Spatiotemporal evolution and assembly processes of ammonia-oxidising prokaryotic communities in coastal reclaimed soils

Sarfraz Hussain Nanjing Agricultural University College of Life Sciences Yifan Yin Nanjing Agricultural University College of Life Sciences Senlin Liu Nanjing Agricultural University College of Life Sciences Shanshan Yan Nanjing Agricultural University College of Life Sciences Dongjie Chen Nanjing Agricultural University College of Life Sciences Feng Wang Institute of Eco-Environmental Sciences, Ningbo institute of Agricultural Sciences, Ningbo,China Cao Hui (  hcao@njau.edu.cn ) Nanjing Agricultural University College of Life Sciences https://orcid.org/0000-0003-3406-3803

using the approach described above, we sought to quantify the possible AOA and AOB in ecosystem functioning along chronosequence of soil formation. We hypothesised that (1) along with the chronosequence of reclaimed soils, AOA and AOB community composition and structure varied distinctly due to the strong in uences of soil physicochemical properties and geographic distance, (2) ecosystem function provides quantitative information about microbial communities' response to environmental conditions, and (3) with the change of environment and land management, there is a speci c corresponding relationship between ammonia-oxidising prokaryotic communities and ecosystem function.

Materials And Methods
Study area description and eld soil sampling The present study was conducted on the south coast of Hangzhou Bay (121°05′-121°35′′E, 30°0′-30°25′N) in Cixi County, Zhejiang Province, China. The study area is a typical northern subtropical monsoon climate zone, with a mean annual precipitation of 1273 mm and temperature of 16.0°C. Salt marshes formed from the Qiantang Estuary saltwater sediments and diked lands were exposed to tidal effects at an elevation range from 2.6 to 5.7 m [23]. The tidal salty marsh was eventually reclaimed for crop growth, beginning 1000 years ago with the formation of dikes at different historical stages. Based on the construction times of dikes, the years of soil reclamation were calculated, and a detailed description is available online in Chinese at (www.cixi.gov.cn). In this study, around 1000 years of ancient spanning of soil chronosequence was identi ed, including undisturbed coastal salty marsh (T0) and soil reclaimed 5years (T5), 20-years (T20), 50-years (T50), 60-years (T60), 120-years (T120), 200-years (T200), 220-years (T220), 280-years (T280), 500-years (T500), and 1000-years (T1000) ago.
Distances between these soils were no more han 30 km in a similar topography. Soil samples were collected and mixed from six random locations, and then six mixed soils sampled from the same reclaimed year were considered replicates. A stainless-steel auger (3-cm diameter) was used for soil collection. Brie y, soil samples were taken from the plough layer (0-20 cm), and a total of 66 samples were obtained representing speci c reclamation years (Fig. S1). The taxonomic class of soil samples was uvisol as per Food and Agriculture Organization/United Nations Educational, Scienti c and Cultural Organization (FAO/UNESCO), which was formed from uvial deposits around the Qiantang River, and the soil texture was silty loam to clay-loam [24,25]. In December 2015, reclamation site samples (T5-T1000) were collected from a traditional soybean-broccoli rotation eld during active broccoli growth from crop free sites to avoid the possible impact of plant roots. Compound fertilisers were applied @ 900 kg·ha -1 , of which nitrogen (N):phosphorus (P 2 O 5 ):potassium (K 2 O) was estimated to be 1:1:1. The salt marsh soils (T0) were sampled in the high-tide area under Phragmites australis. Samples were immediately taken to the laboratory on ice packs and then sieved through a 2 mm mesh to remove roots and another residue.
After that, sieved soils were divided into three parts; two parts were kept at -80°C and 4°C for total DNA extraction, mineralised N content, and nitri cation analysis. The third part was air-dried for the determination of physicochemical parameters of the soil.

