Species, Abundance and Function of Ammonia-oxidizing Archaea in Inland Waters across China

Ammonia oxidation is the first step in nitrification and was thought to be performed solely by specialized bacteria. The discovery of ammonia-oxidizing archaea (AOA) changed this view. We examined the large scale and spatio-temporal occurrence, abundance and role of AOA throughout Chinese inland waters (n = 28). Molecular survey showed that AOA was ubiquitous in inland waters. The existence of AOA in extreme acidic, alkaline, hot, cold, eutrophic and oligotrophic environments expanded the tolerance limits of AOA, especially their known temperature tolerance to −25 °C, and substrate load to 42.04 mM. There were spatio-temporal divergences of AOA community structure in inland waters, and the diversity of AOA in inland water ecosystems was high with 34 observed species-level operational taxonomic units (OTUs; based on a 15% cutoff) distributed widely in group I.1b, I.1a, and I.1a-associated. The abundance of AOA was quite high (8.5 × 104 to 8.5 × 109 copies g−1), and AOA outnumbered ammonia-oxidizing bacteria (AOB) in the inland waters where little human activities were involved. On the whole AOB predominate the ammonia oxidation rate over AOA in inland water ecosystems, and AOA play an indispensable role in global nitrogen cycle considering that AOA occupy a broader habitat range than AOB, especially in extreme environments.

Community structure and diversity of AOA in inland waters. We examined the diversity of AOA and differences among AOA community structures in the inland water ecosystems. A database of 729 archaeal amoA gene sequences from this study was constructed and the community structures of AOA were investigated. After the diversity analysis using DOTUR software, a total of 216 unique OTUs (based on a 2% cutoff) were recovered. The observed Chao1 richness estimate and Shannon diversity index were as high as 324.90 and 4.88, indicating a high biodiversity of AOA. Thirty-four species-level OTUs (each representing a specific AOA species), were obtained from our sequences, using 85% amoA gene sequence identity as a species threshold 18 . The evolutionary relationships between these species and published amoA gene sequences are shown in Fig. 2. The AOA species in Chinese inland water ecosystems were widely distributed in three AOA lineages (group I.1b, I.1a, and I.1a-associated), which are the known lineages of AOA together with the ThAOA (Thermophilic AOA) group 13 . Most of the sequences clustered into the group I.1b (28 species representing 563 sequences), while a small quantity were affiliated with the group I.1a (4 species representing 135 sequences), and the remaining two species (representing 31 sequences) were assigned to the group I.1a-associated.
Among the 34 species obtained in this study, three were found to be closely related to Candidatus Nitrososphaera gargensis (species-3), Nitrososphaera viennensis EN76 (species-7) and Ca. Nitrosotalea devanaterra (species-30) respectively, and one (species-33) was similar to both Ca. Nitrosoarchaeum limnia SFB1 and Ca. Nitrosoarchaeum koreensis MY1 (with identities > 85.0% on the nucleotide level). This indicates that the above five known species of AOA exist in Chinese inland water ecosystems. The other 30 AOA species found in inland water sediments had not been previously characterized, which  Table 1. Different colors represent different types of inland waters as shown in the legend. The map were come from web of "Data Sharing Infrastructure of Earth System Science" http://www.geodata.cn. All of the maps used in the manuscript are free. Continued reminds us that our knowledge of AOA in natural environments is incomplete, and more unidentified AOA species probably exist 13 . Most of our sampled inland waters harbored species-12 (24 inland waters) and species-3 (21 inland waters). 18 species existed in three or fewer of the sampled inland waters, and 11 species were specific to a particular sampling site (Fig. 2). Although some AOA species were capable of living in a broad range of environments, most of them were selective and quite a few were sensitive to different environments. The distribution of AOA species in inland water ecosystems was nonrandom on a geographical scale. As a result, the community structure of AOA varied greatly among inland waters (Supplementary Table S1). This is also a reason for the high biodiversity of AOA in inland water ecosystems.
