Effects of spatial expansion between Phragmites australis and Cyperus malaccensis on temporal variations and bioaccumulation of vanadium in coastal marshes of the Min River estuary, Southeast China

Vanadium (V) plays an important role in physio-ecological processes of marsh plants. The effects of spatial expansion between invasive species (Phragmites australis, PA) and native species (Cyperus malaccensis, CM) on temporal variations and bioaccumulation of V in coastal marshes of the Min River estuary were investigated by space-for-time substitution method. In situ filed sampling was conducted in PA marsh (PAM, before expansion), CM marsh (CMM, before expansion) and ecotonal marsh (EM, during expansion, marsh plants were denoted by PA' and CM') in different seasons. Results showed that, over all sampling seasons, the mean V contents in marsh soils ranged from 99.71 to 108.41 mg·kg−1 which exceeded its background value in soils of Fujian province (78.3 mg·kg−1). The V levels in soils differed among seasons or marshes and higher values in PAM, EM and CMM soils were generally observed in spring and winter. The temporal variation of V levels in EM soil rested with the alterations of pH, SOM and plant ecological traits during spatial expansion. The V contents in PA, PA', CM' and CM differed among tissues and higher bioaccumulation occurred in roots. The V levels in tissues differed among species or seasons, which could be ascribed to the differences in ecological traits among plants and the competitive absorption for V by plants during spatial expansion. This paper confirmed that the V in marsh soils of the Min River estuary existed enrichment process and the spatial expansion between PA and CM promoted its enrichment in soils and its bioaccumulation by plants. The findings of this study were favorable for understanding the biogeochemical behaviors of V in marsh ecosystem and strengthening the marsh conservation by regulating its bioavailability in soils.


