The study area was situated in Ningbo City, Zhejiang Province, China, spanning 120°55′–122°16′E and 28°51′–30°33′N. Ningbo is in the central part of China’s coastline; its topography is characterized by higher elevations in the southwestern region and lower elevations in the northeastern sector. Urban zones have altitudes ranging from 4 to 5.8 m, whereas suburban ones from 3.6 to 4 m. The region has a subtropical monsoon climate characterized by mild and humid conditions during the four seasons. The city’s mean annual temperature is 17.5°C, with the highest recorded temperature in July reaching 29.1°C and the lowest in January decreasing to 5.8°C. The average annual precipitation is 1,530 mm, of which 60% falls between May and September (Gao et al., 2023). The total land area of Ningbo is 9,816 km². By the end of 2023, the city’s permanent population was 9.697 million, of which 79.9% was urban (http://www.ningbo.gov.cn/col/col1229099787/index.html, 03 September 2024).

The Zhejiang Province had reclaimed a total of 108,760 ha of coastal land by 2010, with Ningbo City’s reclamation area and wetlands being the most extensive reclaimed (37.8%) and wetland (47.97%) areas in the Province (Yang et al., 2018). The study was performed in the Dasong Enclosure and adjacent farmlands in Zhanqi Town, Yinzhou District of Ningbo City (121.80°–121.89°N, 29.68°–29.7°E). Formerly, this area was an open tidal flat. It was enclosed by seawalls (dikes) in 2012 and featured a typical DDSM habitat. Subsequently, it became an enclosed area named the Dasong Enclosure. The latter was used for agriculture, allowing a few farmers to breed fish, shrimp, shellfish, ducks, and sheep. However, fish breeding and poultry rearing were prohibited in the Dasong Enclosure in 2019, allowing plants that prefer moist habitats, such as Phragmites australis Trinius, Spartina alterniflora Loisel, and Arundo donax Linnaeus, to grow and dominate the Dasong Enclosure (unpublished data from Lihu Xiong). Based on the results of a field investigation in September 2022, in contrast to the outside tidal flats, the Dasong Enclosure presented a reduction in soil moisture, a decline in salinity, and a notable increase in hydrophytic vegetation (unpublished data from Lihu Xiong). In contrast to adjacent crop fields with rice and wheat rotation models, the Dasong Enclosure exhibited a higher plant richness, composition, and coverage, as well as greater bird and mammal diversity (unpublished data from Lihu Xiong). However, the soil animal community remains largely unexplored in the DDSM habitats within the Dasong Enclosure and its adjacent farmlands.

DDSM and wheat farmland sites selected for the present study (Supplementary Table S1) were established to investigate the biodiversity of the soil collembolan community within the northern region of the Dasong Enclosure (NE, 121.80°–121.89°N, 29.69°–29.79°E) and the southern region of the Dasong Enclosure (SE, 121.84°–121.86°N, 29.68°–29.71°E). The NE and SE habitats were originally separated by a river. After the establishment of the Dasong Enclosure, the river channel was modified. As a result, the NE and SE habitats are now separated by a 10-m-wide partially man-made river, and connected only by a 5-m-wide and 15-m-long stone bridge (Supplementary Figure S1).

Wheat farmland (WF) sites were situated in the agricultural region of Zhanqi Town, which have been reclaimed for agricultural land use for more than 40 years. The WF sites were distant approximately 3.2 km from the Dasong Enclosure area. Ten plots, each measuring 10 × 10 m, were established with a minimum separation of 50 m from one another in the NE, SE, and WF habitats (Supplementary Table S1; Supplementary Figure S1). Soil samples of Collembola were collected from these 30 plots.

