The wetlands are designated as the Liao Estuary National Natural Reserve since 1988 and became a Ramsar site with international importance in 2005 (Tian et al., 2017). Mean annual precipitation in the study area is 612 mm, and mean annual temperature is 8.3°C (Wang et al., 2017).

Seven sites were selected along the Liao River on September 15-17, 2018 ( Figure 1 ). These sites were all directly influenced by both the river and wave activity. At each site, we sampled three wetland types: bare, unvegetated sediments (‘bare’), Phragmites australis marsh (‘Phrag’) and Suaeda heteroptera marsh (‘Suaeda’) which were present next to each other at each site, within 200 m. We assumed that the original soils, sediment inputs, tidal fluctuation and regional climate of the three vegetation types within each site were more similar to each other than between sites, and the differences were due to changes in vegetation. Three replicate samples with 10 cm soil depth were collected in each vegetation site. Each sample was over 20 m apart from each other. Longitude and latitude were recorded GPS. Soil samples were stored in a cooler in 4°C.

Samples were first dried to a constant weight at 60°C. Samples were then homogenized, and sieved with a mesh screen of 2-mm. We determined (1) soil element concentrations, (2) soil organic matter content, (3) soil salinity, (4) soil pH, (5) water content, and (6) electrical conductivity. Element concentrations were measured by ICP-MS analysis (NexION 350D, PerkinElmer, US). Total nitrogen was determined by Kjeldahl method using a SKALAR SAN++ continuous flow analysis (Skalar, Netherlands). Organic matter content was measured by dichromate oxidation and titration with ferrous ammonium sulfate. Soil salinity, electrical conductivity and pH were measured with a water-soil ratio of 5:1 by using a glass electrode. Soil water content was measured as a ratio of water weight to the weight of dried soil.

The number of elements in this study were determined by which elements can be detected by the ICP-MS equipment. Mean values of concentration of 43 elements and five environmental factors (soil organic matter content, salinity, pH, water content, electrical conductivity) were calculated for bare, Phrag, and Suaeda. One-way ANOVA and a subsequent Tukey’s test were used to analyse the concentrations of 43 soil elements and the five environmental factors between wetland types. Significance was set at a level of 0.05. Data were log-transformed to meet normality and homogeneity assumptions of ANOVA. Spearman correlation analysis was performed to study the relationships of soil element concentration with organic matter content. The analysis was carried out in SPSS 17.0. Redundancy analysis (RDA) was conducted to study the relationship between element concentrations (43 elements) and five environmental variables (organic matter content, salinity, pH, water content, and electrical conductivity) using CANOCO 5.0. Environmental factors in the final model were selected by the forward selection procedures with 499 permutations. Non-metric multidimensional scaling (NMDS) in CANOCO 5.0 was performed to study the similarities of element composition among the three wetland types (bare, Phrag, and Suaeda).

The values of environmental variables and element concentrations are presented in Table 1 . Soil organic matter content (OM) in the Phragmites marshes and Suaeda marshes was significantly higher than in bare, unvegetated sites. The electrical conductivity (EC) and soil salinity (SS) in the Suaeda marshes was significantly higher than the Phragmites marshes and bare, unvegetated sites. The soil pH in the bare, unvegetated sites was significantly higher than the Suaeda marshes, while the soil water content (WC) in the bare, unvegetated sites was significantly lower than the Suaeda marshes. Even though the values for EC, pH and SS differed significantly, the ranges in values were very small.

In total, 43 elements had detectable concentrations in the soils ( Table 1 ). Concentrations of Hg, K, S, and Se did not significantly differ between types of vegetation. The concentrations of 15 soil elements increased from bare, unvegetated sites to Phragmites marshes to Suaeda marshes. The concentrations of 14 soil elements in the Suaeda marshes were higher than the bare, unvegetated sites and Phragmites marshes. The concentrations of 6 soil elements in the Suaeda marshes and Phragmites marshes were higher than the bare, unvegetated sites. The concentration of Ba decreased from bare, unvegetated sites to Phragmites to Suaeda marshes. The concentrations of Na and Sr in the bare, unvegetated sites were higher than the Phragmites and Suaeda marshes. The concentration of Ca was lower in Phragmites marshes than the bare, unvegetated sites and Suaeda marshes.

Redundancy analysis (RDA) showed that organic matter content explained most variation in soil element concentrations (42.7%), followed by soil electrical conductivity (7.5%), pH (3.0%), and soil salinity (2.7%) ( Figure 2 ; Table 2 ).

Organic matter content was positively related with most of the elements (37 elements) including the macro-elements Ca, S, N, P, Mg and most metals ( Figure 2 ; Table 3 ). Organic matter content was negatively related with Ba, Sr and Na. Two examples were shown for N and Ba ( Figure 3 ).

The distribution of elements in soils was studied with NMDS analysis ( Figure 4 ). Almost all the bare, unvegetated sites were distributed in the right quadrant, whereas almost all the Suaeda marsh sites were distributed in the left quadrant ( Figure 4 ). The Phragmites marsh sites were distributed across the entire NMDS ordination graph ( Figure 4 ).

