Response of Soil Fauna to the Shift in a Riparian Landscape along an Urban–Rural Habitat Gradient

Urbanization is accelerating worldwide, resulting in drastic alterations of natural riverbanks, which seriously affects the ecological functions and services of riparian landscapes. Our understanding of how anthropogenic activities influence soil animal communities within riparian zones is scarce. The soil fauna represents an important biotic component of the soil ecosystem and greatly contributes to soil structure and fertility formation. We investigated the richness, abundance, diversity, and distribution of soil animal groups, including macro- and mesofauna, in different riparian landscapes along an urban–rural habitat gradient. In natural riparian zones with permeable revetments, the soil fauna was richest and most abundant, mainly because of the low levels of human disturbance and the more suitable habitat conditions. Different soil animal groups responded differently to revetment type and distance from the water flow. The hygrophilous soil mesofauna, requiring a more humid environment, was more sensitive to shifts in revetment types, the location on the riverbank, and the seasons. In summer, when precipitation in the study area was highest, the abundance of the hygrophilous soil mesofauna was significantly higher than in autumn. Distance from the water flow significantly affected the abundance of the hygrophilous soil mesofauna. Our results demonstrated that hygrophilous soil mesofauna can serve as a good indicator in riparian zones, reflecting the hydrological conditions. We also observed interactions between revetment type and distance from the water flow; the distance effect was stronger in the natural riparian zone with a permeable revetment type. Our results highlight the importance of anthropogenic effects on soil ecosystem processes and functions in riparian landscapes, and the necessity of protecting and retaining the natural riverbank and native vegetation patches in riparian landscape planning and construction.


Introduction
Riparian zones represent ecotones between aquatic and terrestrial habitats. They are characterized by heterogeneous environmental conditions and various vegetation communities, playing an important role in providing ecosystem services and functions such as flow regulation, bank stabilization, refuges for local species, water purification, and green space for recreation [1,2]. Riparian zones are generally regarded as functionally unique ecosystems [1], and numerous studies have investigated the biodiversity of plants [3], birds [4], insects [2,5], and microbes [6] in these river-land interfaces. However, these ecosystems face multiple stresses such as global climate change, land-use conversion, reshaped water regimes, increasing human interference, biological invasion, and other site-specific issues which are mainly the results of increasingly expanding urbanization [7]. 70% of that of the entire year (https://rp5.ru/, accessed on 27 October 2021). The soil was classified as Ochri-Aquic Cambosols in the Chinese soil taxonomy, with an average pH of 6.64 and soil bulk density of 1.25 g·cm −3 (Table A1). Based on urban planning, distance from the city center, population density, vegetation characteristics, and a preliminary field survey, three riparian landscapes-including a natural riverbank zone (NR), a mixed-type revetment zone (MR), and a hard-type revetment zone (HR)-along an urban-rural gradient were selected to investigate the responses of the soil fauna to shifts in the revetment (or riverbank) type ("revetment type" is used in the following text for simplicity); these were located at three main rivers in Wenjiang district ( Figure 1, Table A2). The "HR" plots, located in the city, with impervious masonry, fewer natural meanders, and more exotic or cultivated plants, showed the highest degree of human disturbance. The "MR" plots, located in suburban areas, were usually composed of stones and soil, resulting in some water permeability. The anthropogenic interferences of "MR" plots were usually at a medium level due to less human activity. The "NR" plots, located in rural areas, showed the lowest degree of disturbance and were characterized by natural habitats with abundant native and spontaneous vegetation. Each riparian landscape plot was replicated three times (located at three main rivers), with a total of nine sampling plots; in each plot, three sampling locations with different distances from the edge of the riverbank (0-5-10 m) were selected to assess the effects of water flow on various soil animal groups. A distance of 0 m refers to sampling in an area (50 × 50 cm) very close to the edge. In all sampling plots, iButton thermometers (Maxim Integrated, San Jose, CA, USA) were arranged to record the soil temperature.

