Ammonia-oxidizing archaea respond positively to inorganic nitrogen addition in desert soils

In soils, nitrogen (N) addition typically enhances ammonia oxidation (AO) rates and increases the population density of ammonia-oxidizing bacteria (AOB), but not that of ammonia-oxidizing archaea (AOA). We asked if long-term inorganic N addition also has similar consequences in arid land soils, an understudied yet spatially ubiquitous ecosystem type. Using Sonoran Desert top soils from between and under shrubs within a long-term N-enrichment experiment, we determined community concentration-response kinetics of AO and measured the total and relative abundance of AOA and AOB based on amoA gene abundance. As expected, N addition increased maximum AO rates and the abundance of bacterial amoA genes compared to the controls. Surprisingly, N addition also increased the abundance of archaeal amoA genes. We did not detect any major effects of N addition on ammonia-oxidizing community composition. The ammonia-oxidizing communities in these desert soils were dominated by AOA as expected (78% of amoA gene copies were related to Nitrososphaera), but contained unusually high contributions of Nitrosomonas (18%) and unusually low numbers of Nitrosospira (2%). This study highlights unique traits of ammonia oxidizers in arid lands, which should be considered globally in predictions of AO responses to changes in N availability.

ammonia oxidation, amoA, arid land, niche differentiation, nitrogen enrichment, Thaumarchaeota

INTRODUCTION

Since the early and influential work of Winogradsky (1890), bacteria were thought to be the only biological agents of ammonia oxidation (AO). However, the deployment of molecular detection techniques in the last three decades has revealed that Thaumarchaeota in the Archaea domain contribute to AO as well (Konneke et al., 2005). High-throughput sequencing and molecular-fingerprinting studies show the presence of genes attributable to diverse groups of ammonia-oxidizing archaea (AOA) and bacteria (AOB) in a wide variety of environments (Purkhold et al., 2000; Leininger et al., 2006; Prosser and Nicol 2008; Pester et al., 2012). Even though AOA outnumber AOB in many ecosystems (Leininger et al., 2006; Adair and Schwartz 2008; Wessen et al., 2010), this dominance does not always equate to AOA contributing to AO more than AOB (Jia and Conrad 2009; Di et al., 2009; Adair and Schwartz 2011). It remains unclear why the abundance of AOA is often unrelated to AO rates (Shen et al., 2008; Wessen et al., 2010). AO fluxes may depend not only on population size but also on community composition due to differential substrate affinities and ecophysiological sensitivities among and within the AOA and AOB (Kowalchuk and Stephen 2001; Bollmann, Bar-Gilissen and Laanbroek 2002; Schleper and Nicol 2010).

A review of literature reveals that mixed ammonia-oxidizer communities are often dominated by one particular phylotype (Prosser 1989; Kowalchuk and Stephen 2001; He, Hu and Zhang 2012; Zhalnina et al., 2012). However, it is uncertain whether and how this outcome is determined by environmental properties. For instance, while culture work shows that Nitrosomonas strains (AOB) prefer ammonia-rich conditions (Taylor and Bottomley 2006), Nitrosospira-related clusters (AOB) commonly outnumber Nitrosomonas spp. in fertilized soils and also in low-NH₄⁺, pristine soils (Jordan et al., 2005; Chu et al., 2007). Additionally, AOB are preferentially enriched after inorganic nitrogen (N) fertilization in the ecosystems studied to date—such as in relatively low pH soils that receive high rates of precipitation or water inputs—while AOA may respond positively only in cases when NH₃/NH₄⁺ is supplied through organic matter mineralization (Offre, Prosser and Nicol 2009; Hatzenpichler 2012; He et al., 2012; Levicnik-Hofferle et al., 2012). These examples suggest that indeed changes in N availability such as through N deposition or fertilization may control AO rates in soils through community compositional shifts (Avrahami and Bohannan 2007; Tourna, Freitag and Prosser 2010; Prosser and Nicol 2012).

Arid environments are vastly underrepresented in the AO research literature (Johnson et al., 2005; Marusenko et al., 2013a; Sher, Zaady and Nejidat 2013), but there is reason to believe that arid lands may harbor populations with different adaptations compared to the more studied temperate soils. For example, arid soils are exposed to prolonged drought and rapid pulses of precipitation and nutrients (Schimel, Balser and Wallenstein 2007; Collins et al., 2008), which require complex and fast genetic regulation from soil microbes (Rajeev et al., 2013). Furthermore, arid soils are often alkaline and can reach up to 50°C in the summer. They are typically dry with low organic matter content and low N mineralization rates especially in non-vegetated areas between shrubs (Austin et al., 2004; Schade and Hobbie 2005), which may select for the most oligotrophic of ammonia oxidizers. These ubiquitous soils also experience intensive management, including watering and fertilizer inputs, both in agricultural and urban residential areas (Warren, Sud and Rozanov 1996; Davies and Hall 2010). As a result, anthropogenic activities and atmospheric deposition are altering resource availability and the N cycle in soils of water-limited environments (McCrackin et al., 2008; Hall et al., 2009, 2011; Marusenko, Huber and Hall 2013b).