Besides the spatial distribution of AOA in Chinese inland water ecosystems, the temporal distribution of AOA was studied in the Pearl River. The AOA community structure in sediments of the Pearl River in winter and summer were analyzed (Fig. 3). Results showed that a population shift in AOA over different seasons occurred in the Pearl River, with most of the AOA in summer clustered into group I.1b, while winter AOA species distributed evenly in group I.1a and I.1b. In summary, this study documented the spatio-temporal divergences of AOA community structure.
AOA in extreme environments. pH. Species-20 and 30 were detected in the sediments of the Tieshanping River with a pH as low as 3.9 (Table 2), indicating the strong acidity tolerance of these two species. Species-30 was identified as Ca. Nitrosotalea devanaterra, which has been detected in acidic soil with a pH of 4.5 and incubated under a pH of 4.0-5.5 19 . This study extended their tolerance limit to pH of 3.9. Ca. Nitrosotalea devanaterra (Species-30) was found to dominate the AOA population (75%) in the sediments of the Tieshanping River, with the remaining population belonging to species-20. In the soil of the Jiaxing paddy field with a pH of 6.55, Ca. Nitrosotalea devanaterra were also detected, accounting for a small part of the AOA population (20%). In other inland waters with a higher pH, no Ca. Nitrosotalea devanaterra were detected. The results provided evidence that Ca. Nitrosotalea devanaterra may exist only in acidic and neutral environments. The simple composition of the AOA population in the sediments of the Tieshanping River also reflected that only a few specialized AOA species could survive in environments with extreme acidity.
Substrate. The sediment samples in this study had a wide gradient of ammonium concentrations from 0.10 mM (Tianchi Lake), to 42.04 mM (Baiyangdian Lake). Five uncharacterized AOA species belonging to group I.1b were detected in the oligotrophic Tianchi Lake ( Table 2), indicating that these five species can live in environments with very low substrate concentration (0.10 mM).
Ca. Nitrososphaera gargensis, N. viennensis EN76 and four other species were observed in the sediments of Baiyangdian Lake (Table 2), indicating their strong tolerance to high ammonium concentration (up to 42.04 mM). Ca. Nitrososphaera gargensis was detected in environments with ammonium concentration of 5.6 mM 22 , and N. viennensis EN76 was incubated under ammonium concentration up to 15 mM 20 . This study expands their tolerance to high substrate content to 42.04 mM.
Temperature. Eight AOA species were observed in Aydingkol Lake with a high surface temperature (up to 75 °C) and six in the Songhua River with a low temperature (as low as − 25 °C) ( Table 2). Species-12 and 17, Ca. Nitrosoarchaeum limnia and Ca. Nitrosoarchaeum koreensis existed in both of the extreme-temperature environments, which expands their recognized temperature tolerance to between − 25 °C and 75 °C. In addition, the moderately thermophilic Ca. Nitrososphaera gargensis, which had been detected in thermal spring microbial mats at 45 °C 22 , was also observed in Aydingkol Lake, which

Abundance of archaeal and bacterial ammonia oxidizers.
Results of qPCR showed that the abundances of AOA ranged from 8.5 × 10 4 to 8.5 × 10 9 copies g −1 dry sediment in Chinese inland waters. As the counterpart of AOA, AOB had the abundance of 2.9 × 10 3 to 4.3 × 10 9 copies g −1 (Fig. 4A). The relative abundance of AOA compared with AOB varied throughout the sampling sites (Fig. 4). and AOA outnumbered AOB in almost all the lakes and rivers, while the opposite was true in the other inland waters with strong human activities such as paddy fields, reservoirs, polluted groundwater, tidal land and constructed wetland. This result implies that AOA predominates the ammonia oxidizer population in inland waters with less human activities, while AOB dominates in inland waters with more human disturbance. Spearman correlation analysis between archaeal & bacterial amoA abundance and environmental variables revealed that different parameters were related to the size of the AOA and AOB population, indicating a niche differentiation between these two groups. Archaeal amoA abundance had an obvious negative correlation with TC (r = − 0.755, p < 0.01) and TS (r = − 0.748, p < 0.01), while bacterial amoA abundance had an obvious positive correlation with NO x content (r = 0.497, p < 0.01), TN (r = 0.529, p < 0.01) and TS (r = 0.484, p < 0.01) (Supplementary Table S2).