Collection and Identification of the Soil Fauna
For the convenience of practical research, the classification methods for soil animals according to their body size are generally used to determine their functions in soil ecosystems [20,21], and our study was no exception. Soil macrofauna taxa, such as earthworms, ants, spiders, and beetles, acting as ecosystem engineers or predators, are generally sampled via hand-sorting. Generally, the soil mesofauna is divided into two groups according to the extracting methods [22,23]. Among them, microarthropods are generally extracted using Tullgren funnels, and the hygrophilous mesofauna is generally extracted using Baermann funnels. Considering the effects of the hydrological conditions in riparian landscapes on the soil fauna, we classified the soil fauna into three groups (macrofauna, microarthropods, and hygrophilous mesofauna), based on body size and habitat preference.
Soil sampling was conducted in nine sampling plots for soil fauna extraction and analyses in summer (mid-July) and autumn (mid-October) of 2020, respectively. Overall, 162 soil samples were collected in each season (9 sampling plots × 6 soil samples × 3 soil animal groups). From each sampling plot, six blocks of soil (30 × 30 cm, 5 cm deep) were dug up and placed in a white porcelain tray for hand-sorting of the soil macrofauna. The macrofauna individuals picked from the soil were preserved in 75% ethanol and transported to the laboratory for further identification. Soil macrofauna abundance was expressed as individual density per square meter. For microarthropods, seven soil cores from the 0-5 cm soil layer were collected randomly (the sampling depth of soils was only 5 cm due to the thinner soil layers of urban soils) using a soil corer 2 cm in diameter, and mixed together to obtain one composite sample, with a total of six composite samples at each sampling plot. For the hygrophilous mesofauna, five soil cores from the 0-5 cm soil layer were collected randomly using a soil corer 2 cm in diameter, and mixed to obtain one composite sample, with a total of six composite samples at each sampling plot.
was replicated three times (located at three main rivers), with a total of nine sampling plots; in each plot, three sampling locations with different distances from the edge of the riverbank (0-5-10 m) were selected to assess the effects of water flow on various soil animal groups. A distance of 0 m refers to sampling in an area (50 × 50 cm) very close to the edge. In all sampling plots, iButton thermometers (Maxim Integrated, San Jose, CA, USA) were arranged to record the soil temperature. Figure 1. Distribution of the sampling plots along three main rivers in Wenjiang district. Sichuan Province is located in the southwest of China (A); Wenjiang is in the middle of Chengdu (B). Different urban-rural habitats are marked with round color blocks (C). Green color block represents a rural habitat, blue color block represents a suburban habitat, red color block represents an urban habitat (C). All sampling plots are marked with colored dots (C,D). Green dots represent a natural riverbank zone in a rural habitat, blue dots represent a mixed-type revetment zone in a suburban habitat, and red dots represent a hard-type revetment zone in an urban habitat (C). Yellow dots, purple dots, and black dots represent the sampling locations with different distances from the edge of the riverbank in each type of revetment zone (D). (E) Natural riverbank zone (NR); (F) mixed-type revetment zone (MR); (G) hard-type revetment zone (HR). Image (C) is exported from Planning Cloud (www.guihuayun. com, accessed on 27 October 2021).
All soil samples were placed into nylon bags, stored in a cooler, immediately transported to the laboratory, and processed within 48 h. For the extraction of microarthropods, all Tullgren funnels were covered with 1 mm mesh to prevent springtails from escaping. The total period of specimen extraction for microarthropods was 48 h. For the hygrophilous soil mesofauna (here, mainly nematodes, also including a few enchytraeids and Diptera larvae), the extraction periods with Baermann funnels were 4, 12, 24, and 48 h. Most of the enchytraeids were collected in the 4 h collection period. The final data for each soil sample using Baermann funnels were the sum of the results of four extraction periods. During the entire extraction process, the air temperature ranged between 35 and 40 • C, regardless of the funnel type. All specimens extracted from the soil were preserved in 75% ethanol. Macrofauna individuals retrieved from the field and mesofauna individuals extracted from funnels were counted and identified to the family or subfamily level under a microscope. Based on their different ecological functions, larvae and adult specimens were classified into different groups. For identification, Pictorial Keys to Soil Animals of China [23] was used.

Vegetation Investigation
To capture the composition and diversity of the vegetation in the different riparian landscapes, nine tree quadrats (3 riparian landscapes × 3 replicates), 10 × 10 m each, were established at the soil fauna sampling plots in summer (mid-July) and autumn (mid-October) of 2020. Five shrub quadrats (2 × 2 m) and five herb quadrats (1 × 1 m) were set up at two diagonals and the center point of each tree quadrat [31]. The species name, DBH (diameter at breast height), tree height, height below branches, crown depth, and canopy density of the trees were recorded in each tree quadrat. In the shrub and herb quadrats, the species names, individual number, height, coverage, and origin of the plants were recorded. Plant identification was based on the "Chinese Plant Illustrated Book" [32].