Within each plot, three soil samples were collected at a depth of 0–10 cm using an auger with a 7-cm inner diameter. Three samples were randomly collected at 3-m intervals. Subsequently, soil samples from the 0–10 cm were collected to measure soil moisture within a 2-m diameter circle around the center of each plot. Soil moisture was quantified using a gravimetric method. The height (cm) of 10 plants was randomly measured within a 2-m circle around each soil sample in each plot. Plant coverage, ranging from 0 to 1, was measured for each plot. Field collection was conducted on 16 April 2023. Ninety soil samples were collected: 3 soil samples/plot × 10 plots/habitat × 3 habitats.

Collembola were extracted from each soil sample by 10 days using dry Tullgren funnels, preserved in 95% ethanol, and stored at −20°C for further identification. Collembola were identified to the species level according to relevant publications (Christiansen and Bellinger, 1998; Potapov, 2021; Xie et al., 2019; Yin et al., 1998) using a set of stereomicroscopes (Olympus SZX16, Tokyo, Japan; equipment sourced from China; and Nikon Eclipse 80i; Shanghai, China).

All data were analyzed using R 4.4.1 (Team, 2024). The full list of species and their corresponding abundances in the NE, SE, and WF habitats are presented in Supplementary Table S2. Species richness (number of species) and abundance (number of individual species) were used to describe the quantitative characteristics of each collembolan community. Consistent with the quantitative classification criteria outlined in the literature (Wei et al., 2022), a dominant species was defined as presenting an abundance of 10% or more within the community. Common species represented 1%–10% of overall abundance. In contrast, species classified as rare amounted to less than 1% of the community.

To evaluate the diversity of the collembolan community in each habitat, Shannon and Simpson indices were calculated by the “diversity” function from the vegan package (Team, 2024). The Shannon index (Shannon, 1948) measures both richness and evenness. Its usual values vary around 1.5–3.5, with values greater than 3.5 indicating that habitat is highly important for biodiversity (Magurran, 2004). The Simpson index (Simpson, 1949) measures evenness, and varies from 0 to 1, with values closer to 1 indicating better evenness among species.

An extrapolated species diversity analysis method was employed to determine whether the field survey of each collembolan community was sufficient (Chao et al., 2014). This method reasonably extrapolates the observed species richness, that is, the number of species, based on existing trends, thereby obtaining corresponding theoretical values. The values were calculated using the “iNEXT” function from the iNEXT package (Hsieh et al., 2024). This study used observed abundances to calculate species richness for rarefaction and extrapolated samples. Rarefaction/extrapolated curves were plottted (Colwell et al., 2012).

To verify Hypothesis 1, a non-parametric statistical test, the Kruskal–Wallis test, was employed to assess variations in the richness, abundance, Shannon, and Simpson indices of collembolan communities, as well as the abundance of individual species across NE, SE, and WF habitats. This approach was necessitated by the noncompliance of these variables with the assumptions of normality and homogeneity of variance. Post-hoc multiple comparisons utilizing pairwise Wilcoxon rank-sum tests were conducted to delineate specific habitats exhibiting significant differences. The Holm method was used to adjust the p-values to control for the error rates of multiple comparisons. The “shapiro.test” function from the stats package was used to assess the normality of the variables, while the “leveneTest” function from the car package was employed to evaluate the homogeneity of variance (Fox and Weisberg, 2019). The “kruskal.test” function from the stat package was employed to conduct the Kruskal–Wallis test, while the “pairwise.wilcox.test” function from the stat package was used to conduct pairwise Wilcoxon rank-sum tests (Team, 2024).

To test Hypothesis 2, the composition of soil collembolan communities in NE, SE, and WF habitats was analyzed using non-metric multi-dimensional scaling (NMDS) based on the Bray–Curtis dissimilarity index. NMDS is a sorting method that simplifies multi-dimensional space objects into lower-dimensional spaces while retaining the original relationships between objects. In cases with many samples and species, the NMDS model can more accurately reflect the numerical sorting information of the distance matrix and is therefore considered the most robust non-restrictive sorting method (Minchin, 1987). This visualization was achieved through the “metaMDS” function from the vegan package (Oksanen et al., 2024). Only the species present in a minimum of three sampling plots were considered (Xie et al., 2021). This exclusion was based on the observation that species found in only one or two plots exhibited consistently low abundance.

