A case study of soil food web components affected by Fallopia japonica (Polygonaceae) in three natural habitats in Central Europe

The experiment was conducted in a valley near the village of Opátka in South Eastern Slovakia, Central Europe. This region has a temperate climate, with an annual average of 40 summer days per year and a warm, moderately dry sub-region with a mild winter. The average daily temperature in January ranges from 1.5 to 4.0°C, the average daily temperature in July ranges from 16.0 to 18.5°C, while the average annual temperature ranges from 5.0 to 7.0°C. The mean annual precipitation is 650 to 700 mm. The soils are characterized as Fluvisols, and the vegetation zone is characterized as Carpathian oak-hornbeam forest. The landscape is patchy, with deeply undulating uplands (Miklós, 2002). The first F. japonica specimens appeared in the village of Opátka around 1992 (personal communication with forester of the cadastre). It was later probably transferred to the entire valley (7 km) below the village, invaded the banks of the creek and then spread to adjacent habitats thus creating large monocultures.

For studying the impact of F. japonica on soil physical properties, microbial, and soil nematode communities, we selected three habitats in the valley, namely, forest, grassland, and wetland and corresponding areas adjacent to them invaded by F. japonica.

Forest (F) (48°47.63′N, 21°03.43′E; 455 m a.s.l.): covered by a natural, undisturbed, 100 years old deciduous Querco-Fagetea forest, mainly consisting of Quercus robur, Q. cerris, Carpinus betulus, Acer campestre, and many shrub species such as mostly Viburnum sp. and Prunus spinosa.

Forest edge invaded by F. japonica (FF): a nearly monospecific stand of F. japonica covering an area of 500 m², with an estimated time of invasion of 15 years.

Grassland (G) (48°48.14′N, 21°03.40′E; 392 m a.s.l.): covered with indigenous multispecies vegetation dominated by Dactylis glomerata, Lolium perenne, Trifolium pratense, and Achillea millefolium; irregularly mown.

Grassland edge invaded by F. japonica (GF): an adjacent of F. japonica covering an area of 250 m², with an estimated time of invasion of 15 years.

Wetland (W) (48°48.34′N, 21°03.35′E; 386 m a.s.l.): covered by Petasites hybridus, Caltha palustris, Galium aparine, Equisetum sp., Ranunculus sp., and Urtica sp.; the soil is regularly flooded mostly in the spring, and the vegetation is mown once a year in the autumn.

Wetland edge invaded by F. japonica (WF): an adjacent area of 200 m², with an estimated time of invasion of 10 years, the soil is regularly flooded mostly in the spring.

We selected a 25 m × 25 m area of the three different habitats (F, G, and W) which was not yet colonized by F. japonica. The distance between F, G, and W along the valley was approximately 1,500 m. A pair of the invaded and the uninvaded areas which did not differ in elevation, inclination, exposition, or management were chosen and the distance between the invaded and the uninvaded areas was 50 m. In each invaded area, we installed five randomly chosen 1 × 1 m plots (approx. 10 m apart) which had similar cover of F. japonica. Similar, five 1 × 1 m plots with random distribution were installed in corresponding uninvaded F, G, and W areas. This resulted in 30 plots (five plots × two invasion state [invaded and uninvaded] × three habitats [F, G, and W]). The uninvaded areas were assumed to represent the situation prior to the invasion of F. japonica.

The soils were sampled using a garden trowel to depths of 0 to 20 cm in May 2016. A quadrat sampling method was used. Five soil subsamples were collected from each quadrat (1 m²), one from each corner and one from the center. The subsamples from each quadrat were then bulked to obtain five representative soil samples (1 kg) for each area. The soil samples were transferred to the laboratory in plastic bags. The bags were stored at 5°C until processing (storage time of soil samples were no longer than one week). Each sample was gently homogenized manually before processing

Soil pH was determined for air-dried soil samples in a 1:3 solution of soil: 0.01 M CaCl₂ using a pH meter inoLab pH 720-WTW GmbH, Weilheim, Germany. Soil moisture content was measured gravimetrically after the soil had been dried to a constant weight in an oven at 105°C for 24 hr. All determinations were performed in triplicate.