Total DNA extraction
The soil total genomic DNA was extracted using the Soil DNA kit (Omega Bio-Tek, Inc., Norcross, GA, USA) according to the manufacturer's instructions from 0.5 g of soil. The purity and quality of DNA were measured by 1% agarose gel electrophoresis and a spectrophotometer, respectively (Nanodrop, PeqLab, Germany).
Quanti cation of the amoA gene in AOA and AOB To estimate the amoA functional gene for AOA and AOB, quantitative PCR (qPCR) was performed. The ammonia-oxidising archaeal amoA gene was quanti ed by uorescence qPCR using the primers shown above (section 2.3). Absolute quanti cation was carried out on the Applied Biosystems QuantStudio TM 6 Flex Real-Time PCR System (Life Technologies Corporation, Carls-bad, CA, USA) ampli er using Hieff ® qPCR SYBE ® Green Master Mix (YEASEN). The uorescence qPCR reaction volume was 20 μl, containing 10 μl 2 × SYBR Green Mix, 0.8 μl upstream and downstream primers (10 pmol / μl), 1 μl diluted total DNA template, and 7.4 μl double distilled water. The protocol for the PCR reaction was as follows: predenaturation at 95°C for 5 min; 40 cycles at 95°C for 30 s, annealing at 56°C and 58°C for 60 s, and extension at 72°C for 60 s and 35 s (archaea and bacteria, respectively); the melting curve temperature range was 65-95°C.

Sequencing data processing and analysis
Single-end reads were assigned to samples based on their unique barcode, and Cutadapt (V1.9.1) was used for quality ltering to obtain high-quality clean reads [26]. Reads were analysed by the Silva database using the UCHIME algorithm to classify and delete chimera sequences [27,28]. The Uparse program grouped the puri ed sequences (Uparse v7.0.1001). The same operational taxonomic units (OTUs) were assigned sequences with 97% identity [29]. The Mothur algorithm [30] was used to perform annotated OTU analysis to obtain representative species and count the number of OTUs per sample in taxonomic information (Kingdom, Phylum, Class, Order, Family, Genus, and Species level). Archaeal and bacterial OTUs were identi ed and assigned in the archaeal and bacterial Silva Database [27]. For multiple sequence alignment, MUSCLE software (Version 3.8.31) was used to examine the phylogenetic relationship between different OTUs and the classi cation of the dominant species in different samples [31].

Phylogenetic diversity matrices analysis
The Mean Phylogenetic Distance (MPD) determined by Net Relatedness Index (NRI) for all co-occurring populations revealed 'basal dispersal' within the population, while the Net Taxon Index (NTI) measures the Mean Nearest Taxon Distance (MNTD) between populations, thus calculating the population 'terminal' phylogenetic dispersion [32]. The MPD and MNTD standardised effect size was measured using the ses.MPD and ses.MNTD commands in the 'Picante' package of R [33]. The distribution of 999 null values, computed by shu ing the tip labels in the tree, was used to account for temporal changes of the species pool. ses.MPD. observed ses.MNTD. The observed values were multiplied by -1 to be equivalent to the NRI and NTI [33].

Statistical analyses
Microbial diversity and a comparison of gene copy numbers were evaluated by one-way analysis of variance (ANOVA), followed by Duncan's multiple range test at SPSS 25.0 for Windows. Discrepancies were considered statistically signi cant at P < 0.05. Soil ecosystem function (potential nitri cation and nitrogen mineralisation rates) correlation with environmental parameters and amoA gene abundance, as well as reclaimed soil age, were calculated by Spearman's correlation analysis. Alpha diversity analysis namely observed-species, Chao1, Shannon, Simpson, ACE, and good-coverage, were calculated with QIIME (Version1.7.0). Beta diversity analysis was performed on QIIME software (Version1.7.0) to evaluate the differences in community complexity among samples and to calculate the Unique Fraction (UniFrac) distance. The Unweighted pair group method with arithmetic mean (UPGMA) sample clustering was conducted in R software (Version 2.15.3). NMDS (non-metric multi-dimensional scaling) analysis and non-parametric Analysis of Similarity (ANOSIM) based on the Bray-Curtis distance for analysing compositional differences between ammonia oxidiser prokaryotic communities were performed in R software (Version 2.15.3). Differential abundance of ammonia oxidiser between individual samples was calculated by MetaStat analysis [36].
