Responses of plant species diversity and soil physical-chemical-microbial properties to Phragmites australis invasion along a density gradient

The invasion of ecosystems by strongly colonising plants such as Phragmites australis is viewed as one of the greatest threats to plant diversity and soil properties. This study compared a range of diversity measures including soil properties and mycorrhizal potential under different degrees of Phragmites density among three populations in coastal wetland, Victoria, Australia. Species richness, evenness and Shanon-Wiener index had significantly higher values in low degree of Phragmites density in all populations. Higher densities had the lowest diversity, with Shannon-Wiener index = 0 and Simpson’s index = 1 indicating its mono-specificity. Significant alterations in soil properties associated with different degrees of Phragmites density were noticed. These had interactive effects (population × density) on water content, dehydrogenase activity, microbial biomass (C, N and P) but not on pH, electrical conductivity, phenolics, organic carbon, and spore density. Furthermore, the study elucidated decrease of competitive abilities of native plants, by interfering with formation of mycorrhizal associations and biomass. Overall, our results suggest that significant ecological alterations in vegetation and soil variables (including mycorrhizal potential) were strongly dependent on Phragmites density. Such changes may lead to an important role in process of Phragmites invasion through disruption of functional relationships amongst those variables.

Biological invasions by strongly competitive native and non-native species, (commonly referred to as invasive species), is a worldwide phenomenon that threatens to dramatically change communities and ecosystems 1,2 . The expansion of invasive plant species both with and without human influence, has caused striking modifications of natural ecosystems that threaten critical ecosystem services 3 . Invasive plants are characterized by high productivity associated with several key competitive mechanisms that assist invasion, including greater nutrient use efficiency, early maturation, and allelopathy 4 . Many invasive species are tolerant of or even adapted to environmental disturbance, which is likely to accelerate their expansion 5 . As a result, they often displace less competitive native species, resulting in the loss of plant diversity 1 , which in turn, weakens the stability and functioning of ecosystems 6 .
Wetlands, including brackish marshes, represent some of the most productive natural ecosystems. They support various biological communities and biogeochemical functions providing heterogeneous nutrient conditions and diverse habitats, and they are key drivers of global nutrient cycling 7 . Topographic features of wetlands, such as their location in natural depressions and their frequent connectivity to both terrestrial and aquatic systems, make these ecosystems vulnerable to anthropogenic disturbances 8 . These features can facilitate the invasion of both non-native and native plants.
Phragmites australis (hence after Phragmites) is one of the most aggressive and widely studied invasive plant globally both within its native range and in areas where it is introduced 9 . The exact native range of Phragmites is Results Impact on diversity in plant communities. Along the transects, the density of Phragmites shoots increased from none to maximal values over a distance of only 12 m (Fig. 1) and the density was significantly varied among invasion categories within each population (Kruskall-Wallis: χ 2 = 13.62, 13.59 & 13.54 with df = 4 and p ≤ 0.05 for population 1, 2 & 3 respectively). The maximum density (285) was found in population_1 compared to others but there was no significant difference among the populations (ANOVA: F 2,42 = 0.212, p ≥ 0.05). A total of 9 plant species was identified in the random quadrats of the three populations ( Table 1). The number of species per quadrat was on average highest in the low density Phragmites stand and had been lowering as Phragmites density increased (Fig. 2). The diversity indices: species richness (S), Shannon-Wiener diversity index (H), species evenness (E) and Simpson's reciprocal index (D −1 ) differed significantly among the density categories within each population (For example, in population_1: Kruskall-Wallis: χ 2 = 13. 27, 11.18, 11.28 & 11.01 with df = 4 and p ≤ 0.05 for S, H, E & D −1 respectively) but there were no significant differences among three populations (ANOVA: F 2,42 = 2.45, 0.709, 0.095 & 0.550, p ≤ 0.05 for S, H, E & D −1 respectively). Values of all indices were significantly higher in low density compared to severe density in all populations. (Fig. 2: Dunn-Bonferroni, p < 0.05 for all pairwise comparisons). In all populations, the severe density had the lowest diversity, with H = 0 and D −1 = 1, i.e. they were monospecific stands. Again, the regression analysis showed that Phragmites stem density could significantly predict all the indices of the diversity in the community (For example, in population_1:    Table 1. A total of identified species in the invaded zone of three populations. Impact on soil properties. ANOVA results revealed that the degrees of Phragmites density × population interactions had significant effects on soil properties: water content (WC), dehydrogenase activity (DHA), microbial biomass (C, N and P), except pH, electrical conductivity (EC), phenolics, organic carbon (OC), and spore density (SD) ( Table 2). Relatively higher and statistically significant WC, EC, phenolic contents, OC with lower pH were found in highly density sites than in other density categories in all populations (Fig. 3). For example, in population_1, the physico-chemical properties like WC (F 4,10 = 3.98, P < 0.03), pH (F 4,10 = 4.58, P < 0.02), EC (F 4,10 = 9.25, P < 0.002), phenolics (F 4,10 = 12.75, P < 0.001), and OC (F 4,10 = 7.55, P < 0.005) varied significantly across the density categories. Significantly higher DHA, microbial biomass C, N and P with lower AMF spore density were found in highly density categories than in others (Fig. 4). Biological properties like AMF spore density (F 4,10 = 6.88, P < 0. Correlations and PCA. Significant negative correlations were found among soil properties and biodiversity indices in each population, except pH and AMF spore density (Table 3). Within each site, PCA consistently revealed two components of variation among variables. The loadings on the two principal components were similar at all sites except population_1 (Fig. 5). In population_1, two principal components were identified with eigenvalues larger than 1, explaining approximately 81% of the total variance. The first component, which explains 68% of the total variance within the dataset, is characterized by high positive loadings for all biodiversity indices with soil pH and AMF spore density.