Soil microbial respiration (SMR) was measured by the amount of CO₂ released from 100 g of field-moist soil and absorbed by NaOH (μg C-CO₂/g soil) in hermetically sealed bottles (Alef and Nannipieri, 1995) at 25°C for 24 hr. Microbial biomass carbon (MBC) content was determined using the method of Islam and Weil (1998), as oven-dried equivalent (ODE) of field-moist soil adjusted to 80% water-filled porosity was irradiated twice by microwave (MW) energy at 400 J g^-1 ODE soil to kill the microorganisms. The time settings and MW oven power depended on the total amount of soil in the MW oven. After cooling, soil samples were extracted with 0.5 M K₂SO₄. Carbon content (C_irradiated) in the extract was quantified by the oxidation with K₂Cr₂O₇ dissolved in H₂SO₄ and titrimetrically by (NH₄)₂Fe(SO₄)₂. The same procedure was done with a non-irradiated sample (C_{non-irradiated}). The microbial biomass carbon was then determined as MBC = (C_irradiated-C_{non-irradiated})/K_ME, whereby extraction efficiency factor K_ME = 0.213. The activities of acid and alkaline phosphatase were determined by the modified method of Grejtovský (1991) using p-nitrophenyl phosphate as a substrate with incubation at 37°C for 24 hr. Urease activity was determined using urea as a substrate with incubation at 37°C for 3 hr as described by Chazijev (1976), and invertase activity was determined using sucrose as a substrate with incubation at 37°C for 24 hr as described by Schinner and Vonmersi (1990). The control measurements for enzymatic activity did not use the substrate. The activity of all enzymes was measured spectrophotometrically by create a reference curve.

Nematodes were isolated from 100 g of the mixed fresh soil samples by a combination of Cobb sieving and decanting (Cobb, 1918) and a modified Baermann techniques (Van Bezooijen, 2006). Nematodes were extracted from aqueous soil suspensions using a set of two cotton-propylene filters. Subsamples were removed after extraction for 48 hr at room temperature. The aqueous suspensions containing nematodes were examined under a stereomicroscope, excessive water was removed, and the nematodes were fixed in hot fixative 99:1 solution of 4% formaldehyde: pure glycerol and evaluated on permanent glycerine slides (Southey, 1986). All isolated nematodes were microscopically examined at 100, 200, 400, 600, and 1,000 × magnification, identified from permanent glycerine slides mostly to species level (juveniles were identified to genus level) using an Eclipse 90i Nikon, Japan light microscope, with original species descriptions, and several taxonomic keys: Brzeski (1998), Loof (1999), Siddiqi (2000), Andrássy (2005, 2007, 2009), and Geraert (2008, 2010).

Cysts of Heterodera juveniles were extracted by floatation (Sabová and Valocká, 1980) from 100 g of soil for species identification based on morphological markers and morphometric data for both cysts and juveniles.

Nematode species were assigned to trophic groups: bacterivores, fungivores, omnivores, predators, plant parasites, root-fungal feeders, and insect parasites, according to Yeates et al. (1993) and Wasilewska (1997).

The total number of species, total nematode abundance, mean number of nematodes per trophic group, and the Shannon and Weaver species diversity index (H’spp.) (Shannon and Weaver, 1949) were determined. Basic ecological indices were used to assess the status of the soil habitats using nematode communities. The maturity index (MI) for free-living taxa and the plant parasite index (PPI) for plant-parasitic taxa (Bongers, 1990), the enrichment (EI), structural (SI), channel (CI) (Ferris et al., 2001), and basal (BI) indices (Berkelmans et al., 2003) were calculated using the online program ‘NINJA: An automated calculation system for nematode-based biological monitoring’ (Sieriebriennikov et al., 2014; http://spark.rstudio.com/bsierieb/ninja).