Temporal variations in phylogenetic beta diversity were analysed by Permutational Multivariate Analysis of Variation (PERMANOVA) using R function ADONIS based on the Bray-Curtis distance. The effect of physical and chemical factors on ammonia-oxidising microbial communities was analysed by canonical correspondence analysis (CCA). The cca-env t function was used to test the impact of each environmental factor on the distribution of species. Spearman rank correlation coe cient analysis was used to correlate the relationship of environmental factors and reclaimed soil stages with ammonia oxidiser microbial abundance (alpha diversity) and species distribution. We conducted correlation analyses to assess the relationship of environmental variables with NTI and NRI to examine the change in community structure by Spearman's correlation. The mechanisms of ammonia oxidiser community assembly were determined by calculating βNTI. Brie y, if βNTI > 2 or βNTI < -2, deterministic processes may be important in shaping the community composition across all sites, whereas stochastic processes may play a signi cant role in community assembly processes when the values of βNTI are between -2 and 2. The βNTI > 2 revealed signi cantly more phylogenetic turnover than predicted, which is often interpreted by chance as variable selections, while βNTI < 2 referred to less phylogenetic turnover than expected, i.e., homogeneous selection ). If the measured βMNTD values did not bring clarity to signi cant differences from the null distribution of βMNTD, e.g. |βNTI|<2, the observed phylogenetic variability was not the consequence of selection [37]. To overcome this problem, the Bray-Curtis-based Raup-Crick metric (RC bray ) was further determined as described by [38]. Brie y, the values of (RC bray ) ranged between -1 and 1, and we compared |βNTI|<2 and (RC bray ) values. The relative contribution of dispersal limitation was estimated as the percentage of pairwise comparison between | βNTI| < 2 and (RC bray ) values > 0.95, whereas |βNTI| < 2 and (RC bray ) values > -0.95 indicated homogenous dispersal. The undominated process was calculated as |βNTI| < 2 and (RC bray ) values > 0.95. The undominated concept described a state wherein the primary cause of variations between population compositions was neither dispersion nor selection, namely ecological drift (population sizes uctuating due to stochastic birth and death events) [39]. We applied ordinary least-squares regression analysis to determine the slope of the association among phylogenetic relatedness with environmental factors. The Mantel tests assessed the association of phylogenetic distance and environmental parameters with geographic distance. Using partial Mantel tests with Pearson's correlation coe cient and 999 permutations, the relationships between βNTI, geographical distance, and Euclidean distances in environmental parameters were analysed.

AOA and AOB community composition
A total of 840 and 714 archaeal and bacterial OTUs, respectively, were obtained, and community' annotations were performed on the OTU sequences by comparison to the Silva132 database. The archaeal community shared OTUs (14.4%) in all soil samples (Fig. 1A), and coastal salt marsh soil exhibited the highest number of unique OTUs (55.8%) compared with reclaimed soils. The archaeal community had 457 annotated OTUs (54.40%). The overall proportion of annotated OTUs at the phylum level was 24.17%. The cumulative archaeal community composition is shown in Fig. 1B. The relative abundance of AOA phylum Crenarchaeota in coastal salt marsh soil was 19.1%, and it increased to 48.9% after ve years of reclamation and then decreased (from 12.3 to 6.6% in the soil 20 and 50 years after reclamation, respectively), showing an asymmetrical distribution along with chronosequence. A linear increase in relative abundance was found with reclamation time ranging from 60 to 280 years (from 39.1 to 56.6%, respectively) and reached the highest abundance in soil reclaimed 500 years ago (63.2%). Abundance decreased in the 1000-year-old