To test Hypothesis 3, environmental factors, including soil moisture, plant height, and plant coverage, were used (Supplementary Table S1). Spearman’s correlation was used to evaluate the relationships between species richness, abundance of collembolan communities, and environmental factors. Spearman correlation was conducted using the “cor” function from the corrplot package (Wei and Simko, 2021). General linear regression was used to evaluate the importance of environmental factors affecting species richness and abundance of collembolan communities in each habitat using the “lm” function from the stats package (Team, 2024). Distance-based redundancy analysis (DB-RDA) (Legendre and Legendre, 2012) was used to identify environmental factors related to collembolan community composition in each habitat. Moreover, hierarchical partitioning (HP) (Lai et al., 2022) was introduced to DB-RDA to describe the relative contributions of individual environmental factors by the “rdacca.hp” function from the rdacca.hp package (Lai et al., 2022). To eliminate the influence of extreme values and reduce the weight of high-abundance species, community data in each habitat were standardized using Hellinger transformation by the “decostand” function from the vegan package (Oksanen et al., 2024). The variation explained by environmental variables was assessed using adjusted R ² values. The permutation test of hierarchical partitioning for canonical analysis with 999 permutations was performed to evaluate significance using the “permu.hp” function in the rdacca.hp package (Lai et al., 2022).

Overall, 43 collembolan species were identified among 2,405 individuals. Species richness was 16, 21, and 26 in NE, SE, and WF habitats, respectively (Supplementary Table S2). Species richness was significantly higher in the WF habitat than in the NE habitat (Figure 1A). The abundance of collembolan communities in NE, SE, and WF habitats was 471, 1,126, and 808, respectively. No significant differences in abundance of collembolan communities existed among the three habitats (Figure 1B).

Based on the current field survey, species rarefaction curves in all habitats tended to level off gradually. When the number of Collembola collected in NE, SE, and WF habitats reached 942, 2252, and 1,616 individuals, respectively, the extrapolation curves suggested that the corresponding richness was expected to be 18.06, 23.52, and 27.47. These values would have represented an increase of 12.88%, 10.71%, and 5.65%, respectively, in species richness (Figure 2).

The Shannon indices of the collembolan community in NE, SE, and WF habitats were 1.147, 1.755, and 2.141, respectively, with the latter two being significantly higher (p < 0.05) than the first one (Figure 3A). The Simpson indices of the collembolan community in NE, SE, and WF habitats were 0.516, 0.724, and 0.808, respectively, with the latter two being again significantly higher (p < 0.05) than the first one (Figure 3B).

Bionychiurus changbaiensis Sun and Wu (2012) and F. ozaena sp1 Yosii (1977) were prevalent in the WF habitat but rare in the NE and SE habitats (Figure 4; Supplementary Table S2). The abundance of B. changbaiensis was notably higher in the WF habitat than in NE (p < 0.001) and SE (p < 0.001) habitats (Figure 5A). The abundance of F. ozaena sp1 was significantly higher in the WF habitat than in the NE (p < 0.05) habitat (Figure 5D). Sinella curviseta Brook (1882) and Ceratophysella sp2 were dominant in the SE habitat but rare in NE and WF habitats (Figure 4; Supplementary Table S2). S. curviseta was significantly more abundance in the SE habitat than in NE (p < 0.001) and WF (p < 0.001) habitats (Figure 5E). Instead, no significant differences in the abundances of Ceratophysella sp2 were observed among the three habitats (Figure 5B). Sinella sp was uniquely prevalent in the NE habitat, where its abundance was significantly higher than that in both SE (p < 0.05) and WF (p < 0.05) habitats (Figure 5F). Instead, Desoria choi Lee (1977) was dominant across all habitats, and its abundance in the WF habitat differed significantly from that in NE (p < 0.001) and SE (p < 0.001) habitats (Figure 5C).