The differences in nematode characteristics (total nematode abundance, abundance of nematodes per trophic group, and species diversity) and basic ecological characteristics (MI, PPI, EI, SI, CI, and BI) were analyzed with two-way ANOVA with ‘ecosystems’ (F, G, and W), ‘invasion status’ (invaded, uninvaded), and their interactions as factors. To meet the assumptions of these parametric tests, Box-Cox transformation was applied with the maximum likelihood approach and Golden Search iterative procedure on. If there was an interaction between ‘ecosystem’ and ‘invasion status’ (total nematode abundance and mean number of bacterivores), post hoc Fisher LSD test was applied separately for each ecosystem to determine the effect of ‘invasion status’. Otherwise, main factor ANOVA with two factors (without interaction) was applied. Consequently, in the case of confirmed significance of ‘ecosystem’, post hoc Fisher LSD test was used.

As untransformed soil physical and microbial properties (pH, SMR, MBC content, and enzymatic activities) were not normally distributed (Shapiro–Wilk test) and transformation did not improve normality, nonparametric statistics were applied. Differences among six combinations of ‘ecosystem’ and ‘invasion status’ were tested separately with Kruskal–Wallis ANOVA, followed by a post hoc multiple comparisons.

The above mentioned statistical analysis were performed using Statistica Cz, version 12.0 (Statsoft, Inc., 2013) and significance of all tests was determined at p<0.05, 0.01, and 0.001.

Relationships between plots, nematodes, and selected environmental characteristics (soil pH and soil microbial respiration as constrained variables) were analyzed by ordination techniques. Redundancy analysis (RDA) was performed using Canoco 5 (Ter Braak and Šmilauer, 2012), because response data were compositional and had a gradient of 1.7 standard deviations. The significance of the axis was tested by a Monte Carlo permutation test.

Non-metric multidimensional scaling (NMS) ordination was used to examine any changes in the structure of nematode community for the invaded and the uninvaded habitats. A three-dimensional solution was executed by Autopilot, with the slow and thorough mode and Sørensen (Bray-Curtis) distance (recommended for community data). PC-ORD (McCune and Grace, 2002; McCune and Mefford, 2011) was used for the NMS analysis.

The soil physical properties, soil moisture, and pH differed substantially (Kruskal–Wallis statistics with p < 0.001) among the investigated plots (Table 1). The soil moisture content varied from 9.1% in F to 12.4% in W and from 11.7% in GF to 21.5% in WF. Multiple post hoc comparisons confirmed significantly (p < 0.05) higher soil moisture only in the invaded FF than the adjacent uninvaded F. The pH varied from 5.2 in F to 6.8 in W and from 6.4 in FF to 7.2 in WF, but no significant differences were observed between the invaded and the uninvaded plots.

SMR and MBC content were higher (but not significantly) in all plots with F. japonica (FF, GF, WF) than in the uninvaded plots (F, G, W) (Table 1). Opposite trends were found for acid phosphatase; it was lower (but not significantly) in the invaded (FF, GF, WF) than the uninvaded plots (F, G, W). Urease, alkaline phosphatase, and invertase enzymes were influenced by F. japonica invasion in different ways in the different ecosystems, but no significant differences were found.

A total of 9,452 individual nematodes were isolated and identified. The total number of species and genera of soil free-living and plant-parasitic nematodes was higher in the uninvaded (58 and 46) than in the invaded plots (49 and 40). The species found in the uninvaded (F, G and W) but absent in the invaded plots were as follows: six bacterivores Acrobeles cylindricus, Eucephalobus oxyuroides, Mesorhabditis sp. juv., Plectus tenuis, Prismatolaimus dolichurus, and Punctodora sp. juv.; one fungivore Tylencholaimus stecki; one omnivore Aporcelaimellus obtusicaudatus; two predators Coomansus parvus and Thonus ettersbergensis; two plant parasites Heterodera hordecalis and Meloidogyne hapla; and two root-fungal feeders Boleodorus thylactus and Coslenchus costatus. On the other hand, only the invaded plots (FF, GF, WF) contained the omnivore Axonchium coronatum (GF), the predators Coomansus zschokkei (GF) and Trischistoma monohystera (WF), and the plant parasites Hemicycliophora typica (WF) and Heterodera sp. 1 (FF) (Table 2). Mean nematode abundance was significantly higher in the uninvaded compare to the invaded plots (p < 0.001). Subsequent post hoc Fisher LSD test confirmed a significant effect of ‘invasion status’ for F compare to FF and for W compare to WF (both p<0.05), but not for G compare to GF. Species diversity index did not differ significantly between the invaded and the uninvaded plots (Table 3).