reclaimed soil (10.3%). The relative abundance of the phylum Thaumarchaeota decreased linearly with reclaimed years, with 27.0% in the marsh soil and 0.1% in the 1000-year reclaimed soil. The community distribution of AOA in the 1000-year reclaimed soil was relatively low as determined by meta stat analysis, and there were signi cant differences in archaeal community composition among the soils with different reclamation years (Fig. S2). The proportion of OTUs shared by ammonia-oxidising bacteria in the soil during different reclamation years was 16%, with the lowest unique OTUs in T500 (0.5%) and the highest in T0 (17%) (Fig. 1C). The abundance of Proteobacteria, the most predominant bacterial phylum across all soil samples (19.1%), was detected, and it was the highest (61%) in the salt marsh and decreased with the number of reclaimed years (from 42% in 5-year reclaimed soil to 3% in 1000-year reclaimed soil). The relative abundance of Nitrosospira and Nitrosomonas was 5 and 3%, respectively, in 5-year reclaimed soil and decreased at later reclamation stages (Fig. 1D). Metastat analysis (Fig. S3) further proved that the abundance of proteobacteria community composition signi cantly decreased from the coastal salt marsh to 1000-year-old reclaimed soil, while a contradictory trend was noted in the composition of uncultured bacteria. The community composition of others (uncultured bacteria and uncultured prokaryotes) signi cantly decreased in 5-to 20-year-old reclaimed soil. amoA gene copy number and its association with nitri cation rate To quantify the number of amoA gene copies in archaeal and bacterial communities across all reclaimed soils, quantitative PCR was used ( Fig. 2A). Reactions were performed in triplicate for all soils, with an e ciency of 92% and 96% for archaeal and bacterial amoA genes, respectively (r 2 = 0.99). Archaeal amoA gene abundance increased progressively with reclaimed time (from marsh to 1000-year reclaimed soil). Bacterial amoA gene abundance showed a similar trend along reclamation time. In the reclaimed soils, both archaeal and bacterial aomA gene abundance was greater than that in the marsh and youngest reclaimed soil. In addition, Pearson's correlation analysis showed that the abundance of the archaeal amoA gene was negatively correlated with pH, electron conductivity (EC), and available potassium (AK) and positively correlated with soil organic matter (SOM), available nitrogen (AN), total nitrogen (TN), and reclamation years (Table S1). Bacterial aomA gene abundance was positively associated with AP and negatively associated with EC. The nitri cation rate was signi cantly and positively correlated with archaeal aomA gene abundance (P < 0.01), while the net nitrogen mineralisation rate was not signi cant. Furthermore, bacterial amoA gene abundance was not associated with nitri cation rate and net nitrogen mineralisation.

Taxonomic and phylogenetic alpha diversity analysis
Faith's phylogenetic diversity (PD) values of ammonia oxidiser communities generally decreased along with soil chronosequence. To determine whether the AOA and AOB community structure assembled via stochastic or deterministic processes among reclaimed soil, the NRI and NTI were calculated. Based on the NRI and NTI values, we found that AOA and AOB NRI and NTI values were above 2 in all reclaimed soils, except for the AOA community (NTI < 2) 50 years after soil reclamation and the AOB community (NTI > -2) in marsh soil ( Figure S4). Most of the NRI and NTI values showed that populations of cooccurring ammonia oxidisers were more phylogenetically related than predicted by chance. In addition, in both communities, we correlated NTI and NRI values with environmental variables, including AN and TN, which were not signi cant. AOA NTI and NRI values were signi cantly and positively associated with pH (P < 0.01) and were not correlated negatively with NO 3 --N (P < 0.01) EC and AK ( Figure S5). AOB NTI and NRI values were negatively associated with EC and AK, whereas other environmental factors were not signi cantly associated.