The abundances of Arrhopalites sp1 (Supplementary Figure S2A), Ceratophysella sp1 (Supplementary Figure S2E), Parisotoma ekmani Fjellberg (1977) (Supplementary Figure S2G), and Sinella sp5 (Supplementary Figure S2H) were notably higher in the WF habitat than in NE and SE habitats (p < 0.05, p < 0.01, p < 0.001, respectively). The abundance of Ceratophysella skarzynskii Weiner (1996) (Supplementary Figure S2B, p < 0.05) and Isotoma pinnata Fabricius (1781) (Supplementary Figure S2F, p < 0.001) in the SE habitat significantly exceeded that in NE and WF habitats, respectively. The abundance of Ceratophysella sp1 in the SE (p < 0.05) and WF (p < 0.01) habitats was notably higher than that in the NE habitat (Supplementary Figure S2C). Additionally, the abundance of Desoria sp12 was significantly higher in the NE habitat than in SE (p < 0.001) and WF (p < 0.001) habitats (Supplementary Figure S2D). No significant differences in abundance were observed among the three habitats for the other species.

According to NMDS results, the distribution of collembolan species in NE and SE habitats was relatively dispersed; whereas in the WF habitat, it was more concentrated. Collembolan communities in the three habitats partially overlapped near the middle, indicating a small overall similarity in species composition. The collembolan communities in NE and SE habitats showed a relatively high degree of overlap, suggesting that species composition in these habitats was more similar than that in the WF habitat. The second axis separated the collembolan community in the WF habitat from that in the other two habitats (Figure 6).

Species richness correlated significantly with soil moisture (p < 0.05) and plant coverage (p < 0.01). Abundance in the collembolan community correlated significantly with soil moisture (p < 0.01) in the NE habitat according to Spearman’s correlation results (Figure 7A). Instead, species richness and abundance were not significantly correlated with environmental factors in SE and WF habitats (Figures 7B,C).

Based on the outcomes of general linear regression analysis, only species richness showed a significant increase with enhanced plant coverage in the NE habitat (R ² = 0.157, p < 0.05) (Figure 8A). Conversely, no other significant associations were identified between species richness, abundance, and the environmental factors examined in each habitat (Figures 8B–D).

The three environmental factors collectively explained 11.6% of total variance in collembolan community composition in the NE habitat (adjusted R ² = 0.116). Plant coverage contributed significantly to collembolan community composition (adjusted R ² = 0.794, p < 0.01). Plant coverage and soil moisture exerted a high synergistic effect on collembolan community composition (adjusted R ² = 0.233) owing to larger collinearity between the two factors. However, these three environmental factors could not explain the relatively large residual (88.4%) (Figure 9A).

Environmental factors collectively explained 8.7% of total variance in collembolan community composition in the SE habitat (adjusted R ² = 0.087). Plant height contributed significantly to collembolan community composition (adjusted R ² = 0.056, p < 0.05). Plant height, soil moisture, and plant coverage showed a high synergistic effect on collembolan community composition (adjusted R ² = 0.015). Nevertheless, a substantial proportion of the residue (91.3%) was unaccounted for by plant and soil variables (Figure 9B).

Environmental factors collectively explained 2.9% of total variance in collembolan community composition in the WF habitat (adjusted R ² = 0.029), although none of them contributed significantly. Overall, 97.1% of variance could not be explained by these environmental factors (Figure 9C).

The first part of Hypothesis 1 was confirmed: the collembolan community’s species richness was lower in the DDSM habitat than in wheat farmland. In the NE habitat, species richness was significantly lower than in wheat farmland. Contrary to Hypothesis 1, the average abundance of collembolan communities in the DDSM habitat was almost equivalent to that in wheat farmland. Similarly, Yahya et al. (2020) found that collembolan abundance was lower in wetlands than in wheat-rice farmland, but the difference was not significant.