Soil Properties
During the field sampling in summer and autumn, undisturbed soil samples from the soil surface were collected using a soil corer 5 cm in diameter at each sampling plot and transported to the laboratory for soil bulk density (SBD) analysis [33]. Six soil samples (about 1 kg each) from the 0-5 cm layer were collected from each sampling plot with a small shovel, placed in airtight polyethylene bags, and brought to the laboratory. Subsequently, they were air-dried and ground to pass through 2.0 mm sieves for the determination of soil pH, ammonium nitrogen, available phosphorus, and available potassium. The soil water content (WC) was determined gravimetrically by oven drying the freshly collected soil at 105 • C to a constant mass for 48 h [34]. The soil pH was determined using a digital pH analyzer. After calibration of the pH analyzer, 10 g of soil and 25 mL of distilled water were shaken in a test tube and precipitated for 30 min. Subsequently, the pH of the soil slurry (1:2.5, soil/distilled water, [w/v]) was determined [35]. Ammonium nitrogen (AN) was determined using the phenol-hypochlorite method [36], and available phosphorus (AP) was analyzed using a spectrophotometer, following the Olsen method [37]. Available potassium (AK) was determined using an atomic absorption spectrophotometer [38]. Soil temperature at a depth of 5 cm (iButton thermometers) and air temperature were recorded at each sampling plot both in summer and autumn.

Anthropogenic Disturbance Measurement
Human traffic density was used to assess the anthropogenic disturbance degrees in various riparian landscape zones [39]. In mid-July and mid-October of 2020, anthropogenic disturbance was measured in nine sampling plots. We recorded the number of pedestrians walking across each sampling plot (10 × 10 m) from 8 a.m. to 8 p.m. and calculated the mean values and standard deviations for human traffic density per hour.

Calculation and Statistical Analysis
Soil fauna richness and abundance were measured using the taxonomic group number and individual density per square meter, respectively; the density-group index (DG) was used to determine the soil fauna diversity [40]. The diversity of the vegetation in each sampling quadrat was measured using the Shannon index (H ), the Pielou index (J), and the Simpson index (D).
The equations were as follows: For soil fauna, S is the taxonomic group number in each habitat; G is the total taxonomic group number in all habitats in each season; N i is the individual number of family i collected in each habitat; and N imax is the maximum individual number of family i in all habitats [40]; F i /F is the frequency of family i.
For vegetation, P i is the ratio of the individual number of species i to the individual number of all species in each quadrat [41]; N i is the individual number of species i; N is the individual number of all species in each quadrat; S is the number of species in each quadrat. According to the vertical structure of the community, the diversity indices of the tree layer, shrub layer, and herb layer should be weighted when calculating the total diversity index of the whole community [42,43], using the following equation: where C i is the coverage of the specific growth form, and i represents the growth form of vegetation including tree layer, shrub layer, and herb layer; C is the sum of C i ; H i is the average thickness of the leaf layer for the specific growth form; and H is the sum of H i . The thickness of the leaf layer for trees was calculated using 1/3 of the height of the tree layer, whereas the shrub layer was 1/2 and the herb layer was 100%. Wi is the weighted parameter of the diversity index of the ith growth form.
The community diversity index was calculated as follows: All statistical analyses were performed using R v. 4.1.2 [44]. Generalized linear models were performed using the MASS package and the following multiple comparisons using the multcomp package [45]: non-metric multidimensional scaling (NMDS), detrended correspondence analysis (DCA), canonical correspondence analysis (CCA), and redundancy analysis (RDA) were performed using the vegan package [46], and indicator species analysis (ISA) was performed using the indicspecies package [47]. Richness and abundance data were analyzed using generalized linear models with negative binomial distribution due to the overdispersion of the count data and the Gaussian distribution for the diversity indices. A Student's t-test was applied to examine the seasonal effects on the richness, abundance, and diversity indices of the soil fauna. Soil physical properties were analyzed using one-way analysis of variance (ANOVA), and the data were first tested for normal distribution and homogeneity of variance. If the data did not meet the requirements, a logarithmic, square root, or arcsine transformation was performed, followed by multiple comparisons using Tukey's HSD method. The NMDS using the Bray-Curtis distance was performed based on the individual numbers of soil animals to visualize the differences in communities among three revetment types, three distances, and two seasons. Permutational multivariate analysis of variance (PERMANOVA) with 1000 permutations was performed to test the effects of revetment type and distance gradient on the community structure of the soil fauna in each season. To select the best analysis for the associations between soil animal community and environmental factors-including soil water content, soil bulk density, soil pH, ammonium nitrogen, available phosphorus, available potassium, soil temperature, and air temperature-DCA was performed first to calculate the gradient length [48]. The longest gradients were >4 for all analyses except for the hygrophilous mesofauna in summer. Therefore, CCA was used for all soil animal groups in autumn, as well as the macrofauna and microarthropods in summer; RDA was used for the hygrophilous mesofauna in summer (the longest gradient was <3). Only statistically significant environmental variables were plotted in CCA and RDA. The ISA based on individual numbers was conducted using 999 permutations to determine whether there were some specific families that may reflect the characteristics of environmental conditions such as revetment type, distance gradient, and season. In the ISA, both the abundance and frequency of soil animals were taken into account, and statistics were provided for each indicator taxon [49].