Taxonomic and phylogenetic beta diversity analysis Distribution of the AOA and AOB community were distinct in later reclamation stages based on NMDS analysis (Fig. 3). The composition of ammonia oxidisers in the coastal salt marsh was signi cantly different from all reclaimed soils. Similarly, AOA and AOB communities 5 and 1000 years after soil reclamation also separated from other reclaimed soils. ANOSIM analysis revealed signi cant differences in AOA and AOB community composition (Table S2). The AOA community showed non-signi cant variations between T120 and T200 (R = 0.183, P = 0.077), T20 and T50 (R = -0.0260, P = 0.571), T60 and T120 (R = 0.067, P = 0.206), and T280 and T500 (R= -0.022, P = 0.546), indicating that temporal variations in the AOA community were noted with a long reclamation time. The community structures of AOB were similar after 5-120 years of soil reclamation. The community structure of AOB was not signi cant between T20 and T60 (R = 0.098, P = 0.111), T120 and T200 (R = 0.107, P = 0.188), T60 and T120 (R= 0.007, P = 0.444), and T20 and T50 (R= 0.119, P = 0.157). The distance-based community dissimilarity (Weighted UniFrac distance) and βMNTD were used to measure the dissimilarity between different reclaimed soils. The weighted UniFrac values were greater than βMNTD, and both analyses showed that dissimilarity increased with the number of reclamation years (Fig. S6).
Quantitative analysis of the ammonia-oxidising community assembly process The ammonia oxidiser community assembly process was calculated by the βNTI and RC bray to reveal whether community assemblage mechanisms could explain the assembly process of ammonia oxidisers (Fig. 4). By counting the deviations of phylogenetic turnover, we found that AOA and AOB community assembly mechanisms showed that deterministic processes were dominant (84.71 and 55.2%, respectively) with βNTI were greater than 2 or less than -2. The stochastic process was secondary (15.29 and 44.80%, respectively) with βNTI values between 2 and -2. Variable selection contributed a larger fraction to the ammonia-oxidising community, followed by dispersal limitations. Furthermore, at each reclamation time, the AOB community in marsh soil and 60 years of reclaimed soil were assembled by stochastic processes (79.50 and 50.50%, respectively), which were in uenced by dispersal limitations (Table 2). Dispersal limitations and ecological drift (undominated processes) were secondary variables in the AOA community assembly process.

Spatiotemporal variations in phylogenetic beta diversity and community composition
The spatiotemporal variations in ammonia oxidiser community composition and phylogenetic beta diversity were analysed by weighted UniFrac dissimilarity, βMNTD, βNTI, and PERMANOVA based on ADONIS the Bray-Curtis distance. The PERMANOVA results showed that there were signi cant variations in archaeal and bacterial communities between salt marsh and reclaimed soils (P < 0.05) (Table S3). Between undisturbed salt marsh and reclaimed soils, the most considerable variations in community structure were noted (F = 26.8, R 2 = 0.73, P < 0.001). Signi cant differences among reclaimed soils were found in AOA and AOB communities, while the maximum variation was noted in soils 50-500 years after reclamation (F = 73.5, R 2 = 0.88 and F = 14.0, R 2 = 0.58, respectively). AOA and AOB communities showed non-signi cant variations in reclaimed soils between adjacent close reclamation years (P > 0.05).
Temporal variations were detected 0-1000 years after soil reclamation, indicating signi cant changes in the relative OTU composition. The weighted UPGMA cluster analysis indicated signi cant phylogenetic turnover in ammonia-oxidising prokaryotic community composition (Fig. S7A, B). Phylogenetic turnover at the phyla level in the AOA community from the coastal salt marsh to long-term reclaimed soil was associated with shifts in beta diversity composition, and AOB community composition was not correlated with phylogenetic turnover. Based on βMNTD, we found that the dissimilarity of AOA and AOB communities increased with geographic distance and environmental variables (Fig. 5). To validate whether different geographic distance-controlled community assembly, we calculated the correlation between βNTI values and spatial distance (Table S5). The results indicated that the deterministic process of geographic distance was greater than that of an environmental variable (r = 0.172, P = 0.001 and r = 0.119, P = 0.005) in the process of AOA community assembly. In contrast, the AOB community was partially negatively correlated with environmental variables and geographic distance. (VIFs). They explained 57.4% and 58.8% of the variance in AOA and AOB, respectively (Fig. 6A, B). Variance partition analysis (VPA) revealed that the relative in uence of environmental and spatial parameters on the composition of AOA and AOB communities was 55.43% and 42.55% included geographic distance (7.25% and 7.07% respectively). The community composition of AOA in uenced by NH 4 + -N and AOB in uenced by EC (3.66% and 6.77% respectively) (Fig. 6C, D). Spearman correlation coe cient analysis was used to correlate AOA and AOB phyla with environmental variables. Archaeal phylum Thaumarchaeota positively associated with pH and EC and negatively correlated with AP and NO 3 --N. Furthermore, phylum Crenarchaeota was signi cantly and positively associated with TN and negatively associated with NO 3 --N and AK (Table S4) and NH 4 + -N (Fig. S8).