Introduction
Vanadium (V) is vital beneficial element, which plays a very important role in physio-ecological processes of marsh plants (Zhu et al. 2016). Similar to molybdenum (Mo), V is indispensible beneficial element which not only can promote the biological fixation of nitrogen but also can enhance the absorption and utilization of iron (Fe) by plants and eventually influence the biosynthesis of chlorophyll (Wang and Wei 1995;Nawaz et al. 2018). However, excessive amount of V might inhibit the adenosine triphosphatase in cell membrane of roots, retard the growth of marsh plants and reduce the absorption of calcium and phosphate by plants (Jiao and Teng 2008). Marsh soil is the primary stock of beneficial elements and the bioaccumulation for V by plants not only can reflect its bioavailability in soils but also can indicate its biogeochemical behaviors in ecosystem (Li et al. 2020a).
Coastal marsh is one of the most sensitive ecosystems where the material-energy exchanges between fluvial ecosystem and marine ecosystem generally occur (Simas et al. 2001). As affected by the mixing of freshwater and seawater, the physico-chemical conditions in environmental medium (e.g., temperature, grain particle, salinity, pH and redox) are greatly and frequently changed (Moran et al. 1996). Coastal marsh is also affected by severe hydrodynamic forces, frequent erosion and deposition and high-intensity human activities. All these induce the biogeochemical processes of elements in coastal marsh are very complicated (Sun et al. 2017). In the past decade, there has been an ever-increasing interest in discussing the distribution and stock of biogenic elements (e.g., C, N, P and S) (Korol et al. 2016;Herbert et al. 2018;Wan et al. 2020;He et al. 2020) and heavy metals (e.g., Pb, Cu, Zn, Cr and Ni) (Chen & Ma 2017;Keshta et al. 2020) in plant-soil system of coastal marshes, while information on bioaccumulation of beneficial elements (e.g., V and Co) by plants in coastal marsh is poorly documented.
The Min River is one of the biggest rivers flowing into the East China Sea in the Fujian Province of southeast China. Shanyutan is the largest marsh in the Min River estuary (with an area of 893 ha), where Scirpus triqueter, Phragmites australis, Cyperus malaccensis and Cyperus compressus are the most common plants . Thereinto, C. malaccensis, S. triqueter and C.compressus are three native species, while P. australis is an invasive species. Local historical records have indicated that P. australis first colonized Shanyutan at about 30 ~ 40 years ago due to its dispersal from middle and upper reaches of the Min River . Thereafter, the marsh originally dominated by C. malaccensis is gradually expanded by P. australis, resulting in which becomes a single dominant community or even forms an ecotonal community with C. malaccensis (approximately 100 ~ 120 m-wide). In recent years, the expansion velocity of P. australis showed an increasing trend and the width of ecotonal community in the middle-west area of Shanyutan exceeded 150 m (Zhang et al. 2020). The spatial expansion between P. australis and C. malaccensis actually reflects the competitions of the two species for environmental resources such as light, water and nutrient (Li et al. 2020b). Existing studies have indicated that the spatial expansion of dominant species not only greatly altered the sedimentary environment and the physical or chemical conditions of marsh ecosystem (Ewanchuk and Bertness 2004), but also significantly influenced the ecological traits of plants and the key biogeochemical processes of elements by exerting strong effects on biotic and abiotic variables of marsh (Zhang et al. 2010;Vilà et al. 2011). As jointly affected by hydrodynamic forces, tide and community succession, the biotic and abiotic conditions in ecotonal marsh of the Min River estuary might be more complex (Li et al. 2020a), which directly or indirectly influences the distribution, stock and biogeochemical behavior of elements in plant-soil system. Although considerable efforts have been conducted in the Min River estuary to investigate the biogenic element levels in plant-soil systems of coastal marshes (Wang et al. 2018a;Chen et al. 2018;Zhang et al 2020), insufficient information is available concerning the bioaccumulation of beneficial elements by plants. Particularly, little is known about the temporal variations of V bioaccumulation by P. australis and C. malaccensis during their spatial expansion.
In this paper, three typical marshes (P. australis, C. malaccensis and ecotonal marshes) in the Min River estuary were studied and the influences of spatial expansion between P. australis and C. malaccensis on temporal variations of V bioaccumulation in coastal marshes were investigated by space-for-time substitution method. Understanding the distribution and transference of V in plant-soil system was of significance in exploring its biogeochemical behavior in marsh ecosystem. It was hypothesed that the V bioaccumulation in P. australis and C. malaccensis differed among seasons, which might be significantly affected by their spatial expansion. Objectives of this paper were: i) to explore the temporal variations of V levels in soils of different marshes; ii) to investigate the V bioaccumulation in different plants over all sampling seasons; and iii) to determine the key factors influencing the bioaccumulation and transference of V in marsh plants.

Study region
The Min River estuary lies in the transition region of mid-subtropical zone and south subtropical zone. The marshes in the Min River estuary distribute along the riverbank or estuary and start from Zhuqi in the west and end in Chuanshi Island in the east, with a total area of 980.6 km 2 (Liu et al. 2006). This study was conducted in intertidal zone of the Shanyutan (26°00′36″N ~ 26°03′42″N, 119°34′12″E ~ 119°40′40″E), which is located in the south of the Min River estuary (Fig. 1a). The tide is typical semidiurnal tide and the mean tidal range is 4.37 ~ 4.46 m (Dai 2004). During each tidal inundation, the marsh in intertidal zone is generally submerged for 3 ~ 3.5 h. The climate is warm and wet, with a mean annual temperature of 19.6 °C and a mean annual precipitation of 1350 mm (Zheng et al. 2006). The marsh soil is dominated by saline soil and the main vegetations include P. australis, C. malaccensis, C. compressus and S. triqueter.