Phragmites infested soil also significantly reduced the AGB, BGB and AMF colonization in Melaleuca ericifolia across the density categories in all populations (Fig. 6).

Discussion
A negative linear relationship between invasive species and species diversity in a given area is a general trend, which might be changed by a variety of environmental stressors. For example, soil moisture (water availability) may influence the relative performance of native and non-native invasive species. Phragmites can gain higher competitive advantage over other plant species (natives) in aquatic (highly moisturised) vs. terrestrial ecotypes 25,26 . Similarly, natives have superior physiological performance with diversity compared with invasive under less soil moisture condition 27 . Considering these, our findings regarding the relationships between the cover of Phragmites/others and plant community diversity were well supported with other studies 26 i.e. less cover of Phragmites had low soil resources (moisture and nutrients) with high biodiversity indices. Again, native species are often considered to be the driving force in increasing biodiversity at local scales 28 . In the present study, the native plants in the communities invaded by Phragmites contributed to increasing plant diversity when the density of Phragmites was at slight and moderate level. Once the cover exceeded, a negative effect of density on local  Table 2. Results of two-way ANOVA (F-ratios) evaluating the density-dependent role of Phragmites australis on soil properties. WC, water content; EC, electrical conductivity; OC, organic carbon; SD, spore density; DHA, dehydrogenase activity; MBC, microbial biomass carbon; MBN, microbial biomass nitrogen; and MBP, microbial biomass phosphorous. Level of significance: ***, **, * and ns indicate significant difference at P ≤ 0.001, P ≤ 0.01, P ≤ 0.05 and non-significant respectively. species diversity appeared, possibly due to disequilibrium of communities caused by the rapid growth of invasive plants 29 .
We also found the examined biodiversity indices (species richness, species diversity, species dominance, and species evenness) did not linearly decrease/increase with increases of Phragmites density. Instead, Phragmites slightly increased the plant diversity of the invaded community within a certain range of density values (slight density phase), possibly due to direct or indirect facilitation of non-native species during their initial introduction 30 . For example, invasive Sargassum muticum increased native species richness at a low percent cover 31 . This finding is consistent with a study by Melo et al. 32 indicating that human disturbance at an intensity below a certain threshold would not trigger irreversible biodiversity loss, and the delivery of ecosystem services would increase up to a point with increasing human disturbance. In this study, all the three populations were monospecific at high and severe density levels. However, Phragmites does not always lead to monospecific stands. If the area presents suboptimal conditions for Phragmites, such as fairly low water table or low nutrient availability, or if strong competitors are present, Phragmites may not be vigorous enough to competitively displace existing species and may coexist with them 33 . In this study, we observed that adjacent to the monospecific Phragmites stand in Cherry Lake, a zone of very tall and dense populations of the equally invasive Typha orientalis (or Typha spp.), overtopped the Phragmites and probably prevented it from becoming more dominant.
