Discrepant Effects of Flooding on Assembly Processes of Abundant and Rare Communities in Riparian Soils

Numerous rare species coexist with a few abundant species in microbial communities and together play an essential role in riparian ecosystems. Relatively little is understood, however, about the nature of assembly processes of these communities and how they respond to a fluctuating environment. In this study, drivers controlling the assembly of abundant and rare subcommunities for bacteria and archaea in a riparian zone were determined, and their resulting patterns on these processes were analyzed. Abundant and rare bacteria and archaea showed a consistent variation in the community structure along the riparian elevation gradient, which was closely associated with flooding frequency. The community assembly of abundant bacteria was not affected by any measured environmental variables, while soil moisture and ratio of submerged time to exposed time were the two most decisive factors determining rare bacterial community. Assembly of abundant archaeal community was also determined by these two factors, whereas rare archaea was significantly associated with soil carbon–nitrogen ratio and total carbon content. The assembly process of abundant and rare bacterial subcommunities was driven respectively by dispersal limitation and variable selection. Undominated processes and dispersal limitation dominated the assembly of abundant archaea, whereas homogeneous selection primarily driven rare archaea. Flooding may therefore play a crucial role in determining the community assembly processes by imposing disturbances and shaping soil niches. Overall, this study reveals the assembly patterns of abundant and rare communities in the riparian zone and provides further insight into the importance of their respective roles in maintaining a stable ecosystem during times of environmental perturbations.


Introduction
Soils in aquatic-terrestrial ecotones (i.e., riparian zones around rivers, lakes, streams) are regularly influenced by intermittent flooding triggered by natural rainstorms or artificial hydraulic regimes [1]. The resulting alternation between submersion and drying regulates a series of biochemical processes in soil ecosystems, such as nitrogen transformation [2], greenhouse gas emissions [3], and phosphorous release [4]. Moreover, consequent changes in soil moisture content can greatly influence microbial activities, and alter both the soil structure [5] and nutrient availability [6]. Flooding therefore strongly shapes community structures of soil microorganisms. Microorganisms are an essential part of soil, which are crucial for the decomposition of organic matter, the cycling of soil nutrients, material and energy, and the fixation of carbon and nitrogen [7]. Soil microorganisms thus play an irreplaceable role in maintaining ecological functioning and stability.
Different soil microbial communities typically exhibit an unbalanced distribution, which consists of vast numbers of rare taxa and a limited number of abundant taxa [8]. The highly abundant taxa constitute the main component of a given microbial community and are deemed most important in maintaining fundamental functions of the ecosystem, such as biomass generation and carbon cycle [8,9]. Rare taxa commonly known as the "rare biosphere" have, however, been receiving increasing attention from the scientific community [8, [10][11][12][13]. They serve as "ecological insurance" for the microbial community by supporting crucial ecosystem processes [10]. For example, rare taxa had a vital role in regulating denitrification, nitrification, and sulfate reduction [14][15][16]. Meanwhile, rare taxa are comprised of more metabolically active microorganisms that act as an endless pool of genetic diversity, and that achieve disproportionately diverse functions in consideration of their low gene abundance [10]. Moreover, rare taxa aid in essential functions related to nutrient cycling by enhancing the functionality of abundant taxa [13]. Recent studies conducted on diverse ecosystems showed that abundant and rare microorganisms generally exhibit divergent community distribution and functional profiles [17][18][19]. Hence, uncovering the fundamental processes of microbial assembly in abundant and rare subcommunities is crucial for the deeper understanding of the microbe-driven ecosystem functioning.
In the assembly processes of the microbial community, the relative contribution of stochastic processes (such as homogenizing dispersal, dispersal limitation, and undominated processes) and deterministic processes (such as homogeneous and variable selection) are frequently evaluated [10,20,21]. Based on the null model, homogenizing dispersal means uniform species distribution caused by high dispersal rates; dispersal limitation, conversely, represents high species turnover due to low dispersal rates [22]. Homogeneous selection indicates that environmental filters without spatiotemporal variations are the main drivers for community turnover; variable selection suggests the role of changing environmental conditions on community turnover; while the term of undominated processes is a generalization that community turnover is governed by ecological drift and inappreciable dispersal and selection effects [22]. A scarce number of studies investigating community assembly processes of abundant and rare taxa currently exist, however. It was previously found that stochastic and deterministic processes primarily drive the community assembly of abundant and rare taxa, respectively [18,23,24]. While opposing patterns have been observed for bacteria in grassland soils, oil-contaminated soils, as well as in rivers and reservoirs [9,25,26]. These studies suggest that community assembly processes often differ between habitats for ecologically diverse taxa [10]. Here, environmental factors can significantly regulate the relative importance of deterministic and stochastic processes in community assembly. A previous study showed that the extent to which homogeneous selection influenced the community assembly of abundant bacteria increases with increased soil pH [27]. In addition, augmentations in dissolved oxygen have been observed to enhance the importance of stochastic processes on the community assembly of both rare and abundant bacterioplankton [21]. Soil moisture could be the most important element in determining the processes of both abundant and rare bacteria assembled in coastal wetlands [24]. Moreover, increased soil organic matter has been shown to result in a shift of dominant assembly process from homogeneous selection to variable selection during succession [28].
Aquatic-terrestrial ecotones affected by periodic flooding are characterized by dynamic environmental conditions, which often result in distinct microbial communities [1,6]. Nevertheless, studies evaluate the assembly processes of abundant and rare communities and corresponding determinants in aquatic-terrestrial ecotone ecosystems with environmental gradients remain scarce. In the present study, we focus on the abundant and rare subcommunities of both bacteria and archaea in a riparian zone that experiences periodic water flooding. We specifically hypothesized that (i) the structures of abundant and rare bacterial and archaeal communities are significantly influenced by flooding; (ii) abundant and rare bacterial and archaeal communities are structured by different assembly processes; (iii) flooding exhibits differentiated influences on the community assembly processes of abundant and rare bacteria and archaea. To validate these hypotheses, we employed high-throughput sequencing and null model analysis to estimate the effects of flooding on community structures and their assembly processes along a riparian elevation.