Here, we tested the effect of long-term inorganic N addition on AO processes and ammonia-oxidizing microorganisms (AOM) in arid land soils, assessing the AO kinetics in bulk soil and characterizing AOA and AOB by sequencing the environmental amoA gene, which encodes a subunit of the ammonia monooxygenase (AMO) enzyme. We hypothesized that N addition would cause the absolute and relative abundance of ammonia oxidizers to shift from AOA-dominated in oligotrophic native (unfertilized) soils to AOB-dominated in NH₄⁺-rich conditions, as has been found in other soils. Consequently, this population replacement would enhance overall AO rates and cell-specific AO rates, but decrease affinity between the enzyme and the substrate. Using common patch types in arid lands, we further expected that the decline of AOA relative to AOB under N addition would be less dramatic in relatively fertile soils under shrubs than in areas away from plants.

MATERIALS AND METHODS

Study area description

Our site is in the northern Sonoran Desert at ∼620 m elevation in Lost Dutchman State Park, AZ, USA (coordinates: N 33.459372 S –111.484956), located east of the Phoenix metropolitan area and within boundaries of the Central Arizona–Phoenix Long-Term Ecological Research area (http://caplter.asu.edu). Soils are classified as Typic Haplargids, a subgroup of Aridisols. We measured soil AO rates and community parameters from two randomly assigned 20 m × 20 m plots, one that received N fertilizer as NH₄NO₃ (applied as solid by hand biannually at 60 kg N ha⁻¹ yr⁻¹ from 2005 to 2012) and another that served as an unfertilized control (see Hall et al., 2011 for further description about the plots). Nitrogen deposition in this area is 7.3 kg N ha⁻¹ yr⁻¹ (Cook 2014). Plant cover (∼60%) within our study plots is dominated by the native shrubs creosote bush (Larrea tridentata [DC.] Coville) and bursage (Ambrosia spp.). Plots did not contain any N-fixing trees. Mean annual temperature is 22.3°C, with the coldest and warmest months averaging 3.7 and 41.9°C, respectively (2005–2012; NCDC 2013). Mean annual precipitation is 272 mm but is highly variable year to year. Rainfall is bimodally distributed between summer monsoon events and low-intensity winter storms (WRCC 1985).

Sample collection

Surface soil samples were collected in late January of 2012, one month after winter storms. In each of the control and N addition plots, three soil samples were collected from each of two patch types to explore N treatment effects in typical desert environments: between plants (hereafter called ‘interplant’) and under canopies of the common shrub L. tridentata (‘under plant’). Mature/dark soil biological crusts were low in abundance within the plot area and were avoided for sampling. Early colonization by biocrust organisms is widespread in the region (Rosentreter, Bowker and Belnap 2007) but is not yet formed to visibility at our site locations. Each soil sample consisted of two 0–5 cm (depth) × 7 cm (diameter) cores taken 5 cm apart. In total, we collected 12 soil samples consisting of 3 replicate soil samples from each plot (treatment, control) and patch type (interplant, under plant) (3 × 2 × 2 = 12 samples). Soil samples were processed independently for all analyses (soil properties, AO rates, quantitative PCR, pyrosequencing).

Laboratory methods and soil properties

Following collection, samples were transported on ice to the lab, sieved to <2 mm and homogenized. Soils were at 3–5% soil moisture upon collection and were analyzed within 24 h for all soil properties and processes. Two subsamples (2 g each) from each homogenized soil sample were frozen in liquid N and stored at −80°C until DNA extraction within one month. Duplicate DNA extracts were combined prior to molecular processing methods to obtain one determination per sample.

Soils were processed for pH (1:2 soil to DI H₂O), water holding capacity (% WHC; gravimetrically), soil organic matter content (% SOM; loss on ignition), and extractable NH₄⁺, nitrite (NO₂⁻) and nitrate (NO₃⁻) content (2M KCl extraction, colorimetric analysis), following standard methods (Sparks et al., 1996; Marusenko et al., 2013b). Data reported for each of the three field replicates is an average of laboratory triplicates.