Contributions of AOA and AOB to microbial ammonia oxidation. Because AOA were found to be ubiquitous and abundant in Chinese inland water ecosystems, we expected that they would play a significant role in ammonia oxidation. To examine this assumption, potential nitrification rates (PNRs) were measured to estimate the combined activity of archaeal and bacterial ammonia oxidizers. The values ranged from 0 to 146.91 nmol N g −1 h −1 (n = 72, Fig. 4A). Spearmen correlation analysis between PNR, archaeal & bacterial amoA abundance and environmental variables showed bacterial amoA abundance and pH were significantly correlated with PNR (Supplementary Table S2). The archaeal amoA abundance showed no correlation with PNR. Multiple linear regression (stepwise regression) on PNR also showed that bacterial amoA abundance was the most determining variable for nitrification followed by pH (bacterial amoA abundance explained 29.5% of the variability of PNR, while pH explained 3.2%, n = 45, Supplementary Table S3). These results led to a possibility that AOB, rather than AOA, contribute more to nitrification in Chinese inland water ecosystems. This hypothesis was further tested in the Tiaoxi River on a spatio-temporal scale. Sediment samples from different seasons were collected from four sites in the Tiaoxi River ( Supplementary Fig. S1) under different ammonia loading levels (Supplementary Table S4). The community structure of AOA (Fig. 5A) and abundance of archaeal & bacterial amoA genes (Fig. 5B) were detected. AOA from different sites showed quite different community structures, although species-12 and 25 existed in all of the four sites. AOA outnumbered AOB in almost all of the sediment samples, while the correlation analysis showed that the abundance of AOB not AOA had a significant correlation with PNR (Supplementary Table S5), indicating that AOB might contribute more to ammonia oxidation than AOA on spatio-temporal scale.

Discussion
To the best of our knowledge, this is the first report of the large-scale occurrence, ecological behaviors, biodiversity and potential function of AOA in inland waters. They were found to be ubiquitous, have high biodiversity and diverse community structure, have strong adaptability to extreme conditions, abundant and play a less important role in ammonia oxidation than AOB in inland water ecosystems.
The ubiquity of AOA in inland waters was demonstrated through sediment samples (n = 100) from 28 inland waters, including six sites with extreme pH, temperature or substrate conditions. The ubiquity of AOA may be explained by their high diversity and strong tolerance to extreme conditions. High diversity helps increase the capacity to adapt to environmental change, and the strong tolerance to extreme conditions extends the occurrence to a largest scope 1,26 .
The similar phenomenon also appears in other ecosystems. AOA is also ubiquitous in soil ecosystems with quite high diversity 27 . There are few studies focused on the biodiversity of AOA in marine ecosystems on a large scale, but the observed unique archaeal amoA sequences specific to an individual sample location has indicated the high biodiversity of AOA in marine ecosystems 28 . For the whole natural environment, the considerable global AOA diversity was observed in a phylogenetic analysis on 12356 publicly available archaeal amoA sequences from different ecosystems 18 . These results demonstrate the high diversity of AOA in natural environments. As to the adaptability to extreme conditions, AOA were found in a wide pH range (2.5 to 9.0 24,29 ), and temperature range (0.2 to 97 °C [23][24][25], indicating the great potential of AOA in adapting to extreme acidity and alkalinity, and extreme low and high temperatures. In addition, AOA had a preference for low ammonium content 30,31 , and survived extreme low ammonium concentrations (≤ 10 nM 20 ) with half-saturation constant (Km) of 133 nM total NH 4

+32
. The detection of AOA in extreme conditions in this study broadened their known limits, especially those of identified species ( Table 2). The cold-tolerance of AOA was expanded to − 25 °C, and the tolerance to high substrate content was extended to 42.04 mM. The results point to a function of AOA in extending ammonia oxidation to a much greater range of habitats. As a result, AOA are widely distributed in marine, soil and inland water environments 1,13,33 , and exist in extreme conditions such as hot springs 34 , Antarctic waters 35 , acid soils 19 and oligotrophic environments 32 . High AOA diversity imply a strong capacity for ammonia oxidation in various ecosystems and an important role in global nitrification.   Fig. 2.