Study Area and Sampling
In order to explore the processes underlying abundant and rare community assemblages across environmental gradients, a riparian zone located in a natural wetland reserve (Baijiaxi: 31°09′02″ N, 108°33′45″ E) on the Pengxi River of the Three Gorges Reservoir, China was selected as a study area (Fig. 1a). This area is characterized by a humid, midsubtropical monsoon climate. The annual air temperature is 10.8-18.5 °C with a mean annual precipitation of 1200 mm. The zonal soil type is classified as purple soil (entisols soil) dominated by a silty clay texture [6]. As one of the main tributaries of the Three Gorges Reservoir, the water level of the Pengxi River periodically fluctuates as a result of the Three Gorges Dam operations since the dam's completion in 2010; the level begins to rise in September due to the reservoir filling for power generation, reaching its peak of 175 m in November, and then subsequently drops to a minimum of 145 m in April of the following year [6].
Soil samples were taken on October 10, 2013; April 1, 2014; September 23, 2014, and May 6, 2015. These dates represent the period immediately before and after the rising of the water level over two consecutive inundation cycles.

3
The five sampling sites, which were at an elevation of 150, 155, 165, 170, and 175 m along the riparian zone, represented different levels of flooding (Fig. 1b). At each sampling site, four subsamples from a depth of 0-10 cm were taken using a Petersen grab if below the water level, or a stainless-steel core sampler if above the water level. The subsamples at each site were sealed in a plastic bag to be mixed into a composite sample. A total of 20 composite soil samples were obtained during the four sampling excursions. Samples were put on ice and transported to the Key Laboratory of Reservoir Aquatic Environment, Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences. Samples destined for DNA extraction were immediately stored at -20 °C upon arrival. The remaining samples were stored at 4 °C for physicochemical analysis. For detailed data of the soil physicochemical properties, please refer to a previous article from the authors [6].