Sample collection
The space-for-time substitution method was used to investigate the influence of spatial expansion between P. australis and C. malaccensis on temporal variations and bioaccumulation of V in coastal marshes. The P. australis (PA) and C. malaccensis (CM) communities represented the stage of before expansion, while the P. australis-C. malaccensis (PA'-CM') community in ecotone represented the stage of during expansion. The elevations and hydrological regimes of different communities were similar. Three experimental plots (50 m × 50 m) were randomly laid in intertidal zone of the northwest Shanyutan (Fig. 1b, c). At each plot, three subplots (20 m × 20 m) were laid in PA marsh (PAM), ecotonal marsh (EM) and CM marsh (CMM), respectively. Field sampling was conducted at above subplots in March, July, October and January in 2016, which represented spring, summer, autumn and winter, respectively.
Aboveground and belowground biomasses were determined using quadrat method (50 cm × 50 cm) at spatial scale in each subplot (three replications). The aboveground part of plants in the quadrat was clipped near the ground and the roots were dug out. The height and density of plants were measured, the roots were washed and the stem, leaf and standing litter were separated. All plant samples were washed thoroughly with deionized water and then were oven-dried at 80 ℃ for 48 h. After dry weight measurement, the samples were ground into fine powder. Because the substantial roots (> 98%) of P. australis and C. malaccensis were distributed in 0-60 cm depth (Li et al. 2020a), the V levels in soils of this depth were studied. Three columnar samples (0-60 cm) were obtained from the same position with plant samples. After the columnar samples were extracted, they were divided at 10 cm interval. The soil samples were air-dried, ground and sieved through a 100-mesh nylon sieve. Soil pH and electrical conductivity (EC) in different depths (at 10 cm interval) were determined in situ by portable pH meter (IQ150, Spectrum, USA) and Soil & Solution EC meter (Field Scout, Spectrum, USA), respectively. Three single soil cores (5.0 cm diameter) were sampled from each layer and weighed for bulk density (BD) and moisture determination after being oven-dried at 105 °C for 24 h. The physical and chemical properties of topsoil in different marshes were shown in Table 1.

Sample analyses
A 0.0500 (± 0.0005) g homogenized sample was digested with 2 mL HNO 3 (70%) and 2 mL H 2 O 2 (30%) at 180℃ for 15 h. The residue was diluted to 40 mL with deionized water for analyzing V levels by inductively coupled plasma mass spectrometry (XSeriesII, Thermo Company, USA). Quality assurance and quality control were assessed using duplicates, method blanks and certified reference material (GBW10020) from the National Research Center for Standards in China with each batch of samples (two blank and one standard for each 20 samples). The recoveries of samples spiked with standards ranged from 83.2% to 111.2%. Soil organic matter (SOM) was determined by soil nutrient analyzer (TFW, Wuhan Tianlian Apparatus Company, China). Soil particle size was analyzed using a laser particle size analyzer (Mastersizer 2000, Malvern Instruments, UK).

Parameter calculations
The V stock (T V , mg·m −2 ) in soil was calculated by the following equation (Wu et al. 2020): where BD i (g·cm −3 ) is soil bulk density of the i layer; V i (mg·kg −1 ) is V level in the i layer; and h i is soil depth (10 cm).
The V stocks in tissues of plant (root, stem, leaf and standing litter) (V i , mg·m −2 ) were calculated according to Li and Redmann (1992): where C i (mg·g −1 ) is V content in the i part; and B i (g·m −2 ) is biomass of the i part.
Bioconcentration factors [BCF] in relation to the V contents in soil was calculated by the ratio [Element] plant and [Element] soil (Duman et al. 2007): where C root , C stem , C leaf , C litter and C soil were the V contents (mg kg −1 ) in root, stem, leaf, litter and soil, respectively.
The V concentration quotients for roots/stems (R/S), roots/leaves (R/L) and stems/leaves (S/L) were calculated according to Dahmani-Muller et al (2000): where C root , C stem and C leaf were the same as above.

Statistical analyses
Statistical analyses were performed using SPSS 19.0 and Origin 8.5 for Windows. The Shapiro-Wilk test was applied to identify the normality of data before the statistical analyses were conducted. The analysis of variance (ANOVA) test was used to determine if the V levels in soils differed significantly among seasons or marshes. Also, it was used to determine if the V contents in plants differed significantly among seasons or tissues (p < 0.05). If ANOVA showed significant differences, multiple comparison of means was undertaken by Tukey's test with a significance level of p < 0.05. The stepwise linear regression analysis was used to explore the key factors affecting the variation of V levels in marsh soil based on environmental variables. The principal component analysis (PCA) was used as a first exploratory analysis to better visualize the possible environmental gradients determining the variations of V contents in soils of different marshes.