Overall, species richness and abundance exhibited different responses in the two habitats. The results of this study partially align with the findings from a study on reclaimed coastal areas along the west coast of the Yellow Sea in China, which detected higher taxonomic richness and abundance of soil macrofauna in wheat farmlands and lower values in unutilized wetlands (Ge et al., 2014b). The expected outcome of species richness could be due to frequent waterlogging, anaerobic conditions, and a relative lack of microorganisms in the DDSM habitat of the study area (Sławska, 2000). For example, the present study area was found to harbor significantly fewer soil microorganisms in the halophyte plants Suaeda salsa Linnaeus, S. alterniflora Loisel, and Phragmites communis Trinius growing in coastal wetlands than in adjacent rice fields (Li and Shen, 2011). Additionally, specific plot conditions in DDSM habitats with low plant richness and coverage could contribute to lower species richness. Plots NE9, NE10, SE4, and SE8 showed lower plant coverage, especially the NE9 and NE10 plots, which were located in the drainage engineering area and frequently subject to disturbances from workers, vehicles, and construction activities (Supplementary Figure S3). As vegetation cover and land use affect the taxonomic richness of soil fauna in coastal wetlands and farmland (Ge et al., 2014b), we speculated that poor conditions with low vegetation cover or bare land in certain areas of DDSM habitats lowered species richness of the collembolan community. Furthermore, species richness of the collembolan community was significantly lower in the NE habitat than in the SE habitat. This coincided with lower elevation of the NE habitat (0.04 ± 0.9 m) compared with the SE habitat (1.08 ± 1.71 m) (unpublished data from Lihu Xiong), suggesting that the soil drainage process is likely slower in the NE habitat. Consequently, variations in elevation may have led to more extensive and frequent waterlogging in the NE habitat than in the SE habitat (Sławska, 2000). Elevation and drainage processes within DDSM habitats can influence soil salinity, waterlogging conditions, and microorganisms (Li and Shen, 2011). Indeed, soil salinity (Owojori et al., 2009), waterlogging environments (Sławska, 2000), and soil microbial biomass (Filser et al., 2002) have been shown to affect abundance, survival, and juvenile production of collembolan communities. Consequently, we propose that changes in salinity, waterlogging environments, and microbial communities, may indirectly affect species richness across DDSM habitats.

The significantly higher abundance of dominant species in the DDSM habitat may be attributed to unexpected results (Wiwatwitaya and Takeda, 2005). D. choi was dominant in NE and SE habitats, representing 13.01% and 21.58% of total abundance among collembolan communities in this study (combined collected Collembola), respectively. The collembolan community tends to be dominated by a handful of species across various habitats, thereby contributing to abundance and composition of the community (Sousaa et al., 2004; Wiwatwitaya and Takeda, 2005).

Species composition of the collembolan community differed between the DDSM habitat and wheat farmland, confirming Hypothesis 2. Several collembolan species, inhabit wetlands, and many of them are found exclusively in these habitats (Sławska, 2000). A similar phenomenon was observed in this study. C. skarzynskii, Desoria sp12, I. pinnata, and Sinella sp were only present in the DDSM habitat, in which they were also dominant. Accordingly, they might be specialized for such areas. Microtubercules and hair are key adaptations enabling Collembola to thrive in moist environments (Marx et al., 2012). Both C. skarzynskii and I. pinnata exhibited these features, which may explain their dominance and exclusivity in DDSM habitats. Arrhopalites sp1, Heteraphorura seolagensis, P. ekmani, and Sinella sp5 were exclusively distributed in wheat farmland and were significantly dominated in this habitat in the present study. Previous studies have reported that Collembola from the genera Ceratophysella, Isotomurus, Ballistura, Arrhopalites, and Sminthurides were stenotypic species in mire habitats (Sławska, 2000); whereas Isotogastrura mucrospatulata Palacios-Vargas, Lima and Zeppelini (2015) was endemic in tidal flats (Lima et al., 2023). Similarly, we found that certain Collembola from the genera Ceratophysella, Desoria, Sinella, and Isotoma were preferentially distributed in DDSM habitats. We also observed that certain species from the genera Arrhopalites, Heteraphorura, and Parisotoma were particularly well-adapted to the cultivated environment of wheat farmland. Therefore, we suggest that these species, which dominate and are exclusive to DDSM and wheat farmland, could serve as potential indicators for restoring and protecting soil biodiversity in reclaimed coastal wetlands.