DNA Extraction and Sequencing
DNA was extracted from 0.25 g soil using the PowerSoil DNA Isolation Kit (Mobio, USA), following the manufacturer's instructions. The quantity and quality of extracted DNA were determined using a NanoVue Plus Spectrophotometer (GE Healthcare, UK) and 1% (weight/volume) agarose gel electrophoresis, respectively. Bacterial and archaeal 16S rRNA gene was amplified using primers 338F/806R [29] and A364aF/A934bR [30], respectively. Amplicons were purified using an AxyPrep DNA Gel Extraction Kit (Axygen, USA), and subsequently sequenced on the Illumina MiSeq PE300 Platform at Majorbio Bio-Pharm Technology Co., Ltd., Shanghai, China (http:// www. Major bio. com).
Raw fastq files were demultiplexed and quality-filtered using Trimmomatic (version 0.30) [31]. Reads with ambiguous characters were removed [9]. Sequences were subsequently clustered into operational taxonomic units (OTUs) based on 97% sequence similarity. The phylogenetic affiliation of each 16S rRNA gene sequence was analyzed by RDP Classifier against the SILVA 138 database (https:// www. arb-silva. de/ docum entat ion/ relea se-138/). To minimize sequencing errors, singletons (OTUs with only one sequence in all samples) were eliminated, and sequences were thereafter rarefied to an even sequencing depth based on the sample with the lowest sequence reads, thereby making samples comparable. Alpha-diversity indices, i.e., Chao1 richness and Shannon diversity, were computed based on rarified sequences using the "Picante" package in R (v. 4.0.5). Sequences were deposited in the NCBI under accession number PRJNA863415 for bacteria, and PRJNA394057 for archaea.

Classification of Abundant and Rare Taxa
Operational taxonomic units (OTUs) were defined as abundant and rare according to 1% and 0.01% thresholds in the relative abundance, respectively [32]. All OTUs were further classified into six categories [33,34]: always abundant taxa (AAT), conditionally abundant taxa (CAT), always rare taxa (ART), conditionally rare taxa (CRT), moderate taxa (MT), conditionally rare and abundant taxa (CRAT) ( Table 1). To focus the comparison of abundant taxa and rare taxa, and to avoid confusion, AAT, CAT, and CRAT were combined into abundant taxa, while ART and CRT were combined into rare taxa [24,35].

Null Model Analysis
With a null model approach, the mean nearest taxon distance (MNTD) and nearest taxon index (NTI) were calculated to characterize the phylogenetic clustering of abundant taxa and rare taxa using the "Picante" package with the "ses.mntd" function in R v. 4.0.5 [21,23]. NTI is a standardized measure of phylogenetic distance between each taxon in a sample and the taxon most closely related to it is also present in the sample. It corresponds to the standardized measure of the observed MNTD relative to the mean of the null distribution with 999 randomizations. For a single community, a mean value of NTI > 0 suggests coexisting taxa are more phylogenetically clustered, while a mean NTI < 0 suggests the taxa are overdispersed [20].
The relative contributions of different assembly processes to abundant and rare subcommunities were further determined according to the β-nearest taxon index (βNTI) and Raup-Crick index (RCI) again with a null model approach [21]. A value of |βNTI|> 2 indicated that the primary influence on the turnover of a given community was caused by deterministic processes (variable selection + homogeneous selection), with βNTI > 2 reflecting the effect of variable selection and βNTI < − 2 homogeneous selection [22]. Stochastic processes (dispersal limitation + homogenizing dispersal + undominated processes) were identified as the main driver when the value of |βNTI|< 2. Particularly, the turnover of a community was governed mainly by dispersal limitation when |βNTI|< 2 and RCI > 0.95, whereas the turnover of a community was driven primarily by homogenizing dispersal when |βNTI|< 2 and RCI < − 0.95. Moreover, values of |βNTI|< 2 and |RCI|< 0.95 indicated the effect of undominated processes, that is, selection and dispersal limitation, did not strongly drive compositional turnover [22].