In this study, AOA were found to numerically outcompete their counterpart AOB in inland waters with less human disturbance, while AOB dominated in inland waters affected strongly by human activities. Evidence can also be found in previous studies. AOA appear to be numerically dominant in marine environments [36][37][38] and soils 17,[39][40][41] , while in environments involved with more human activities like fertilized agricultural soils [42][43][44] , polluted wetlands 2 , wastewater treatment plants 45 and bioreactors 46,47 , AOB outnumbered AOA. The environmental changes caused by the human activities seem to create a position of advantage for AOB over AOA. This also demonstrated the niche differentiation between AOA and AOB. The factors influencing the relative abundance of AOA to AOB were various, including the ammonia concentration 20,48,49 , the dissolved oxygen concentration 50 and the pH 49 . Considering the complex status of various ecosystems, the relative predominance of AOA and AOB may be not affected solely by a single parameter, but by a combination of influencing factors 51 .
With the increasing understanding of AOA, the function of AOA & AOB in ammonia oxidation was called into question. This study, based on a large number of samples, documented that AOB contributed more to nitrification than AOA in Chinese inland water ecosystems. Even so, the function of AOA can't be ignored, especially in extreme environments. In other small-scale studies, different results have been obtained. For example, nitrification was suggested be driven by bacteria rather than archaea in six nitrogen-rich grassland soils in New Zealand 30 and in sediments of four nitrogen-rich wetlands 2 . In six estuarine sediments, PNRs increased as the abundance of archaeal amoA increased, rather than bacterial amoA 52 , and in the Black Sea water column nitrification was mainly controlled by archaeal amoA expression in the lower oxic zone 53 . The relative importance of AOA vs. AOB in ammonia oxidation needs more researches. The in situ measurement of microbial ammonia oxidation rate and separation of the role of AOA & AOB are two key items. In this study we can make a conclusion that AOA may play an indispensable role in global nitrogen cycle considering that AOA occupy a broader habitat range than AOB, especially in extreme environments.

Methods
Study site background. A total of 100 sediment samples from 28 inland water ecosystems, including lakes, rivers, paddy fields, reservoirs, groundwater, swamp, tidal land and constructed wetland were investigated in Chinese territory (23 to 46° N and 86 to 130° E). Details for every sampling site including location, background and number of samples are listed in Table 1. Other information including the nitrogenous compounds content and some physicochemical characteristics of the sampling sites is listed in Supplementary Table S6.
Among the sampled inland waters, there were six with extreme conditions. Aydingkol Lake had the highest surface temperature (75 °C) and the Songhua River had the lowest temperature (− 25 °C). Tianchi Lake was an oligotrophic lake (NH 4 + as low as 0.10 mM) and Baiyangdian Lake was hyper-eutrophic (NH 4 + up to 42.04 mM). The sediment in Tieshanping River had an acidic pH (as low as 3.9) and Tarim River had an alkaline pH (up to 8.9). These extreme conditions were all relatively stable. Furthermore, sediments from the Pearl River and the Tiaoxi River were sampled in different seasons to verify the results on a spatio-temporal scale.