Some Collembola were found in both DDSM and wheat farmland habitats, but showed habitat preferences. S. curviseta, Ceratophysella sp1, B. changbaiensis, and F. ozaena sp1 existed in both habitats; however, the first two were significantly dominant in DDSM, whereas the last two were significantly dominant in wheat farmland. D. choi was dominant in both the DDSM habitat and wheat farmland, but was significantly more abundant in the former. As Desoria mulyeongariensis prefers wetland habitats in Korea (Lim and Park, 2011), we speculated that D. choi might also prefer and thrive in moist habitats in coastal marshes. Therefore, although distributed in both habitats, certain collembolan species from the genera Sinella, Ceratophysella, and Desoria preferred DDSM habitats, whereas the genera Bionychiurus and Folsomia favored farmland habitats. A similar phenomenon was observed by Chang et al. (2013), who found that species from the genus Folsomia were preferentially distributed in rice and soybean farmlands, with marsh soils and located in temperate zones. This suggests that habitat preferences by collembolan species may be driven by specific environmental factors, such as soil moisture, organic content, and the availability of food resources (Potapov et al., 2020). These factors could play a crucial role in shaping the distribution of these species, with certain genera thriving in the DDSM habitats compared to the more disturbed conditions of farmland. Further research is needed to examine the mechanisms guiding habitat preferences.

The Shannon and Simpson indices of collembolan communities were significantly higher in wheat farmland than in the DDSM habitat, indicating a clear separation between the two areas regarding species diversity and evenness. Additionally, a significant difference in the Shannon and Simpson indices was observed between the NE and SE habitats. The higher species richness and the greater dominance of certain species in the SE habitat, compared to the NE habitat, may have contributed to this difference (Shannon, 1948; Simpson, 1949). Therefore, the biodiversity of the collembolan community across DDSM habitats warrants further attention. Whereas, the composition of the collembolan community was similar in the two reclaimed DDSM habitats, according to the outcomes of NMDS, its composition in wheat farmland differed from that of the DDSM habitat. Xie et al. (2021) found that certain collembolan species contributed to the structure and pattern of different habitats. In the present study, Arrhopalites sp1, Desoria sp7, Folsomia inoculata sp1, Folsomia octoculata, H. seolagensis, P. ekmani, and Sinella sp5 only existed in wheat farmland; therefore, these seven species contributed to the significant separation between wheat farmland and DDSM habitat.

Our findings partially support Hypothesis 3. Plant parameters significantly affect the composition of collembolan communities in the DDSM habitat; however, contrary to our expectations, soil does not significantly influence the collembolan community in wheat farmland.