Statistical Analyses
One-way analysis of variance (ANOVA) based on the least significant difference (LSD) was conducted to determine the significant differences within the Chao1 richness, Shannon diversity, and NTI among samples from different elevations (SPSS Statistics 20.0, USA). Principal co-ordinates analysis (PCoA) based on the Bray-Curtis distance was conducted to explore the distributions of bacterial and archaeal subcommunities by using Canoco 5 software [36]. Permutational multivariate analysis of variance (PERMAVONA) based on Bray-Curtis distance at the OTU level was conducted using the "vegan" package in R (v. 4.0.5) to reveal differences in bacterial and archaeal subcommunities between sampling elevations and their relationship with soil physicochemical properties. Spearman's rank correlation was used to assess the abundance-occupancy relationship of both bacteria and archaea taxa. For this, the log-transformed mean of their relative abundance and the number of sites they respectively occupied was used [37]. Mantel test based on Spearman's correlation was performed to determine the effects of environmental factors (based on Euclidean distance matrices) on the assembly processes of abundant and rare subcommunities using the "ggcor" package in R v. 4.0.5. The linear regression between change in environmental variables and βNTI of different subcommunities was conducted in Origin 2022 (OriginLab, USA).

Characteristics of Periodic Flooding
Since the completion of the Three Gorges Dam in 2010, the soils in the riparian zone had been cumulatively inundated for 16 to 1474 days, depending on the sampling elevation and sampling date that soils were taken (Fig. 1c).  (Fig. 1c). Moreover, the RFE ratios, defined as the accumulated flooding to exposure time (RFE), which was employed to characterize the flooding frequency of each sampling site, were less than 1 for the three sites with an elevation equal to or above 165 m. By contrast, the ratios for sites located at elevations lower than 155 m were consistently greater than 1 (Fig. 1c).

Diversity and Distribution of Abundant and Rare Subcommunities
A total of 508,905 and 561,270 high-quality sequences were respectively obtained for bacterial and archaeal communities from 20 soil samples; these samples were rarefied to an even sequencing depth based on the lowest sequence reads across all samples (19,938 for bacteria and 20,856 for archaea). Subsequently, rarefied sequences of bacteria and archaea were divided into 5062 and 792 OTUs, respectively, at 97% similarity. Thereinto, for bacteria, 73 and 4925 OTUs were respectively identified as abundant and rare taxa, which accounted for 22.7% and 67.4% of the total sequences; for archaea, 63 and 729 OTUs with 95.5% and 4.5% of the total sequences were identified as abundant and rare taxa, respectively (Table S1). Moreover, both abundant and rare bacteria and archaea exhibited a significant abundance-occupancy correlation with stronger correlations observed for rare taxa (Fig. S1). Abundant taxa for both bacteria and archaea were more widespread than rare ones; only 32.2% and 5.5% of rare bacteria and archaea were found in more than 50% of all samples, respectively (Fig. S1). In general, sites at lower elevations showed higher alpha diversity. For the bacterial communities, higher Chao1 richness and Shannon diversity of rare subcommunities were observed at sites with an elevation of 150 m and 155 m (one-way ANOVA, P < 0.05; Fig. 2a, b). Sites at the remaining elevations did not show significant differences in alpha diversity. For both abundant and rare archaeal subcommunities, the sites at 150 m and 155 m had significantly higher Chao1 richness and Shannon diversity than those at the sites of 175-165 m (one-way ANOVA, P < 0.05; Fig. 2d, e). In addition, sites with higher elevation had higher mean values of NTI for both abundant and rare bacteria (one-way ANOVA, P < 0.05; Fig. 2c). However, while sites at higher elevation also showed higher NTI values for abundant archaea, they demonstrated lower values for rare archaea (one-way ANOVA, P < 0.05; Fig. 2f).
Moreover, sites between 150-and 175-m elevation showed significant differences in compositions of both abundant and rare bacterial and archaeal subcommunities based on PCoA (PERMAVONA, P < 0.05; Fig. 3a-d). Furthermore, soil environmental factors, including NH 4 + , Fe 2+ , and Fe 3+ contents, moisture, and sampling elevation were the significant factors shaping the abundant and rare subcommunities for both bacteria and archaea (Table S2). Thereinto, the sampling elevation imposed the most significant effect on the composition of abundant (R 2 = 0.225 and 0.317, respectively) and rare subcommunities (R 2 = 0.221 and 0.122, respectively) for both bacteria and archaea (Table S2).