The ordination analysis identified two explanatory variables (pH and SMR) that accounted for 53.5% of the total variation of nematode species abundance (Fig. 1). Monte Carlo permutation tests identified statistical significance of all axes for this relation (pseudo F = 1.7, p = 0.01). Soil acidity and SMR were negatively correlated with the occurrence and abundance of most nematode species in F, G, and W, but soil pH positively correlated with the abundance of the predators A. coronatum and C. zschokkei in GF. The RDA partly separated the nematode communities of the uninvaded habitats from the invaded habitats. Cephalobus persegnis, Plectus parietinus, Prismatolaimus intermedius, and Mesodorylaimus bastiani were more abundant in the uninvaded plots, but Microdorylaimus parvus, Anatonchus tridentatus, and Rotylenchus robustus were more abundant in the invaded plots (Fig. 1, Table 2).

Figure 1:

The NMS analysis compared nematode composition based on species diversity. The best three-dimensional solution for the NMS ordination had a final stress of 10.40 (p < 0.0001) after 48 iterations, which was supported by a Monte Carlo permutation tests with a significance of p = 0.004 and a mean stress of 10.67 for real data and 250 runs for both real and randomised data. The variances explained by the first and second axes were 48 and 28%, respectively. The NMS analysis identified a notable impact of the invasive plants (Fig. 2). The samples from F, G, and W were clearly separated into one group, and the samples from FF, GF, and WF were separated into another group. Moreover, samples from forest (F) differed the most from the other samples, and were separated to the left in the figure, and were represented mostly by high abundances of Paratylenchus straeleni, Tripyla filicaudata, Filenchus misellus, and Prismatolaimus intermedius.

Figure 2:

A total of 62 nematode species were identified: 18 of which were bacterial feeders (29.0%), 15 were plant parasites (24.2%), 9 were omnivores (14.5%), 7 were predators (11.3%), 6 were fungivores and root-fungal feeders (both 9.7%), and 1 was an insect parasite (1.6%).

Plant parasites were the most abundant trophic group in F and G (Table 3). The abundance of plant parasites was significantly higher in the uninvaded rather than the invaded plots (p < 0.01). Bacterivores were the second most abundant trophic group in F and G and the most abundant trophic group in W. Subsequent post hoc Fisher LSD test confirmed a significantly higher bacterivore abundance (p < 0.05) in F compared to FF and in W relative to WF (p < 0.05), but not for G and GF was observed. The abundance of fungivores was very low in the forest compared to both the grassland and the wetland (p < 0.001), but without significant differences between the invaded and the uninvaded plots. The abundance of root-fungal feeders was significantly higher in the forest and grassland rather than the wetland (p < 0.05) and in the uninvaded than the invaded plots (p < 0.01). The abundance of omnivores was significantly higher in the uninvaded rather than the invaded plots (p < 0.001). No significant differences in the abundance of predators were found.

Maturity and Channel indices were significantly higher in the uninvaded compared to the invaded plots (p < 0.05). Structure and Basal indices did not significantly differ between the invaded and the uninvaded plots. When plotting the Enrichment and Structure Indices, most of the soil samples ended up in Quadrat B for both the invaded and the uninvaded plots (data not shown) suggesting that food webs were highly enriched and structured with both bacterial and fungal decomposition channels and maturing food web condition. No significant effect of ‘ecosystem’ was observed in the case of abovementioned nematode indices (Table 3).