In DDSM habitats, plant height and coverage were important environmental factors affecting the composition of the collembolan community. Collembola are detritivores or fungivores, whose abundance can be quite high in some wetlands (Batzer and Wu, 2020), and are affected by plant variables. Plant species richness, biomass, coverage, and vegetation type contribute to collembolan community composition because diverse plant communities provide a wide range of food resources (Maceda-Veiga et al., 2016; Zhang et al., 2023). Collembola are particularly abundant in wetlands with vegetation and tend to accumulate in areas where plant growth and leaf litter accumulate (Giordano et al., 2014). These microhabitats offer a wealth of food sources and shelter from predators and environmental pressure in wetlands (Giordano et al., 2014; Maceda-Veiga et al., 2016). Similarly, the significant influence of plant height and coverage on collembolan communities observed in this study may be due to the greater availability of food resources and shelter in the DDSM habitat. Although soil moisture exhibited a significant positive correlation with species richness and abundance of the collembolan community (Figure 7), its individual influence was not pronounced based on the DB-RDA results. This study revealed considerable collinearity between soil moisture and factors, such as plant coverage and height. Consequently, the observed positive correlations among soil moisture, species richness, and abundance may be attributed to the substantial contributions of plant coverage and plant height (Figure 8). We did not detect a significant contribution of soil moisture to collembolan community composition in the DDSM habitat; however, other studies did. The lengths of the antenna and body of Collembola were negatively associated with soil water content, which could be explained by there being more opportunities for these organisms to migrate on the surface of a coastal mudflat when it was submerged in water (Li et al., 2023). These findings suggest that the relationship between the collembolan community and soil moisture is case-specific.

Collembola respond to plant and soil variables in farmlands (Chang et al., 2013; Dou et al., 2019; Xie et al., 2021); however, we did not detect significant relationships between collembolan communities and these variables. Studies have reported positive correlations between plant variables and collembolan communities in wheat farmlands. For example, wheat roots (Becker et al., 2001) and crop identity (Chauvat et al., 2014) are key determinants of collembolan community composition in wheat fields. However, other studies have reported different findings; wheat litter (Sereda et al., 2015) and crop species (Bokova et al., 2023) do not significantly influence the collembolan community in wheat farmlands. Additionally, some studies have reported the significant influence of soil moisture on the collembolan community in wheat fields (Asif et al., 2016; Salmon et al., 2021). Suitable soil moisture may increase the accessibility of microorganisms to food, thereby affecting the collembolan community (Meyer et al., 2021). However, some studies have only detected minor or insignificant roles of soil moisture on collembolan composition (Shaki and Ahmed, 2015) and suggested that Collembola tended to be abundant in wetter soil and soil with sufficient amounts of nutrients (Shaki and Ahmed, 2015; Steinberger et al., 1984). In this study, insufficient soil nutrients in the reclaimed farmland may have contributed to the non-significant effect of soil moisture. Therefore, further studies combining soil moisture and nutrients are needed.

Large residuals in ether the DDSM habitat or wheat farmland could not be explained by plant and soil variables. Besides variables measured in this study, other parameters, such as plant genetic diversity, plant roots (Chateil et al., 2013), soil pH or organic matter (Becker et al., 2001), particle size (Salmon et al., 2021), climate [e.g., flooding (Krediet et al., 2023) or warming (Zhang et al., 2023)], topography, and management (e.g., tillage, fertilizer) (Chang et al., 2013) or dikes (Ge et al., 2017) have also been reported as key determinants of collembolan composition in wetlands and wheat farmlands. Therefore, more plant and soil variables and their interactions should be considered in future studies.

Although reclamation generally leads to a loss of marine biodiversity, this study confirmed that reclamation of tidal flats to DDSM habitats could maintain a certain richness and abundance of the soil collembolan community after 11 years. Therefore, ensuring high biodiversity in DDSM habitats and reclaimed farmlands may help maintain local biodiversity, consistent with other studies on reclaimed forests and farmlands (Ge et al., 2014a). Overall, the results of this study provide basic knowledge of the soil faunal biodiversity in DDSM habitats and their targeted farmlands in subtropical areas. Whether DDSM habitats are restored to their original tidal flats or left to succeed to their targeted agricultural farmland, they can maintain a certain unique and general biodiversity of soil fauna, providing a “biodiversity resource.” Future investigations ought to reveal the function of collembolan communities in processes of restoration of biodiversity and the conservation of reclaimed DDSM and agricultural areas. These findings will contribute to the development of restoration strategies for reclaimed coastal areas in southeastern China.