Drivers of Community Assembly Processes
To determine the relative influence of each variable on the stochastic and deterministic processes, the relationship between environmental variables and the βNTI of different subcommunities was assessed based on Mantel tests. The community assembly process of rare bacteria was influenced mainly by soil NH 4 + and Fe 2+ contents, Fe 2+ /Fe 3+ , moisture, and RFE. Abundant archaeal subcommunities were significantly driven by moisture and RFE, while soil total carbon (TC) and ratio of carbon to nitrogen (C/N) were the best predictors of the assembly process driving the rare archaeal subcommunities (Mantel test, P < 0.05; Table S3). Notably, no environmental variable was observed to significantly determine the assembly of the abundant bacterial subcommunity (Mantel test, P > 0.05; Table S3).
Pairwise comparisons of βNTI for rare bacteria were significantly positively correlated with variations in soil NH 4 + and Fe 2+ content, Fe 2+ /Fe 3+ , moisture, and RFE (Fig. 4a-e). This suggests that an increasing divergence in those environmental variables influenced the community assembly of the rare bacteria due to deterministic processes (i.e., variable selection). For abundant archaea, differences in both soil moisture and RFE were significantly positively correlated with βNTI (Fig. 4f, g). This indicates that an increasing discrepancy in soil moisture and RFE led to an increase in the stochastic processes that propelled the assembly of the abundant archaeal subcommunity. Moreover, the significantly positive correlations between rare archaeal βNTI and soil C/N and TC (Fig. 4f, i), indicate the increasing divergence of these two variables resulted in a transition from homogeneous selection to stochastic processes in the community assembly of the rare archaea.

Assembly Processes for Abundant and Rare Subcommunities
For the bacterial community assembly, stochastic processes (70.5%) dominated within the abundant subcommunity, whereas deterministic processes (73.2%) contributed most to the rare subcommunity (Fig. 5a). Moreover, dispersal limitation, which belongs to the stochastic processes, contributed more greatly to the assembly of the abundant subcommunity (62.6%) than to the rare subcommunity (20.0%). Variable selection, which is a type of deterministic process, demonstrated greater effects on the assembly of the rare subcommunity (64.7%) than on the abundant subcommunity (29.5%) (Fig. 5a). At sites at a 150-and 155-m elevation, dispersal limitation (a stochastic process) showed greater contributions to the assembly of the abundant subcommunity (Fig. 5b); the effect of variable selection (a deterministic process) on the rare subcommunity at sites with lower elevations was greater than those at higher elevations (Fig. 5c).
For archaeal assembly, stochastic processes dominated both the abundant subcommunity (90.0%) and the rare subcommunity (60.0%). Undominated processes and dispersal limitation, which are both characterized as stochastic processes, contributed a comparable amount to the assembly of both abundant (46.8% and 41.6%) and rare subcommunities (28.9% and 26.8%), with contributions to abundant subcommunities being higher than to rare subcommunities (Fig. 5a). Homogeneous selection of deterministic processes showed the greatest contribution on the assembly of the rare subcommunity, but had no substantial impact on the assembly of the abundant subcommunity (Fig. 5a). Furthermore, undominated processes in the assembly of abundant archaea decreased with decreasing elevation, while dispersal limitation increased with decreasing elevation (Fig. 5d). At sites at 150-and 155-m elevation, homogeneous selection (a deterministic process) showed greater contributions to the assembly of the rare subcommunity (Fig. 5e).