Discussion
Succession of an ammonia-oxidising prokaryotic community along a coastal soil chronosequence Communities change in an orderly manner with time in a particular assumed environment, which is de ned as succession [40]. Macro-ecologists have shifted their attention from standard vegetation description to the study of community dynamics, while micro-ecologists are still unable to develop a wellestablished framework to address microbes in successional environments [41]. Several studies have shown that the composition and distribution of microbial communities (fungal, archaeal, and bacterial) changes with temporal variations [42][43][44]. Salt marshes are particularly active ecosystems, and apart from their environmental signi cance, few studies have discussed the primary succession of microbial communities in these habitats to investigate the trends and processes at the phylogenetic level that drive archaeal, bacterial, and fungal assembly dynamics [45][46][47][48]. Although much of our understanding of the assembly and dynamics of the microbial community depends on taxonomy-based evaluations (i.e., based on the 16S rRNA gene), less focus has been devoted to the distribution in natural systems of functional genes. In the present study, we investigated the successional patterns AOA and AOB communities in an undisturbed salt marsh and reclaimed cultivated soil spanning 1000 years of ecosystem development, providing a unique and dynamic landscape to study the pattern of functional microbial communities. Some studies indicated that the archaeal amoA gene is more abundant than that of bacteria in marine and terrestrial environments [49][50][51]. Conversely, mounting evidence from various oceans and coasts has shown that the abundance of bacterial amoA genes in some regions is higher than that of archaeal amoA [52][53][54]. We showed that the archaeal amoA gene was more abundant in an undisturbed coastal salt marsh than the bacterial amoA. Along with reclamation years, archaeal amoA gene abundance signi cantly increased, while bacterial amoA gene abundance was asymmetrical and comparatively lower than that of archaea (Fig. 2). These ndings suggested that the AOA amoA gene dominated not only costal marshes but also in reclaimed soil.
Marine N cycling by AOA and AOB can be affected by several factors, including physiochemical variables and external nutrient availability [55]. Here, we also found that abiotic parameters, including EC, pH, NH + 4 -N, NO − 3 -N, and SOM, had signi cant effects on the composition and diversity of ammonia oxidiser prokaryotic communities in reclaimed soils, which is an agreement with previous publications. have been identi ed as essential parameters in uencing the distribution and variety of ammoniaoxidising organisms in various habitats [61]. Previous studies indicated that AOA richness dominated in acidic soil, and AOB dominated in neutral or alkaline soil [62]. The current study revealed that soil pH and ammonia-oxidising prokaryotic community (AOA and AOB) richness continuously decreased with the succession of the soil, indicating that soil pH plays a vital role in ammonia-oxidising microbial composition. The changes in SOM, TN, and AN concentration in the soil environment were closely associated with dissimilarities in AOB community composition [63]. It was previously shown that AOA was more abundant in soil with lower concentrations of AN, while AOB increased correspondingly [64].