Temporal variation of V contents in soils of different marshes
Over all sampling seasons, dissimilar variations of V content in soils of different marshes were observed (Fig. 2). Generally, higher V levels in soils of PAM, EM and CMM occurred in spring and winter ( Table 2). The V contents in PAM, EM and CMM soils between spring and winter, between summer and winter and between autumn and winter showed significant differences (p < 0.05). The variations of V contents in soils also differed among marshes (Fig. 2). With a few exceptions, the V levels in profiles of EM were much higher than those of PAM and CMM. Compared to PAM and CMM, the mean V levels in EM soil over all sampling seasons increased by 7.2% and 2.2%, respectively. Significant differences of V contents in PAM and CMM soils were observed in spring and autumn (p < 0.05), while those in PAM and EM soils occurred during summer and autumn (p < 0.05).

Temporal variation in ecological traits of plants
Over all sampling seasons, the aboveground and belowground biomasses of PA were significantly higher than those of PA' (p < 0.05). Similarly, the aboveground and belowground biomasses of CM were much higher than those of CM' (p > 0.05) (Fig. 3a). For the four plants, stem was main body of aboveground biomass. Particularly, the percent of stem biomass in aboveground biomass of CM or CM' exceeded 99%. The heights of PA were significantly higher than those of PA' (p < 0.05), while the densities of PA were much lower than those of PA' (p > 0.05). On the contrary, the heights of CM were much lower than those of CM' (p > 0.05), whereas the densities of CM were significantly higher than those of CM' (p < 0.05) (Fig. 3b).

Variation of V contents in plants
The V contents in tissues of plants differed among seasons (Fig. 4). Generally, the V levels in roots and leaves of PA or PA' reached the maximums in spring, while those of CM achieved the highest values in autumn. By comparison, the V contents in roots and leaves of CM' reached the maximums during autumn or winter. The highest V levels in stems of PA' and CM occurred in autumn, whereas those of PA and CM' were observed during winter or spring. The V contents in plants also differed among tissues and, over all sampling seasons, almost all values in roots of PA, PA', CM' and CM were significantly higher than those in other tissues (p < 0.05). The V levels in tissues of PA' were generally  higher than those of PA and, similarly, the values in organs of CM' were much higher than those of CM. With a few exceptions, the V contents in tissues of CM' in the four seasons were much higher than those of PA'.

Transfer and accumulation of V in plants
Dissimilar variations of R/S, R/L and S/L ratios in plants were observed in different seasons (Table 3). The R/S and R/L ratios in PA or CM' reached the maximums in summer, while those in PA' achieved the highest values in spring. The highest S/L ratios in CM and CM' occurred in spring, whereas those in PA and PA' were observed in autumn or winter. The R/S and R/L ratios in PA' were generally lower than those in PA, while the values in CM' were much higher than those in CM. Except for spring, the R/S ratios in CM' in other seasons were much higher than those in PA'. The R/S and R/L ratios in PA, PA', CM' or CM were larger than 1, but the S/L ratios were less than 1. Over all sampling seasons, the [BCF] in aboveground parts were generally lower

Temporal variations of V stock and allocation in plant-soil systems
Over all sampling seasons, similar variations of V stock in profiles of different marshes were observed (Fig. 5). The V stocks in soils of EM in the four seasons were much higher than those of PAM and CMM. The higher V stocks in PAM, EM and CMM soils occurred in winter, while the lower values were observed in summer or autumn. The V stock in soil was the key body of total stock in plant-soil system (> 99%) ( Table 5). Allocations of V in organs differed among seasons or species. Roots were the main V stock of plant subsystems, which got the higher values in spring or winter. Allocations of V in aboveground parts of PA', CM' and CM reached the maximums in autumn, while those of PA achieved the highest values in winter.