Organic matter content is found to be the most important factor influencing element mobility and concentrations in freshwater wetlands (Salomons and Förstner, 1984; Kissoon et al., 2015; Yuan et al., 2019; Wang et al., 2020). In our study, organic matter content correlated positively with concentrations of most elements including the nutrients of N, P and S ( Figure 2 ; Table 3 ), which is consistent with other studies in salt marshes (Jokic et al., 2003; Sun et al., 2013; Yu et al., 2014; Qu et al., 2018) and mangroves (Chen and Twilley, 1999; McKee et al., 2007). In addition, similar with our study, sediments in Suaeda salt marshes often show higher concentrations of metal elements (e.g., Pb, Cu, Cr, Zn, Mg, Ni, Ca, Al, V, Ca, and Ni) due to the enrichment of soil organic matter content and Fe-Mn oxides (Williams et al., 1994; Graham and Mendelssohn, 2014). Coastal wetlands had large amounts of Al and Fe, the oxides and phosphates bind to several elements (Salomons and Förstner, 1984; Sun et al., 2015). On the contrary, elements that bind with soil organic matter content in a weak affinity usually are in soils which contain much inorganic particles (Beesley et al., 2010). Concentrations of such elements in soils decreased with the increase of organic matter content (Wang et al., 2019). This explains the negative relationships between organic matter content and alkali metals Na, Ba and Sr ( Figure 2 ).

Even though OM is recognized as the most important factor influencing element concentrations in wetlands, there are striking differences between studies in coastal and inland wetlands, which is that in the coastal systems most metals correlate positively with OM (e.g. Otte et al., 1993; Wang et al., 2022), whereas in the inland systems it is a negative relationship (e.g. Yellick et al., 2016; Zhu et al., 2021). This can be understood if we consider the origin of the metals in those different systems. As Salomons and Förstner (1984) already described in detail, metals in estuaries originate from precipitation with sediments and organic matter. As a result, metal concentrations, OM and fraction of small sedimentary particles strongly correlate (Otte et al., 1993). However, metals in inland wetlands do not originate from sedimentation processes, but are derived from erosion of inorganic materials, in which the metals are occluded (Salomons and Förstner, 1984; Gao et al., 2018). In that case, as OM increases, the proportion of inorganic matter per volume of soil decreases, as do the concentrations of metals associated with that inorganic fraction (Werkmeister et al., 2018; Zhu et al., 2021).

Although pH, EC and SS were identified as statistically significant factors in explaining variation in element concentrations, those relationships likely have no real relevance in explaining spatial variation of elements. This is because although there are numerical differences, the ranges in values are very small. This is evident if the correlation between Na, EC and SS is more closely examined. Na is usually an important element in determining salinity, and usually concentrations of Na increase with increasing salinity and electrical conductivity. That is particularly evident if such gradients are observed along the entire length of estuaries, from freshwater to full seawater salinity (Salomons and Förstner, 1984; Du Laing et al., 2009; Cloern et al., 2017). However, in this study, even though the geographic range of the sampling sites was about 10 km, values for pH, EC and SS varied very little. As a result, Na in our study correlated slightly negatively with salinity, which was an artefact of the narrow range of values. Therefore, although pH, EC and SS varied significantly between vegetation sites and were identified as significant environmental variables, the effects of those environmental variables on metals distributions were insignificant compared to the effects of OM, which varied strongly between types of vegetation.

Our study found relatively high organic matter and high concentrations of most elements in Suaeda salt marshes, similar with other studies in coastal salt marshes (Otte et al., 1993; Yu et al., 2011), most likely due to biological processes (Schmidt et al., 2011). Without vegetation (bare, unvegetated sites) we observed 0.88% organic matter, but this was 1.11-1.12% in the Phragmites and Suaeda marshes. Wetland plants make direct contribution to soil formation by organic matter input (Baustian et al., 2012), which, as discussed above, affects element distributions. Wetland plants can absorb nutrients and metals and release them as they die (Baldantoni et al., 2004). Plant growth, root development and microbial processes regulate carbon accumulation, metal mobility, and nutrient accumulation (Jacob and Otte, 2003; Pennings et al., 2005; Loomis and Craft, 2010; Yu et al., 2016).

So, plants produce OM, which explains the higher levels of OM compared to bare, unvegetated sites, but this has been known for a very long time. The more important outcome of this work, however, is that where no statistically significant differences were found in OM between Phragmites and Suaeda sites, many elements showed statistically different concentrations between sites dominated by Phragmites compared to Suaeda, namely: Ca, Co, Cr, Dy, Er, Eu, Gd, Ho, Mg, Mn, Ni, Pr, Tb, V, and Zn. This shows the power of multi-element analysis. It highlights that it is the plants that are the most important determinant of element distribution in estuarine sediments, through their production of organic matter, but also through the different ways in which different plant species interact with their immediate environment.