Soil Fauna Abundance
Soil fauna abundance varied significantly among the different revetment types, distances, and seasons (Table 1, Figure 2). Specifically, compared with autumn, in summer, the abundance levels of the soil macrofauna, microarthropods, and hygrophilous mesofauna differed more significantly among the three revetment types and the three distances (Table 1). In most plots, soil fauna abundance was higher in the natural riverbank zone (located in rural areas with low urbanization levels) than in hard-type revetment zone (located in urban areas with high urbanization levels); moreover, soil fauna abundance in plots closer to the water flow was generally higher than that in plots further away from the water ( Figure 2, Table A3). Within the same revetment type and the same distance, densities of the entire soil fauna community and the hygrophilous mesofauna tended to be higher in summer than in autumn, whereas for macrofauna, the opposite trend was observed. This verifies the existence of seasonal effects and different responses of soil animal groups to environmental shifts (Table A3). Additionally, regarding soil fauna abundance in summer, a significant interaction between revetment type and distance gradient was found (Table 1). For instance, soil fauna abundance in the natural riverbank zone (NR) differed significantly throughout the distance gradient in summer, whereas in the mixed-type revetment zone (MR) and the hard-type revetment zone (HR), these differences were less pronounced (Table A4).

Diversity and Community Structure
Overall, soil animals from 191 families were collected and identified, including soil macrofauna, microarthropods, and hygrophilous mesofauna (Table A5). Generally, the richness of the entire soil fauna community was significantly higher in autumn, with a rich leaf litter layer (148 families), than in summer (119 families) (Table A5, Figure 3). Hygrophilous mesofauna richness was significantly higher in summer than in autumn ( Figure 3). In NR, soil fauna richness was highest, followed by MR, whereas it was lowest in HR (Table A5). Significant differences were found between natural riverbank zones and hard-type revetment zones ( Figure 3). Soil fauna richness was highest at 0 m (Table A5), although this difference was not statistically significant (Table 1, Figure 3). The interaction effect of revetment type and distance gradient on soil fauna richness was weaker than that on abundance, and was only observed for the entire soil fauna community and the macrofauna community in summer (Table 1) . Different lower-case letters indicate significant differences among revetment types or distances (p < 0.05). The plots of three revetment types and distances in autumn are not presented due to non-significance. NR = natural riverbank zone, MR = mixed-type revetment zone, HR = hard-type revetment zone; SUM = summer, AUT = autumn; 0, 5, and 10 represent the different distances (m) from the land-water boundary. It is the same below.  . Different lower-case letters indicate significant differences among revetment types or distances (p < 0.05). The plots of three revetment types and distances in autumn are not presented due to non-significance. NR = natural riverbank zone, MR = mixed-type revetment zone, HR = hard-type revetment zone; SUM = summer, AUT = autumn; 0, 5, and 10 represent the different distances (m) from the land-water boundary. It is the same below.
The DG indices provided specific diversity information about the soil fauna in different revetment types, distances, and seasons (Table 2). In general, in summer, the indices were more significant compared to autumn. Additionally, the indices for the entire soil fauna community, the macrofauna, and the hygrophilous mesofauna were significantly higher in more natural riverbank zones than in highly artificial ones. Furthermore, the effect of revetment type was stronger than that of the distance gradient. Taxonomic group number is the number of families, except insect larvae, which were at the order level. Each box plot displays the median (bold lines), the interquartile range (box), minimum and maximum values (whiskers), and outliers (circles). Different lower-case letters indicate significant differences among revetment types and distances (p < 0.05). The plots of three revetment types and distances in autumn are not presented due to non-significance.