Flooding-related Distribution of Abundant and Rare Taxa
Generally, the diversity and structure of both the abundant and rare communities were influenced by flooding. Compared to higher elevations, riparian soils at lower elevations were more frequently affected by flooding, which significantly increased the alpha diversity of the microbial community. This may be related to the higher nutrient availability of frequently-flooded soils which have a greater exchange rate with the aquatic system and thus more extensive nutrient cycling [38]. Indeed, flooding can carry large amounts of nutrients from upland soils to later deposit in lower-elevation soil [6]. Periodic flooding can also create a heterogeneous environmental gradient in the riparian zone by altering soil biogeochemical cycles [1,6]. Therefore, the community structures of both abundant and rare taxa exhibited elevation-related distributions. Notably, soils with more frequent flooding (i.e., 155 m and 150 m with RFE > 1) harbored significantly different bacterial and archaeal communities from sites in higher elevations (≥ 165 m). This suggests that there may be a threshold for the frequency of dry-wet alternation which causes significant changes to the microbial communities in soil, such as an RFE equal to 1 in the present study. h-i Correlations between βNTI and changes in soil TC and C/N for rare archaea, respectively. Dotted lines indicate the βNTI significance thresholds of − 2 and + 2. "**" and "***" represent significant correlations at P < 0.01 and 0.001, respectively. RFE, ratio of flooding to exposure Periodic flooding also acts as one of the most stressful environmental transitions known to soil microbial systems [39,40]. The effective utilization of resources, high growth rate, and better adaptability to changing environments allow abundant taxa to play a stable and key role in the biogeochemical cycle in a disturbed environment [8,10]. Compared to abundant taxa, rare taxa may play a more active role during fluctuating environmental conditions (i.e., dry-wet alternation) due to their sturdiness and continuous regrowth [41]. Meanwhile, the high diversity of rare communities underlines their importance as "seed banks" to the diversity of microbial species [9,13]. This entitles the rare taxa with a prominent metabolic potential under appropriate conditions [42]. In this study, most of the rare taxa (97.1% for bacteria and 69.5% for archaea) were conditionally rare taxa (Table S1), which could shift to abundant given the favorable environmental conditions. It implies that rare taxa remain latent until favorable conditions are met, based on the heterogeneous habitats in riparian soils.

Water Flooding Determines the Assembly Processes of a Community
In the present study, the community assembly of the abundant bacteria was mainly governed by stochastic processes (i.e., dispersal limitation), whereas the rare bacterial subcommunity was primarily driven by deterministic processes (i.e., variable selection). For archaeal communities, both the abundant and the rare subcommunities were governed by stochastic processes (i.e., dispersal limitation and undominated processes) although with an indispensable contribution of deterministic processes (i.e., homogeneous selection, 40%) on the community assembly of the rare archaea (Fig. 3). These observed discrepancies in the assembly processes of abundant vs. rare communities were largely attributed to the ability of different individual microorganism to cope with changing environments [43]. Here, the niche breadth of the abundant taxa was observed to be wider than that of the rare taxa (Fig. S1). Abundant taxa have the potential to become more abundant in a given environment due to their efficient utilization of a wider range of resources [10]. Thus, stochastic dispersal is more likely to influence abundant taxa as there are more individuals in the environment [9]. On the contrary, rare taxa with a narrow niche breadth are more sensitive to environmental changes, which thus attributes to a low growth rate and greater competition [21,24,27]. Therefore, rare taxa, especially always rare taxa, which are anticipated to persistently maintain low abundance despite changing environmental circumstances [44], are more easily restricted to limited locations by environmental filtering. Moreover, the different community assembly processes of bacteria and archaea might be explained by their differences in metabolic activity, dispersal capability, and growth habit [45,46]. Notably, the assembly processes of both abundant bacterial and archaeal subcommunities, driven by dispersal limitation, were enhanced with the influence of flooding. Dispersal limitation, belonging to the stochastic processes, is defined as a limited exchange of species between different communities due to their low dispersal rates [22] and is thus a process which is predicted to increase dissimilarities in community composition. Classically, a dispersal process is usually a combination of the source community and physical mediators (e.g., water) that transfer individual species [47]. The presence of continuous water routes may promote microbial mobility over the soil [48]. However, the long-term and periodic flooding formed habitat differences (such as oxygen availability) along the riparian elevation might enhance the physiological barrier effects and thus increase the dispersal limitation [49]. Obviously, the effect of dispersal also depends on the taxonomic composition [47]. Abundant taxa with high local abundance have strong environmental adaptability and thus often exhibit ample changes. They are therefore less likely to be excluded from a given habitat [9]. Therefore, dispersal limitation is to some extent deterministic since the capacities of species to disperse can be influenced by environmental conditions [50]. Particularly, the assembly process of abundant archaea was also significantly affected by undominated processes, and flooding showed opposite effects on this process compared to dispersal limitation. Generally, the undominated processes imply the greater effect of ecological drift and negligible dispersal effect drive community turnover [22]. The diminishing contribution of undominated processes at lower elevations suggests that the role of drift relative to dispersal limitation lessened as elevation decreased. It follows that the effects of stochastic processes (such as undominated processes and dispersal limitation) on abundant archaea may be closely associated with flooding. This was demonstrated by the substantial association between the βNTI of abundant archaea and soil moisture, and RFE (Fig. 4), which suggests that increases in soil moisture and RFE may promote stochastic processes.
For the rare taxa, the community assembly was dominated by two opposing deterministic processes, namely variable selection for bacteria and homogeneous selection for archaea. Variable selection often results in dissimilar community compositions due to divergent selective environments, while homogeneous selection promotes convergent community compositions through a consistent environmental filtering [22]. In this study, 97.1% of the rare bacteria belonged to conditionally rare taxa which could periodically shift between abundant and rare in response to environmental fluctuations [13]. It is clear that periodic flooding can create such fluctuating environmental conditions, thereby making variable selection the primary driver of the assembly of rare bacteria [10]. This was further supported by the substantial relationship between the βNTI of rare bacteria and soil Fe 2+ content, Fe 2+ / Fe 3+ , moisture, and RFE. It suggests that changes in these soil parameters, as triggered by flooding, may be the source of variable selection for rare bacteria. Unlike rare bacteria, a large part of rare archaea (always rare taxa, accounting for 30.5% of total rare taxa; Table S1) permanently persists at low abundances, suggesting their rarity in the riparian soils. Given that these rare species occupy specialized niches and frequently interact with other species, it has been shown that such rarity results from strong homogeneous selection [10]. Aligning with previous study, homogeneous selection was found to explain the prevalence of always rare taxa [44], which suggests a large convergence and a weak distance-decay relationship with these taxa [51]. This in line with the relatively low variation of the rare archaeal subcommunity along the riparian elevation (R 2 = 0.257). Meanwhile, the βNTI of rare archaea was significantly correlated with soil C/N and TC, suggesting the role of C/N and TC as consistent environmental filters in driving the community assembly of rare archaea. Our previous study has proven that C/N and TC in the riparian soil does not vary significantly along the elevation [6].