Despite numerous efforts to examine the influences of invasive plants on soil characteristics (physical, chemical and biological), inconsistent results are often obtained, especially under field conditions 17,22 . These inconsistencies possibly arise from complex and interlinked factors such as large spatial and temporal variation in soil characteristics, and uneven plant diversity between undisturbed and invaded regions. In this study, we selected sites that were located within a short distance of one another, and represented areas of undisturbed topography where former plant diversity was more or less similar between presently invaded and uninvaded areas. We In this study, different degrees of Phragmites density significantly influenced a range of soil physical-chemical-biological characteristics. High soil organic carbon, phenolics contents with low pH in invaded areas can be explained by additional organic matter supply by Phragmites through root exudation 11,14,34 and litter fall associated with their high productivity 15 . Along with the changes in soil physical-chemical characteristics, different degrees of Phragmites density had differential effects on belowground microbial activities. We found Phragmites density accelerated DHA, an indicator of microbial activities, which was consistent with the findings that phenolics may increase microbial activity 35 . These results were well aligned in this study due to linear relationships among increased organic carbon, phenolics contents, and dehydrogenase activity in Phragmites rhizosphere as well other studies Saviozzi et al. 36 stating that soil enzymatic activities are well related to organic carbon and phenolics..
We also found soil microbial biomass C, N and P were greatest underneath Phragmites and this increase relates to the degree of density substantiated with the study of Stefanowicz et al. 37 Martina et al. 38 . Our results clearly showed that soil properties under the canopy of Phragmites were profoundly different from areas outside the canopy even if they were only located a few metres away. The degree of organic matter accumulation in soil is generally dependent on the relationship between C input and decomposition rates from litter of plants. As Phragmites is a perennial graminaceous plant it returns more C to the soil from dying leaves and roots, than annual plants in the surrounding community 39 .This was clearly linked to high organic carbon accumulation in highly density soil in this study. The higher organic carbon in Phragmites invaded soil presumably contained substrates such as sugars, amino acids and organic acids that were readily metabolized, leading to the greater concentration of microbial biomass. The findings are well justified with the results of Yang et al. 40 through confirmation of positive soil microbial biomass carbon correlation with soil organic carbon.
The significant changes in microbial biomass and DHA were clearly associated with the different degrees of Phragmites density. The lower microbial biomass and DHA in the soil beneath the low degree density around patches of Phragmites may be caused by the absence of carbon inputs 37,40 . From the present study, again, we found variations of AM spore abundance and inoculum potential in soil colonised by Phragmites, which is aligned with the study of Sanon et al. 41 whom found lower spore density from invaded sites by the invasive species, Amaranthus viridis. Moreover, some studies have found that increased soil fertility shifted the microbial community structure and biomass, with a noticeable decrease in fungi and increase in bacteria 42 thereby validated in our mycorrhizal study findings. Again, it is assumed that the changes within microbial population are expected to affect the microbial biomass C:N ratio 43 , reflected in our study.. More acidic soils found under high degree density of Phragmites compared to others also supports the hypothesis that Phragmites has a greater influence on the proportion of microbial biomass depending on density.
Densely populated Phragmites soils showed a lower AMF inoculum potential in terms of Melaleuca ericifolia roots colonized by AMF in field collected soils. These results are directly compatible with other studies, most notably in a study of garlic mustard infested soil. Roberts and Anderson 44 reported a significant negative correlation between density of garlic mustard and inoculum potential of AMF. In addition, Stinson et al. 45 found invasive species Alliaria petiolata supressesed the native plant growth by disrupting the mutualistic association of belowground AMF potential with native tree seedlings. The results presented here suggest that reduced growth of Melaleuca ericifolia in Phragmites infested soil was linked to the AMF potential in soil. These findings strongly suggest AMF potential could be characterised as indirect allelopathy of invasive species 45 , for example, Phragmites. However, other factors such as nutrient dynamics 17 may alter plant growth and AMF composition and this complex interplay needs further investigation.

Population_1
Population_2  In conclusion, this study highlighted the variations of plant diversity indices, and soil properties including mycorrhizal inoculum potential influenced by different degrees of Phragmites density. Specifically, Phragmites depending on density greatly reduced the biodiversity indices and increased the DHA and soil microbial biomass (C, N, P). In addition, significant decrease of the mycorrhizal infection and AMF spore density were found at sites of the highest compared with lowest Phragmites density. Additionally, higher phenolic content, lower pH and mycorrhizal inoculum potential of soil in high density sites demonstrate that Phragmites competitive success may be partly achieved due to release of allelopathic compounds; a finding compatible with other studies 11,14,34,[46][47][48] . This study represents a step forward in our understanding of biodiversity indices and soil properties as affected by different degrees of Phragmites density, and provides valuable insights in advancing knowledge regarding the influence of different degrees of Phragmites density on soil microbial biomass and mycorrhizal inoculum potential. However, further research must be undertaken, concerning more native and non-native Phragmites on a wide geographic scale to identify species-specific invasion mechanisms. This work must include allelopathic processes, in particular the soil systems that may lead to a better understanding of the mechanisms related to soil mutualistic community degradation and their associated effects in natural ecosystems

Study site.