Temporal variation of V contents in marsh soils
This paper found that the mean V levels in marsh soils over all sampling seasons ranged from 99.71 to 108.41 mg·kg −1 which exceeded its background value in soils of Fujian province (78.3 mg·kg −1 ) (Chen et al. 1992) but was slightly lower than its background value in terrestrial surface of China (112 mg·kg −1 ) (Nie 2011), implying that the V in marsh soils of the Min River estuary existed enrichment process. It was reported that the V enrichment in soils was mainly dependent on parent material and pedogenesis (Li et al. 2020a), and higher V contents generally occurred in soils originated from parent rock with higher V levels (Wang and Liu 1994). Besides, atmospheric deposition and anthropogenic import also influenced the V enrichment in soils . In this paper, since the study region located in the National Nature Reserve of the Min River estuary where it was strictly protected in recent 20 years, the V levels in marsh soil primarily rested with its geochemical enrichment process. This paper indicated that the V contents in soils differed among marshes or seasons (Fig. 2), which might rest with the differences in physical and chemical properties of soils in PAM, EM and CMM. The stepwise linear regression analyses showed that the variation of V levels in soils of PAM could be better explained by soil temperature (x 1 ) (y = -0.93x 1 + 122.764, R 2 = 0.299, p < 0.001), while those of CMM could be better explained by soil temperature (x 1 ), sand (x 2 ) and SOM (x 3 ) (y = -1.093x 1 -0.362x 2 + 1.798x 3 + 131.3, R 2 = 0.635, p < 0.001). By comparison, the variation of V contents in soils of EM could be better explained by pH (x 4 ) and BD (x 5 ) (y = 6.115x 4 -101.983x 5 + 160.575, R 2 = 0.456, p < 0.001). These implied that soil temperature, SOM and pH might be important factors influencing the temporal variations of V levels in soils of different marshes. Previous studies have reported that thermal conditions and SOM significantly affected the adsorption-desorption of metallic ions in marsh soils (Boyer et al. 2018;Li et al. 2020c) and higher SOM was generally favorable for enhancing V adsorption due to its strong complexing capacity (Du Laing et al. 2009;Zhu et al. 2016). In this paper, the variations of soil temperature over all sampling seasons could partly explain the temporal variations of V levels in soils of different marshes. Moreover, the relatively higher SOM contents in EM soil might also explain its higher V levels (Table 1, Fig. 2). It was also reported that, under acidic condition, metals generally existed in free or ionized state which showed strong mobility (Huang 2003;Lu and Yan 2010). In the study region, the marsh soils were acidity and were greatly affected by acid deposition (Pan 2001;Li et al. 2020c). Although the pH in soils of PAM, EM and CMM showed narrow ranges over all sampling seasons (Table 1), the mobility of V in soils might be active due to the lower pH. This conclusion could partly explain the lower V levels in PAM soil since the lower pH was observed (Fig. 2, Table 1). This paper implied that the spatial expansion between PA and CM generally increased the V contents in EM soil over all sampling seasons (Fig. 2), and, compared with PAM and CMM, the mean values increased by 7.2% and 2.2%, respectively. As shown in Table 1, the physical and chemical properties of EM soil were greatly altered during spatial expansion, which might influence the variation of V levels. Similar results were reported by Ehrenfeld (2003) and Chacón et al. (2009) who reported that the alterations of plant species, community structure and ecological traits during alien species invasion significantly affected the physical and chemical properties of soils. Compared with PA or CM communities, both PA' and CM' in ecotone showed higher densities. However, the former generally occupied the higher spaces, while the latter occupied the lower spaces (Fig. 3b). Just for this reason, the special space combination between PA' and CM' in EM might be more favorable for intercepting the suspended particulate matter in tide. Actually, the fine particles (clay and silt) in topsoil of EM did show increasing trend and, compared with PAM and CMM, the values increased by 13.85% and 29.10%, respectively (Table 1). These indicated that considerable V element might be imported into ecotone simultaneously, resulting in the higher V levels in soils of EM. As shown in Fig. 3, the ecological traits of plants were also greatly altered during spatial expansion, which might affect the variation of V levels in soils. Previous studies have reported that there were great differences in V absorption and accumulation among vegetations (Nawaz et al. 2018;Li et al. 2020a). In this paper, both belowground and aboveground biomasses of plants (PA' or CM') in ecotone were much lower than those in pure communities (PA or CM) (Fig. 3a), which implied that the V absorption amounts by PA' and CM' might not be very high and this could partly explain the higher V levels in EM soil.
In order to better visualize the possible environmental gradients determining the temporal variation of V contents in different marsh soils, the principal component analysis (PCA) was conducted (Fig. 6). In PAM, two principle components explained 92.98% of the variance. Principal component 1 (PC1), which explained 87.19% of the total variance, represented the gradient variations of plant height and aboveground biomass. Principal component 2 (PC2), which explained 5.79% of the total variance, showed the gradient variations of plant density, soil temperature and SOM. Further analyses indicated that the V levels showed strong correlation with PC2. By comparison, 84.73% of the total variance of environmental variables in EM was explained by two principle components (PC1 and PC2). PC1 represented the gradient variation of plant height and aboveground biomass, whereas PC2 showed the gradient variation of plant density and belowground biomass. Generally, the V levels showed close correlation with PC2. For CMM, 59.38% and 13.95% of the total variance of environmental variables were explained by PC1 and PC2, respectively. PC1 represented the gradient variations of plant height, aboveground biomass and pH, while PC2 showed the gradient variations of plant density and belowground biomass. As a whole, the V levels showed strong correlation with PC1. The above analyses indicated that the temporal variation of V levels in soils of EM, to a great extent, rested with the alterations of pH, SOM and plant ecological traits during the spatial expansion between PA and CM.