Surface sediments (0-8 cm) were collected from each sampling site in 2012 and 2013. The samples were placed in sterile plastic bags, sealed and transported to the laboratory on ice. One part of each sample was used for the analysis of physicochemical characteristics immediately after arrival, one part was incubated to measure the potential nitrification rates (PNRs), and the rest was stored at − 80 °C for later DNA extraction and molecular analysis. DNA Extraction, PCR, Cloning and Sequences Analysis. DNA was extracted from about 0.3 g sediment using FastDNA ® Spin Kit for Soil (MP Biomedicals, USA). Concentrations of the extracted DNA were determined by spectrophotometric analysis on a NanoDrop 2000 UV-Vis Spectrophotometer (Thermo Fisher Scientific, USA) and the quality was checked by electrophoresis on a 1% (weight/volume percent) agarose gel. The archaeal amoA (ammonia monooxygenase α-subunit) gene was amplified using primer pairs Archaea-amoAF/Archaea-amoAR according to Francis et al. 28 with an annealing temperature of 53 °C. The sequences of primers and thermal profiles used in this study were shown in Supplementary Table S7. All PCR reactions were performed with the Ex Taq TM polymerase (Takara Dalian, China).
PCR amplified fragments were ligated directly using the pGEM-T ® Easy Vector Systems (Promega, USA) according to the manufacturer's instructions, and then transformed to Escherichia coli JM109 competent cells for cloning. Selected clones were sequenced using T7 forward primers targeting vector sequences adjacent to the multiple cloning sites by an ABI PRISM 3730XL automated-sequencer (Applied Biosystems, USA). Sequences of archaeal amoA genes obtained in this study were deposited in the GenBank under the accession numbers (HM637849-HM637867, HQ538539-HQ538560, HQ538562, JF439021, JF439023-JF439028, JF439030-JF439044, JF439046-JF439066, KC108794-KC108815, KP167639-KP168260). All the sequences were aligned using the ClustalX 1.83 program. Phylogenetic analysis was performed using Mega 5.0 software 54 . Phylogenetic trees were constructed by neighbor-joining (NJ) with the Maximum Composite Likelihood and the robustness of tree topology was tested by bootstrap analysis with 1,000 replicates. The calculation of operational taxonomic unit (OTU) and diversity indices, including Chaol, richness estimate and Shannon diversity index, were generated by DOTUR by employing the furthest neighbor approach 55 . Quantitative PCR analysis. Quantitative PCR (qPCR) was performed on an ABI 7300 real-time PCR instrument (Applied Biosystems, USA) with a SYBR Green qPCR kit (Takara Dalian, China). The qPCR thermal profiles of archaeal and bacterial amoA genes were performed with primers Archaea-amoAF/Archaea-amoAR and amoA-1F/amoA-2R 7 , with the annealing temperatures of 53 °C  ) were extracted from the fresh sediment samples with 2 M KCl solution and measured using a Continuous Flow Analyzer (SAN plus, Skalar Analytical, the Netherlands). The other physicochemical characteristics (total nitrogen (TN), total carbon (TC), total sulfur (TS), total phosphorus (TP)) of the sediment samples were also measured according to Bao 56 . The pH was determined using dry sediments after mixing with water at a ratio (dry sediment/water) of 1:5, and the organic matter was determined by K 2 Cr 2 O 7 oxidation method. All analyses were performed on triplicate samples. Potential nitrification rates (PNRs). The potential nitrification rate was measured using a chlorate inhibition method with minor modifications 57 . Briefly, 3.0 g of fresh sediment was added to 50-mL centrifuge tube containing 20 mL phosphate buffer solution (NaCl 8.0 g L −1 , KCl 0.2 g L −1 , Na 2 HPO 4 0.2 g L −1 , NaH 2 PO 4 0.2 g L −1 , pH = 7.4). (NH 4 ) 2 SO 4 was added to the incubation system to a final concentration of ammonium similar to the in situ condition. Samples were run in triplicate with and without allylthiourea (ATU, an inhibitor of nitrification process) (100 μ M final concentration) to identify the difference between NO 2 − accumulation by aerobic microbial nitrification and chemical ammonia oxidation 58  Software Statistics). Mann-Whitney U test and non-parameter paired test were used respectively for the comparison of two data groups and paired data. The correlations between different types of variables were computed by Spearman correlation analysis. Stepwise linear regression analysis was used to determine the most important factor for a dependent variable. Unless otherwise specified, the level of significance in this study was α = 0.05. Graphing was achieved using Origin 8.0 software.