AO rates using the shaken-slurry assay

In situ net rates of potential AO were measured under various levels of N addition (see ‘ammonia oxidation kinetics’ below) using the shaken-slurry method (hereafter as ‘slurry AO rates’), in which oxygen and substrate diffusion is not limiting (Hart et al., 1994; Norton and Stark 2011). The direct product of AO was measured as NO₂⁻ accumulation after inclusion of chlorate (NaClO₃), a NO₂⁻-oxidation inhibitor (Belser and Mays 1980). Using NO₃⁻ as a proxy for AO was unsuitable since NO₂⁻ build-up is common in natural dryland conditions (Gelfand and Yakir 2008). The shaken-slurry assays contained 10 g soil in 100 mL solution of 0.015 mol·L⁻¹ NaClO₃, and 0.2 mol·L⁻¹ K₂HPO₄ and 0.2 mol·L⁻¹ KH₂PO₄ to buffer pH at 7.2. Slurries and no-soil blanks were continuously aerated in solution by mixing at 180 rpm on a reciprocal shaker in the dark. Homogenized slurry aliquots were removed at four time points over 6 h and amended with several drops of MgCl₂ + CaCl₂ (0.6 M) to flocculate soil particles. Aliquots were then centrifuged at 3000 × g and supernatant was filtered through pre-leached Whatman no. 42 ashless filters. The supernatants were stored at 4°C and analyzed within 24 h. Net rates of slurry AO were calculated as the linear increase in NO₂⁻ content from 0 to 6 h, measured colorimetrically using a Lachat Quikchem 8000 autoanalyzer. Consistent with the literature showing that metabolism of ammonia oxidizers can be activated and responsive to the environment at the scale of hours, especially for AOB (Wilhelm, Abeliovich and Nejidat 1998; Placella and Firestone 2013), NO₂⁻ accumulation in our assays was linear from 0 to 6 h.

AO rates in static incubation

As a secondary method to slurry AO rates, we also measured AO following various levels of N addition in a modified method using NaClO₃ inhibition in static, 48 h aerobic incubations of bulk soil (‘static AO rates’ from here on; Nishio and Fujimoto 1990; Hart et al., 1994; Low, Stark and Dudley 1997). Although substrate diffusion may be limited in aerobic incubations to fully quantify enzyme activity, we used this method to independently assess AO in conditions more representative of the upland desert environment compared to the shaken-slurry assays, which assesses aerated AO potential. Ten grams of soil was brought to 60% WHC using water and NaClO₃ (15 mM) in plastic cups. Soil in one cup was extracted at the onset and a second cup extracted after incubation for 2 days in the dark. Soils were extracted in 50 mL of 2 M KCl followed by shaking for 1 h and filtering through pre-leached Whatman no. 42 ashless filters. The extracts were stored at 4°C and analyzed colorimetrically within 24 h. Net rates of static AO were calculated as the increase in NO₂⁻ content between 0 and 48 h.

AO kinetics

The relationship between substrate availability and reaction rate can be measured to discern functional parameters of microbial communities. Although this is not a true enzyme kinetics study, we applied similar calculations to our concentration-response kinetics approach to model relationships for AO (slurry and static) in bulk soils based on the Michaelis–Menten equation (Martens-Habbena and Stahl 2011; Prosser and Nicol 2012), as has been successfully applied to soil assays (Koper et al., 2010):

\begin{equation*} V = (V_{{\rm max}} \times S)/(K_{\rm m} + S). \end{equation*}

In this equation, the NH₄⁺ concentration (S) and AO rate (V) are used to estimate the maximum AO rate (V_max) and half-saturation constant (K_m; inverse of enzyme and substrate affinity). To estimate AO kinetics under oligotrophic conditions in the shaken-slurry assay, we removed pre-existing NH₄⁺ from soils to obtain the least variable and lowest residual substrate availability (Widmer, Brookes and Parry 1989; Koper et al., 2010; Norton and Stark 2011). Prior to the shaken-slurry assay, 5 g soil was mixed in 45 mL of potassium phosphate solution and centrifuged at 3200 × g for 1 min before discarding the N-containing supernatant. The resulting soil pellet from two preparations was combined to compose 10 g total soil from each plot (treatment, control), patch type (interplant, under plant) and soil sample replicate (× 3). Inorganic N was then supplemented as (NH₄)₂SO₄ mixed with DI water to eight final concentrations in the slurry ranging from 0 to 22.5 mM. In total, we evaluated AO rates using 96 different soil preparations (12 soil samples × 8 NH₄⁺ concentrations) per method (shaken-slurry assay, static incubation). In the static incubations, we excluded the N removal step as to minimize soil disturbance. Soils were supplemented with (NH₄)₂SO₄ in solution to produce final concentrations ranging from 0 to 50 μg NH₄⁺-N g⁻¹ (0–22.5 mM). The 0 μg NH₄⁺-N g⁻¹ addition (only includes pre-existing NH₄⁺) was used to estimate background net rates of AO. As a rough indicator of the N addition effect on the relative importance of NH₄⁺ mineralization and nitrification, we also measured the net rate of NH₄⁺ gain (production processes dominate) and loss (consumption processes dominate) during the static incubation experiment. In the assay, some of the NH₄⁺ consumption processes are likely minimized due to sieving of soil (exclusion of large NH₄⁺-assimilating plant roots) and lower laboratory temperature compared to natural conditions (reduced volatilization).