The DG indices provided specific diversity information about the soil fauna in different revetment types, distances, and seasons ( Table 2). In general, in summer, the indices were more significant compared to autumn. Additionally, the indices for the entire soil fauna community, the macrofauna, and the hygrophilous mesofauna were significantly higher in more natural riverbank zones than in highly artificial ones. Furthermore, the effect of revetment type was stronger than that of the distance gradient.   Table 1. Regarding the macrofauna community in summer, NMDS separations were observed among different revetment types, with only a slight overlap between NR and MR ( Figure 4A). For the microarthropod community in summer, MR was separated from NR and HR, with some overlap between NR and HR ( Figure 4B). Regarding the hygrophilous mesofauna community in summer, NR separated from HR, but there were some overlaps between NR and MR, as well as between MR and HR ( Figure 4C). Furthermore, the NMDS separation of the hygrophilous mesofauna community was highly pronounced between summer and autumn, which was not the case for the macrofauna and microarthropods ( Figure 4D-F). Note: "ns" means no significant difference. Significant p values are shown in bold. α = 0.05. "+" means the value of revetment type and distance on the left is higher than that on the right, whereas "−" has the opposite meaning.
As shown via the NMDS, the soil fauna community showed more distinct separations among revetment types ( Figure 4A-C) than among distances ( Figure A1D-I), and more distinct separations were found in summer ( Figure 4A-C) than in autumn ( Figure  A1A-C); this is consistent with the data shown in Table 1. Regarding the macrofauna community in summer, NMDS separations were observed among different revetment types, with only a slight overlap between NR and MR ( Figure 4A). For the microarthropod community in summer, MR was separated from NR and HR, with some overlap between NR and HR ( Figure 4B). Regarding the hygrophilous mesofauna community in summer, NR separated from HR, but there were some overlaps between NR and MR, as well as between MR and HR ( Figure 4C). Furthermore, the NMDS separation of the hygrophilous mesofauna community was highly pronounced between summer and autumn, which was not the case for the macrofauna and microarthropods ( Figure 4D-F). Permutational multivariate analysis of variance (PERMANOVA) showed that the revetment type had significant effects on the soil fauna community structure, irrespective of the season, whereas distance only significantly affected the soil fauna community structure in summer (p < 0.05, Table 3).

Environmental Factors and Soil Fauna Communities
Canonical correspondence analysis (CCA) and redundancy analysis (RDA) were applied to evaluate the relationship between soil fauna and environmental factors (  (Table A6). The models for soil macrofauna, microarthropods, and hygrophilous mesofauna in summer and microarthropods in autumn were significant (p < 0.05, Table A6), whereas those of soil macrofauna and hygrophilous mesofauna in autumn were not significant (p > 0.05, Table A6).
Different soil animal groups responded differently to environmental factors. Temperature (air and soil) was the most important environmental factor in summer, regardless of the soil animal group ( Figure 5). Specifically, for macrofauna, besides AT and ST, soil pH, BD, and AK also had significant effects on the community in summer (p < 0.05, Figure 5A, Table A6). For microarthropods, only AT and ST were significant factors in summer (p < 0.05, Figure 5B, Table A6), whereas in autumn, the significant variables were soil pH, AP, and WC (p < 0.05, Figure 5D, Table A6). In the case of the hygrophilous mesofauna, besides AT and ST, the community showed a stronger response to the soil water content (WC) in summer because of the high rainfall levels and their strong dependence on water (p < 0.05, Figure 5C, Table A6).

Indicator Species
For 29 families of soil fauna, there were significant associations with different revetment types, distances, and seasons (Table A7), based on indicator species analysis (ISA). For the different revetment types, a total of seventeen indicator families were found in summer, of which eight were associated with natural riverbank types, eight with the mixed-type revetment, and only one with the hard-type revetment. However, in autumn, only eight indicator families were found: three in the natural riverbank type, four in the mixed-type revetment, and one in the hard-type revetment (Table A8). In contrast, six families showed significant associations with distance and were associated with 0 m sites in summer. In autumn, a total of three indicator families were found: one in the 0 m site and two in the 5 m site.