Linkage Between Community Assembly and Riparian Ecosystems
Riparian zones, characterized by their aquatic-terrestrial ecotone ecosystems, have long been regarded as biogeochemical hotspots of nutrient cycling because of the fluctuating environmental conditions and intensive material exchange between soil and water trigged by flooding [52]. Generally, microbial communities contribute immensely to soil biomass [53], underpinning integral biogeochemical processes and ecosystem functioning [11], where abundant and rare taxa each play a vital role. In this study, abundant taxa exhibited a ubiquitous distribution mainly driven by stochastic processes. This indicated that abundant taxa can persist at relatively high abundances across the riparian soils with a low probability of extinction, which ensures the stability of an abundant taxa-mediated ecosystem functioning [8,9]. For example, abundant taxa are regarded to be the most crucial species in carbon cycling [8]; the stochasticity-driven assembly keeps their roles in carbon cycling from changing substantially as a result of great environmental fluctuations caused by flooding. In contrast, rare communities often consist of functionally relevant taxa that perform similar ecological functions, and changing environmental conditions can result in fluctuation in their abundance [10]. Given the predominance of deterministic processes in rare communities, many rare taxa, especially conditionally rare taxa, sustain a low growth rate or even adopt a type of dormancy to cope with unfavorable or harsh local environmental conditions, such as long-term periodic flooding [54]. Undoubtedly, these unfavorable environments may be suitable conditions for other conditionally rare taxa with different ecological niches to bloom, thus promoting the relevant ecosystem functions [10]. This may explain why riparian zones are deemed biogeochemical hotspots. Overall, in the context of climate change, which foresees increased flooding and precipitation in certain areas, uncovering the patterns of microbial community assembly processes in response to flooding can provide insight into their ecological functional roles.