We conducted studies on the three natural stands of Phragmites population adjacent to Cherry lake (37°51′30″S, 144°50′5″E) (Fig. 7). An online tool called Scribble Maps (using the basic version) was used to create maps of (a) Australia; (b) Victoria (one of the states of Australia) and (c) study site (Cherry lake, Altona, Victoria, Australia) 49 . Scribble Maps is an advanced geographic information system and annotation tool that allows for the creation of custom maps 49 . Terrain views were used to identify target areas for the maps of the study sites (Fig. 7). Each population was separated from the others by a distance of at least 700 m. Cherry lake is a large remnant of a historically coastal wetland, Victoria, Australia.

Diversity measurement.
To evaluate the impact of invasion, field works were conducted mid-February 2014. A 75 m baseline was established through each Phragmites population, and 3 perpendicular transect lines, extending from low to high density into the area occupied by Phragmites, were chosen randomly along the baseline (Fig. 7d). The cover values of Phragmites were classified into five gradients, corresponding to zero (0%), slight (5~25%), moderate (26~50), high (51~75) and severe (76~100%) 50 . Alternatively, each transect was arranged into five density categories according to degrees of Phragmites density ranging from zero to severe level. Point-intercept sampling design along a transect line was used to collect soil and plant sample. Five (1 × 1-m) quadrats at 3 m intervals from baseline were established along a 12-m transect line with a density gradient of Phragmites. In total, 45 (3 × 5 × 3) vegetation quadrats from 3 populations were sampled for soil and plant diversity measurement. In each quadrat, all species of vascular plants and individual numbers of each species were recorded. Biological diversity was measured in terms of species richness (S), Shannon-Wiener diversity index (H), species evenness (E) and Simpson's reciprocal index (1/D) according to Keller 51 . Evenness was calculated as H/ln S, where S is the species richness expressed as the number of species.
Laboratory analysis. Soil samples were collected along each transect, encompassing a gradient representing a range of Phragmites densities from absent to dense Phragmites growth. At 3 m intervals along the transect 1 m 2 quadrats were made and three soil samples with 5 cm radius and 10 cm deep of Phrgamites rhizosphere from each density degree were collected. A total of 15 soil cores was taken from each transect i.e. in total, 135 (3 × 3 × 5 × 3) soil samples were collected from three populations. All soil samples were stored in sealed sterile bags and immediately transported back to the laboratory. The soil samples were passed through a 2 mm sieve to remove all debris. The three soil cores from each individual quadrat were thoroughly mixed to form a composite sample and then stored in a refrigerator at 4 °C for further processing. Soil samples were weighed prior to each further physical-chemical-microbial property analysis depending on the methods. Soil water content (WC) was determined by sampling 5 g of composite soil using the gravimetric method 52 . Soil pH and electrical conductivity (EC) were determined with a pH meter ((Pocket digital pH meter, 99559, DickSmith Electronics, Australia) and conductivity meter (TPS Digital conductivity meter, 2100, TPS Pty Ltd., Australia) with a 1: 2.5 w/v (soil: distilled water) ratio 53 . Soil phenolics was determined by sampling 100 mg of air dried soil following the Folin-Ciocalteu method 54 . Soil organic matter (SOM) was measured by sampling 5 g of air dried soil using the loss on ignition method 55 .
Soil arbuscular mycorrhizal fungi (AMF) spores or sporocarps were extracted from 25 g air dried soil in triplicate for each sample by wet-sieving followed by flotation centrifugation in 50% sucrose 56 . The finest sieve was 53 µm. The spores were collected on a grid patterned (4 × 4 mm) filter paper, washed with distilled water to spread them evenly over the entire grid and counted using a dissecting microscope at ×30 magnification. A sporocarp was counted as one spore. The number of spores was expressed as the mean of three replicates. Soil dehydrogenase activity (DHA) was determined using the method of Gu et al. 57 with little modification. Briefly, 2 g fresh sieved soil (2 mm) and 4 ml of 2 g L −1 2, 3, 5-triphenyltetrazolium chloride (TTC) were mixed thoroughly using mixer (Vortex mixer, VM1, Ratek Instruments Pty Ltd., Victoria, Australia) and incubated at 37 °C in dark for 24 h. After that 0.5 ml of 1 M sulphuric acid was added to stop the reaction and centrifuged at 4 °C and 4000 rpm for 10 min with addition of 4 ml of ethyl acetate. The colour intensity of the supernatant was measured at 485 nm in spectrophotometer (Biochrom Libra S12, England) with ethyl acetate as a blank. The DHA has been used to measure overall biological activity in soil 58 .