Our study showed that AN and SOM was signi cantly correlated with the AOB community in reclaimed soils. Moreover, the relative abundance of archaeal phylum Thaumarchaeota was positively correlated with EC and pH, while Crenarchaeota was signi cantly and positively correlated with TN. The bacterial amoA gene harboured in phylum Proteobacteria was positively correlated with pH, EC, AK, and NH 4 + -N and negatively correlated with TN, AN, and SOM concentration. Interestingly, both ammonia oxidiser communities responded differently to the number of reclamation years; therefore, both communities occupied separate ecological niches [14].
Our results suggest that differences in phylogenetic beta diversity of ammonia oxidiser communities could be described by temporal variability in nutrient availability (NH 4 + -N, SOM, AP, and AN), as well as immigration of microbial input (uncultured bacterial species and uncultured prokaryotic accumulation), and the amplitude of variation in environmental factors (EC and pH) from coastal salt marshes to reclaimed cultivated soils. Soil buffering capacity and crops became more dominant as succession proceeded, which possibly reduced the amplitude of variations, resulting in the reduction of phylogenetic turnover in AOB (Fig. 5). These temporal effects were further correlated with edaphic factors from coastal salt marshes to long-term reclaimed soils, indicating that the AOA community was dominantly driven by environmental variables compared to the AOB community (Fig. S4). Archaeal and bacterial richness continuously decreased from coastal salt marshes to long-term reclaimed cultivated soils, while the accumulation of uncultured bacteria species was observed at later reclamation stages. Phylogenetic diverse ammonia oxidiser taxa were observed at later stages than at primary reclamation stages; therefore, our study supports 'the theorem of diversity begets diversity' [65], which assumes community evolution towards species complexity.
Deterministic processes determine the assembly of ammonia oxidiser communities Assembly processes that form the community's structure have recently been of considerable interest [66,67]. Particularly for the long-term time scale, it is essential to consider the drivers in uencing ecological succession in response to environmental disturbance [68]. It has previously been stated that stochasticity has been reduced across successive phases [69,70], but there was no improvement in the assembly mechanism during succession based on the time scale. According to niche-based theory, deterministic factors, including species traits, interspecies interactions, and environmental conditions, govern community structure [71]. Our study showed that AOA and AOB communities were assembled by deterministic process (84.71 and 55.2%, respectively), as well as reclamation of coastal marshes. Our results are contradictory with previously reported studies; the pattern of increasing deterministic process might be due to long-term reclamation of coastal salt marshes over 1000 years, while previous studies demonstrated that short successional stages (more than one century) [69,72], arti cial soil disturbance, and cropping system also in uence the deterministic process [73]. A study reported that spatial distance had a signi cant role in the soil bacterial community [73], and our study also revealed that spatial distance (geographic and environmental) was signi cantly correlated with AOA and AOB community structure. In our study, the AOA community was positively correlated with geographic and environmental distance, while the AOB community was negatively associated with environmental distance and partially correlated with geographic distance. The unmeasured environmental variables also affected phylogenetic community assembly. Previous studies have shown that unmeasured environmental factors have an intensifying impact on deterministic processes that suppress stochastic processes [74]. Our AOB community results showed that phylogenetic turnover is associated with geographical distance instead of environmental variables; therefore, deterministic processes are dominant compared to stochastic processes, which supports the argument given by [75]. Other studies have shown that as geographic distance increases, if phylogenetic turnover also increases, it will signi cantly in uence deterministic processes [74]. In this study, the heterogeneous community selection of AOA was strongly associated with geographical distance. βNTI correlations with environmental variables indicated that EC and AK were negatively related to the AOB community, while AOA community was partially negatively correlated with pH.