Indicator Species
For 29 families of soil fauna, there were significant associations with different revetment types, distances, and seasons (Table A7), based on indicator species analysis (ISA). For the different revetment types, a total of seventeen indicator families were found in summer, of which eight were associated with natural riverbank types, eight with the mixed-type revetment, and only one with the hard-type revetment. However, in autumn, only eight indicator families were found: three in the natural riverbank type, four in the mixed-type revetment, and one in the hard-type revetment (Table A8). In contrast, six families showed significant associations with distance and were associated with 0 m sites in summer. In autumn, a total of three indicator families were found: one in the 0 m site and two in the 5 m site.

Discussion
The temporal and spatial distribution of the soil animal community in riparian zones is a result of the interactions among revetment structure, vegetation composition, location on the riverbank, seasonal shift, and other environmental factors [50]. We hypothesized that the revetment type, which is correlated with the urbanization gradient, and the distance from the land-water boundary in different riparian landscapes induce changes in the communities of different soil animal groups by altering the hydrothermal conditions, shifting the vegetation composition, changing water availability, and varying the intensity of anthropogenic interferences.
As an important component of riparian landscapes, the type of the revetment along an urban-rural gradient significantly affected the soil fauna community in this study, which is in agreement with previous findings [2,51]. We also observed effects of distance from the water, although they were not as pronounced as those of revetment type. Different soil animal groups responded differently to changes in the riparian surroundings, which supports our second hypothesis. As expected, seasonal dynamics of the soil fauna community were also detected due to the different hydrothermal conditions.

Effect of Revetment Type on Soil Fauna
Our data demonstrate that the heterogeneity of microhabitats, resulting from differences in revetment types, triggered a shift in both the composition and distribution of the soil fauna, supporting our first hypothesis. Specifically, soil animals were most abundant in natural riverbank zones, and a high degree of artificiality had negative impacts on the soil fauna. The formation of riparian landscapes with different revetment types affects not only the hydrothermal conditions but also the vegetation composition on the riverbank, ultimately influencing the assemblages and activities of soil organisms through the redistribution of heat, moisture, and litter [14,52,53].
When hard-type revetments are used, such as concrete and masonry, for slope protection, the aquatic ecosystem and the riverbank ecosystem are separated; this hampers the diffusion of oxygen into the soil and the material exchange between the soil and other environmental compartments, with negative impacts on biological diversity, ecological balance, and environmental protection [13]. Flow regulation and canalization reduce the natural dynamics between the river and its banks [13] and drive changes in local habitat characteristics, thereby influencing the activities of soil dwellers [2,54].
Vegetation plays a critical role in riparian zones; for example, it maintains a suitable temperature and provides woody debris to create favorable habitats [55], and acts as a driver of soil biota composition and distribution through litter quality and quantity [56]. Apart from the complexity of vegetation, the ratio of native plant species, the size of the tree, and even the presence of certain tree species may favor soil fauna diversity at the habitat scale [2,51,57,58]. In the present study, plant diversity and the native plant ratio in the natural riverbank zone were higher than in the hard-type revetment zone (Table 1), providing more diverse foods to soil fauna; this partly contributed to the largest catch of soil fauna in the natural riverbank zone, consistent with previous findings [2,51].
Environmental changes are highly important factors that alter the structure of the soil fauna community [2,51]. Investigating how anthropogenic activities may modify the environment and affect the biome is crucial to developing reasonable revetment management [10]. In the present study, the soil fauna was strongly influenced by environmental factors such as temperature, soil moisture, soil bulk density, and pH. Water availability is a key factor that influences vegetation characteristics [59] and soil faunal distribution in riparian zones; compared with HR and MR, NR, with a higher water availability, promoted the abundance and diversity of the soil fauna, especially hygrophilous mesofauna taxa, which are more adapted to humid environments. This is in agreement with previous studies stating that natural or mixed-type revetments with better water permeability provide a relatively sustainable microhabitat and facilitate the survival of soil animals [52,53]. In addition, we found more indicator families in the natural riverbank zone than in the hard-type revetment zone, which might be, at least partly, a result of the larger species pool of the natural riverbank type. It also suggests that the hard-type revetment zone is not beneficial to the soil fauna, with fewer indicator families appearing in this habitat.
Overall, riparian landscapes with different revetments shaped the soil fauna communities through differences in water permeability, vegetation composition, and soil properties, which should be considered when retaining natural riparian habitats.