Soil microbial biomass C, N and P were determined by chloroform fumigation-extraction method using adjusted (water holding capacity) moisturised soil due to reactivate the soil microbes [59][60][61] . The soil samples were incubated in the dark for two weeks at 25 °C. Two sets of samples (5 g of each fumigated and un-fumigated, soil samples were weighed separately into 50 ml centrifuge tubes and 20 ml of 0.5 M K 2 SO 4 added to each. To one set of samples, 0.5 ml of ethanol free chloroform was added. Both fumigated and un-fumigated samples were capped and shaken simultaneously for 1 h. After shaking, the suspensions were allowed to settle for 10 min and the supernatants were filtered through Whatman No. 42 filter. For the sub-samples with chloroform, only the top 15 ml of the supernatant was filtered to reduce the amount chloroform in the filtrate. Filtrates from fumigated soil were also immediately bubbled with air for 30 min to remove any residual chloroform. Blanks were treated in the same manner. Then, soil microbial biomass C, N and P was determined using the filtrates accordingly. Total Organic Carbon (TOC) was determined following the procedures of Wüthrich et al. 62 with TOC analyser (SHIMADZU TOC-5000A). The difference in TOC between the fumigated and non-fumigated samples corresponds to the microbial carbon in the soil. TOC-contents were divided by 0.33 (K EC ) to estimate the C content of the microbial biomass 60 . Total N was determined using the method of Cabrera and Beare 61 with ion chromatograph (Shimadzu Ion Chromotograph, Kyoto, Japan) and calculated using K EN -factor (0.54) 63 . Soil inorganic P was extracted with 0.5 M NaHCO 3 and measured photometrically (UV/Visible spectrophotometer, Biochrom Libra S12, England) at 882 nm as a blue phosphate molybdic acid complex 64 and microbial P was calculated using K EP -factor (0.4) 64 .
Mycorrhizal inoculum potential bioassay. Each pot containing 100 g sieved composite soil from Phragmites density gradient soil was moistened uniformly with distilled water and incubated for 48 hrs at 25 °C. Melaleuca ericifolia seeds were selected for bioassay as this plant is a native wetland shrub and not commonly found in fresh and brackish water swamps across south-eastern Australia. The seeds were surface sterilised by washing in 70% ethyl alcohol for duration of 10 minutes. The seeds were sown on the soil surface in 45 mm plastic pots. Pots were placed in growth chamber (Westinghouse, Electrolux home products, Australia) set to 25/15 °C day/night temperature and a 12 h photoperiod with illumination of 84 µmols −1 m −2 . After 2 weeks, the seedlings were thinned in equal number (10 per dish) for growth and establishment. After three months, plant biomass was measured and the sampling of root tissue was undertaken by obtaining the roots of five plants for each treatment including the control. The staining of the roots for determining AMF root colonisation was completed accordingly 65,66 . Stained root sections were then examined with light microscopy through detecting either the presence or absence of arbuscules or hypha in sample root tissue 67 .

Statistical analyses.
The impact of Phragmites density on individual species was assessed by analysing changes in other species density among the density categories (zero to severe level) in each population. The 'Kruskal-Wallis' test, followed with pairwise comparisons using the Dunn-Bonferroni approach was adopted, as the data did not satisfy assumptions of normality and homogeneity of variance. One-way ANOVA with Bonferroni post-hoc analysis was conducted to compare biodiversity indices among different populations. Regression analysis was conducted to test the responsibility of Phragmites density to cause the variation in the diversity indices and soil properties. Again, one-way ANOVA with Bonferroni post-hoc analysis was conducted separately for each population to compare physico-chemical-biological variables among density categories. The Spearman correlation was used to assess the relationships between soil properties and biodiversity indices. The relationships among components of diversity and soil properties were quantified with correlation and principal components analyses (PCA). Principal components were based on correlation matrices, and were interpreted and presented if eigenvalues were ≥1 68 . Principal-components analyses were conducted individually for each of the three populations. AMF colonization and biomass data from the AMF inoculum potential of soil experiment were subjected to one-way and two-way ANOVA with significant test by LSD.