Accumulation and transference of V in marsh plants
This paper implied that the V contents in PA, PA', CM' and CM differed among tissues and, over all sampling seasons, the values in roots were significantly higher than those in other tissues (Fig. 4). Previous studies have indicated that, as V existed in growing medium, bioaccumulation was a key adaptive strategy for most plants (Saco et al. 2013;Hou et al. 2014). Generally, the bioaccumulation for V occurred in roots, which was 2 ~ 1000 folds of aboveground parts (Aihemaiti et al. 2020). The primary reason was that the V in roots could form stable compound with calcium through the chelating and complexating the polar compound in cytoderm, which retarded the transference of soluble vanadium ion and reduced the V bioaccumulation in aboveground parts (Kaplan et al. 1990). In this paper, the ratios of R/S and R/L in PA, PA', CM' or CM were larger than 1 (Table 3), implying that the V contents in roots of the four plants were significantly higher than those in aboveground parts. The conclusion could be better verified by the higher [BCF] in roots of different plants over all sampling seasons (Table 4). Besides, the S/L ratios in marsh plants were less than 1 ( Table 3), indicating that the limited V transferred from roots to the aboveground parts might be preferentially allocated to leaves and this was favorable for the biosynthesis of chlorophyll and the metabolism of carbohydrates in photosynthesis process (Nawaz et al. 2018). It should be noted that, over all sampling seasons, the V contents in standing litters were c much higher that those in stems and leaves (Fig. 4). One possible explanation could be ascribed to the nutrients (including V) absorbed by plants and the nutrients retained in standing litters (Chen and Sun 2020b). As mentioned above, the V in living bodies could form stable compound with calcium in cytoderm, which indicated that, as plant withered, the V in these compounds might be stranded in standing litters in large numbers due to its poor mobility. This paper showed that the V contents in tissues differed among species. Over all sampling seasons, the V levels in tissues of PA' were generally higher than those of PA and, similarly, the values in organs of CM' were much higher than those of CM (Fig. 4). The probable reason might be related to the alteration of plant ecological traits and the competitive absorption for nutrients (including V) by plants during spatial expansion. Compared to pure community, the aboveground and belowground biomasses of PA' (or CM') significantly decreased (Fig. 3a), indicating that both the living spaces for PA' and CM' in ecotone were severely squeezed and their competitiveness for nutrient might be more intense. Previous studies have found that, in habitat with limited nutrient, the competition advantage of plants generally rested with their conserved utilization for limited resources (Sardans and Peñuelas 2014;Wang et al. 2018b). It was also reported that the spatial expansion of PA and CM in the Min River estuary was bi-directional (He et al. 2018) and the competition between them rested with their ecological adaptation strategies (Li et al. 2020a). In this paper, the V contents and the [BCF] in tissues of CM' over all sampling seasons were generally higher than those of PA' (Fig. 4, Table 4), indicating that the PA' and CM' in ecotone might adopt different strategies for V absorption and utilization to maintain their competitiveness. Compared to PA, the density of PA' increased but its height and R/S and R/L ratios generally decreased (Fig. 3, Table 