Effect of Distance on the Soil Fauna
Soil fauna abundance increased with decreasing distance from the land-water boundary (especially that of the microarthropods and the hygrophilous mesofauna). According to Paetzold [54], the arrival of species in a given habitat is mainly related to habitat features, such as habitat size, distance from sources, and surroundings; moreover, the influences decrease with increasing distance, which generates a distance gradient. In the gradient system, direct gradients (soil, vegetation, water, and heat gradients) represent the habitat conditions and directly influence species diversity, whereas indirect gradients (distance and landscape gradients) influence species diversity by changing the habitat conditions [50,54,60]. As both groundwater and soil moisture are partly recharged by rivers or streams, the distance from the river flow can be regarded as a proxy for water availability; hence, the influence of the river will weaken with increasing distance from the riverbank [54]. It is not surprising that the natural riverbank zone showed a more significant distance effect due to its higher water availability (Table A4). Compared with the microarthropods and the hygrophilous mesofauna, the abundance and richness of the macrofauna along the distance gradient were not obvious, whereas the hygrophilous mesofauna was more sensitive to distance; this supports our second hypothesis, which states that the different soil animal groups respond differently to environmental alterations. Macrofaunal groups, such as spiders and beetles, exhibit greater mobility and environmental tolerance than other smaller animals, which probably helps them survive under less favorable conditions [61]. Wasserstrom and Steinberger [62] found that mite diversity tended to be higher at more distant locations from the seashore and lower at closer locations in a coastal dune field in a Mediterranean ecosystem. The results showed that the gradient of environmental factors, which was generated by the distance from the land-water boundary, affected soil faunal survival, which is similar to our findings. Nematodes, a dominant group in the hygrophilous mesofauna, play an important role in soil ecological processes [24], and their abundance, community composition, and activities (i.e., locate, penetrate, hatch, and mate) can be significantly impacted by hydrothermal changes [63,64]; this is mainly because they are essentially aquatic animals that depend on the water film around soil particles for their development and movement [63,65,66]. Thus, in our study, the hygrophilous mesofauna community showed a more pronounced response to distance from the water than the other soil animal groups.

Effects of Seasonal Shifts on Soil Fauna
We observed significant differences in the abundance and richness of the soil fauna depending on the season (Figures 2 and 3). Moreover, different revetment types and distances often had different seasonal dynamics (Figures 4 and A1). Generally, the seasonal dynamics of soil animal groups are the results of the combined effects of environmental factors such as hydrothermal conditions, amount of plant litter, and soil properties [52,53,65]. In our study, both air temperature and precipitation peaked in summer, with drastic influences on soil temperature and moisture. Given the impacts of soil temperature and moisture on soil faunal feeding and development rate, climate change may have a strong impact on the abundance and community structure of the soil fauna [53,67]. According to a previous study, thermal tolerances may not only constrain where, but also when species can effectively forage for resources; this underscores the importance of climate change, as well as the heterogeneity of soil animal groups in their adaptabilities [68]. In our study, soil fauna abundance was higher in summer than in autumn, and nematodes, as the most abundant group, contributed most significantly to the distribution pattern, most likely because of the high rainfall levels in summer. However, soil fauna richness was higher in autumn than in summer, with a large contribution of macrofauna and microarthropods, which might be a result of the substantial leaf litter layer in autumn. According to Huang et al. [67], the leaf litter provides abundant food and suitable refuge for the soil fauna, and various soil animal groups, including some high-temperature sensitive species, can adapt to the mild environment; this explains the higher soil animal richness in autumn. Soil temperature and air temperature, differing in various riparian landscapes along an urban-rural gradient (Table A2) due to the urban heat island effect, significantly impacted macrofauna, microarthropod, and hygrophilous mesofauna communities in summer. In contrast, in autumn, with a more moderate climate, other soil properties such as soil pH, available potassium, and soil water content had more substantial effects ( Figure 5). Furthermore, soil fauna heterogeneity along the distance gradient was more evident in summer than in autumn. This might be explained as follows: On the one hand, rainfall was higher in summer, which caused an outbreak of the hygrophilous mesofauna. Additionally, the slight differences in soil moisture along the distance gradient caused the significant differences in hygrophilous mesofauna abundance [1,63]. On the other hand, in autumn, the leaf litter layer is thicker than in summer, restricting water evaporation and increasing organic debris input, which homogenized the distance gradient plots [52].