Spatiotemporal variations in ecosystem function and its relationship with environmental variables and abundance of AOA and AOB Nitri cation rate and nitrogen mineralisation rate are fundamental indicators for soil to supply nitrogen for plant growth and nitrogen transformation in the environment [76]. Previous studies indicated that ammonia oxidation was higher in acidic soils with AOA, while bacterial ammonia oxidation was higher in neutral or alkaline soils [77][78][79]. We found that pH decreased from alkaline to neutral with the reclamation of coastal salt marshes, and the relative abundance of the amoA gene and nitri cation rate was higher in archaea compared to the bacterial community. In addition to soil pH, the availability of nutritional gradients SOM and TN played important roles in the nitri cation rate [80][81][82]. The cultivation of crops and utilisation of fertiliser increased soil organic matter and total nitrogen, and a similar trend was observed in our study, suggesting that land management practices, i.e., reclamation of salt marshes to cultivated soil increased soil organic matter and total nitrogen. Both of these factors might contribute to the nitri cation rate in agricultural soil [83]. Another study indicated that the ammonium concentration is the key factor determining the relative contribution of AOB and AOA in nitri cation in agricultural soils [84]. Our study presented contradictory ndings that ammonium concentration was not correlated with nitri cation rate (Table S1). Interestingly, ammonium concentration was also not associated with AOA community distribution and amoA gene abundance (Table S1 & S5), which might be due to the potential nitri cation rate that is also not correlated to ammonium concentration, as AOA plays a signi cant role in nitri cation in agricultural soils compare to AOB [21].
In a grassland ecosystem, nitrogen mineralisation is affected by biotic (animal, soil microbes, and plant) and abiotic (environmental variations and anthropogenic disturbance) factors [85,86]. In previous studies on grassland soils, the net nitrogen mineralisation and net nitri cation rate showed a similar trend [87]; our contradictory ndings showed that the net nitrogen mineralisation rate showed an opposite trend in reclaimed soil. Previous studies revealed that the AOB community was positively correlated with nitrogen mineralisation rate in agricultural land [88], but we found that net nitrogen mineralisation rate was not correlated with bacterial amoA gene abundance. In long-term reclamation, for AOA, amoA gene abundance and nitri cation rates accumulated, indicating that AOA was the dominant nitrifying community. Notably, the comammox organism's discovery and their high amoA gene richness in different soils suggested that they are functionally related to soil nitri cation, thereby modifying the contribution of ammonia-oxidising archaea and bacteria to the nitri cation rate [89]. Thus, net nitrogen mineralisation and potential nitri cation rates were in uenced by environmental factors, reclamation years, and AOA community.

Conclusion
We investigated temporal and spatial variations in an ammonia oxidiser community from a coastal salt marsh (0 years) and long-term (1000 years) reclaimed soils and observed the heterogenicity of diversity, composition, and phylogenetic structure of AOA and AOB communities. During the 1000 years of reclamation, there was a clear evolution of the amoA gene harboured in AOA and AOB communities. The soil ecosystem was signi cantly destroyed by the reclamation of coastal wetlands, which also altered soil physicochemical parameters. We investigated the relative importance of stochastic and deterministic processes in shaping AOA and AOB communities. In the assembly of ammonia oxidiser groups, deterministic processes were dominant over stochastic processes. The major soil physicochemical parameter changes included a decrease in pH, EC, and NH 4 + -N and increase of SOM, AN, and TN. A decrease in substrate concentration and pH subsequently led to a decrease in ammonia oxidiser prokaryotic communities. Temporal differences in soil ecosystem functioning, including nitri cation potential rate and net nitrogen mineralisation rate, were also found, suggesting that reclamation from coastal salt marshes to cultivated soil increased the soil nitri cation activity. These ndings provide a better understanding of how soil N cycling, reclamation of coastal salt marshes, long-term land management, and cultivation of crops affect ammonia-oxidising communities, including assembly dynamics, amoA gene abundance, and community distribution patterns.

Competing interests
The authors declare that they have no competing interests.    Bray-Curtis distance-based non-metric multi-dimensional scaling plot at the OTU level in reclaimed soils.

Figure 5
Boxplot indicating phylogenetic dissimilarity based on βMNTD at temporal scales. Different letters above boxes indicate signi cant differences (Duncan's test, P < 0.05). Scatter plots indicate the correlation between βMNTD values and spatial parameters (geographical distance and environmental parameters).
Spearman correlation (r) and probability values are provided.