3), which implied that the P. australis might compete primarily by increasing the number of tillering and transferring the V in roots to the photosynthetic organ (leaf) preferentially. However, compared with CM, the density of CM' decreased but its height and R/S and R/L ratios generally increased (Fig. 3, Table 3), indicating that the C. malaccensis might resist the spatial expansion of P. australis by increasing the V bioaccumulation in roots, decreasing the number of tillering and expanding the living space of aboveground parts. This paper indicated that the V levels in tissues of PA, PA', CM' and CM differed among seasons, which could be better interpreted by the differences in growth rhythm and ecological traits of the four plants. In this paper, the V contents in roots and leaves of PA (or PA') in spring were the highest, while those in summer and autumn were much lower (Fig. 4), implying that, compared to the vigorous growth stage, the roots of PA (or PA') at initial growth stage showed higher V bioaccumulation and higher V transference from roots to the leaves. By comparison, the V levels in roots and stems of CM (or CM') achieved the higher values in winter (Fig. 4). Previous studies have reported that the aboveground parts of PA (or PA') almost withered in winter, while those of CM (or CM') were not dead (Wu et al. 2020;Li et al. 2020a). Additionally, stem was the main body of aboveground parts of CM (or CM') in winter (Fig. 3). Thus, in order to keep alive and enhance the stress resistance for the lower temperature in winter, the V accumulated in roots of CM (or CM') might be transferred to the stems. This paper also indicated that, except for the V stocks in stems of PA' in spring and winter, the values in its stems in other seasons and the ones in its leaves over all sampling seasons were much higher than those of CM' (Table 5)., The reason probably rested with the different strategies for V absorption and utilization adopted by the two plants during their spatial expansion, and this has been discussed previously. Compared to CM', the PA' in ecotone might preferentially transfer the V in roots to the aboveground parts to maintain its competitiveness, which was an important reason in inducing the higher V stock in its aboveground parts.

Conclusions
This paper investigated the impacts of spatial expansion between P. australis and C. malaccensis on temporal variations and bioaccumulation of V in coastal marshes of the Min River estuary. Results have demonstrated that: i) the V contents in marsh soils over all sampling seasons exceeded its background value in soils of Fujian province; ii) the temporal variation of V levels in EM soil rested with the alterations of pH, SOM and plant ecological traits during spatial expansion; iii) the V contents in PA, PA', CM' and CM differed among tissues and higher bioaccumulation occurred in roots; and iv) the V levels in tissues differed among species or seasons, which were ascribed to the differences in ecological traits among plants and the competitive absorption for V by plants during spatial expansion. This paper found that the V in soils of the Min River estuary existed enrichment process and the spatial expansion between PA and CM promoted its enrichment in soils and its bioaccumulation by marsh plants. Data availability All data produced from this study are provided in this manuscript.
Code availability Not applicable.

Conflicts of interest
The authors declare that they have no competing interests.

Consent for publication Not applicable.
Ethical approval and consent to participate Not applicable.