Effect of Urbanization on Soil Fauna
Globally, urbanization is accelerating, resulting in the destruction of natural riverbanks and, consequently, seriously affecting the ecological functions of riparian zones [7,8]. Similar to most studies on the impacts of urbanization on biodiversity [69,70], we detected an overall negative effect on the abundance and richness of the soil animal community at high urbanization levels. Riparian landscapes are diverse along the urban-suburban-rural gradient, representing a series of distinct subsets of anthropogenic disturbance; this may reduce regional species pools or prevent some species from colonizing a site [71,72], affecting biodiversity and regional ecological processes [73]. Our results show that natural riparian zones, located in rural areas with low urbanization levels, suffered from human interference (measured using human traffic density) less than hard-type riparian zones in urban areas. In our study, the differences in revetment structure, human density, air or soil temperature, and vegetation composition were important embodiments of urbanization levels and were closely related to soil fauna diversity. The most obvious hydrological change associated with urbanization is the engineering of stream channels, in which natural features are replaced by concrete channels and streambank stabilization efforts designed to resist increased flood flows [7]. In such an approach, the riparian vegetation is replaced by an impervious or less-permeable surface, which alters the rates and pathways of water movement into, through, and out of ecosystems [7]. Walsh et al. [74] described the "urban stream syndrome" and suggested that urban riparian zones may exhibit reduced water and nutrient infiltration because natural terrestrial pathways are often bypassed due to stormwater drainage networks. Similar to soil physicochemical properties, vegetation composition is assumed to be a good predictor of the soil biota community [75]. Williams et al. [76] showed that exotic and cultivated plants were overrepresented in the species pool of urbanized regions. In our study, the proportions of exotic and cultivated plants were higher in urban areas, whereas vegetation diversity was higher in rural areas (Table A2), which is consistent with previous research. Lorenzo et al. [77] reported that numerous non-native tree species support relatively lower numbers of arthropod species than native trees, most likely because they excrete allelopathic compounds that harm native plants and some soil organisms. Overall, the occurrence of exotic plant species alters ecosystem structure and function, community assemblage, and native species interactions [78]. In this context, it is not surprising that soil fauna abundance and richness in artificial riparian landscapes with higher urbanization levels were lower than in natural riparian landscapes.

Conclusions
This work provides specific information about the changes in soil animal community traits in riparian landscapes across an urban-rural gradient. Compared with artificial and semi-natural riparian landscapes, the natural riparian landscape in rural areas had the highest soil fauna diversity because of the better permeability of the revetment structure, a less significant urban heat island effect, higher plant diversity, a higher ratio of native vegetation, and lower anthropogenic disturbance. Different soil animal groups responded differently to the revetment types, locations on the riverbank, and seasonal changes. Hygrophilous mesofauna taxa, which are adapted to humid habitats, appeared more sensitive to the environmental shift in riparian landscapes, especially the hydrological conditions. We observed interactions between revetment type and distance from the water flow, suggesting that the distance effect was stronger in natural riparian landscape with a permeable revetment type.
Overall, the data presented herein demonstrate that protecting and retaining permeable revetment and native vegetation are beneficial to soil animal groups and, consequently, to riparian soil ecosystem processes and functions. With the acceleration of urbanization, natural riparian landscapes are subjected to substantial changes, frequently resulting in biodiversity loss and ecosystem degradation. Balancing ecological functions and flood control practicability, as well as protecting or restoring riparian ecosystem services under the urbanization scenario, are challenging for policymakers, urban planners, and landscape designers. Controlled experiments and long-term studies are, therefore, needed to further reveal the biological mechanisms of nutrient cycling and fertility maintenance in urban riparian zones.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy. Note: pH = soil pH value; SBD = soil bulk density; AN = ammonium N; AP = available P; AK = available K; WC = soil water content; AT = air temperature; ST = soil temperature. Different lower-case letters indicate significant differences among revetment types or distances in the same season, and different capital letters indicate significant differences between two seasons in the same revetment type or distance; α = 0.05. Note: Different lower-case letters indicate significant differences among revetment types or distances in the same season, and different capital letters indicate significant differences between the two seasons in the same revetment type or distance; α = 0.05.

Appendix D
Arachnida Acariformes Oppiidae + + + + + + + + + + + +    Note: "+" indicates that there are soil animals of this family in the sampling plot; "−" indicates that there are no soil animals of this family in the sampling plot.  Appendix G