DNA extraction and purification

DNA was extracted using three freeze–thaw cycles followed by 30 min incubation at 50°C with proteinase K and silica bead beating for chemical and mechanical cell lysis (Garcia-Pichel, Lopez-Cortes and Nubel 2001). The lysate was purified by phenol:chloroform:isoamyl alcohol (25:24:1) extraction, followed by DNA precipitation in 100% ethanol for 12 h at –80°C. DNA concentration and quality was assessed on an agarose gel stained in ethidium bromide and imaged using a Fluor-S Multi-Imager (Bio-Rad Laboratories, CA, USA) with an EZ Load Precision Molecular Mass Standard (Bio-Rad). Bands of DNA were excised from a low-melt agarose gel, homogenized with a tip in a microcentrifuge tube, allowed to diffuse out into sterile H₂O for 12 h and followed by 15 min centrifugation to collect DNA in the supernatant.

Quantitative PCR

DNA was used for quantitative PCR (qPCR) with the following amoA primers: CrenamoA616r (GCCATCCABCKRTANGTCCA; Tourna et al., 2008) and CrenamoA23f for the AOA (ATGGTCTGGCTWAGACG); and amoA1f mod (GGGGHTTYTACTGGTGGT; Stephen et al., 1999) and AmoA-2R’ for the AOB (CCTCKGSAAAGCCTTCTTC; Okano et al., 2004; Junier et al., 2008). qPCR reactions contained 10 μL iTaq SYBRGreen Master Mix (Bio-Rad), 250 nM final concentration of each primer (AOA or AOB), 1 ng of environmental DNA and molecular grade H₂O to bring each reaction to a final volume of 20 μL. The reaction conditions were as follows: initial denaturation for 150 s at 95°C followed by 45 cycles of 15 s at 95°C, 30 s at 55°C and 30 s at 72°C, and a final dissociation step to obtain the melting curve at 95°C, 60°C and 95°C for 15 s each. Standard curves were generated using templates from Nitrosomonas europaea ATCC 19718 (bacterial amoA; R² = 0.99) and a putative AOA clone (archaeal amoA; R² = 0.99) for a dilution series spanning 10²–10¹⁰ gene copies per reaction. Melting curves were checked to verify the quality of each reaction, and to ensure the absence of primer dimers. We report only determinations for which C_t values could be interpolated within our standard curves. Each amoA abundance value (number of gene copies) reported is an average of analytical triplicate qPCR reactions of the same DNA extract.

Pyrosequencing

Purified DNA extracts were shipped to a commercial laboratory for standard PCR and bTEFAP pyrosequencing (Dowd et al., 2008). Commercial primers for PCR were amoA-1F (GGGGTTTCTACTGGTGGT; Rotthauwe, Witzel and Liesack 1997) and amoA-2R for AOB (CCCCTCKGSAAAGCCTTCTTC), and Arch-amoAF (STAATGGTCTGGCTTAGACG; Francis et al., 2005) and Arch-amoAR for AOA (GCGGCCATCCATCTGTATGT) used with a HotStarTaq Plus Master Mix Kit (Qiagen, CA, USA). PCR conditions were as follows: 180 s at 94°C followed by 28 cycles of 30 s at 94°C, 40 s at 53°C and 60 s at 72°C, and final elongation for 5 min at 72°C. PCR amplicons were mixed in equal concentrations and purified using Agencourt Ampure beads (Agencourt Bioscience Corporation, MA, USA). Sequencing utilized Roche 454 FLX titanium instruments and reagents.

Bioinformatics and phylogenetic analyses of amoA

Pyrosequencing data were processed and analyzed in Qiime (Caporaso et al., 2010b), with necessary pipeline adjustments to process functional gene data (i.e. amoA) as described in detail in the notes and script file (http://www.yevmarusenko.com/research/Marusenko_Qiime.txt). Sequences (452 bp long) were clustered into operational taxonomic units (OTUs) using UClust (Edgar 2010). Representative sequences (one per OTU) were aligned with Pynast (Caporaso et al., 2010a). Based on nomenclature classification for AOA in Pester et al. (2012) and for AOB in Koops et al. (2006), a taxonomic assignment was made for each OTU using a template reference database created from sequences of known pure isolates, enrichments, and other characterized AOA and AOB from previous studies (reference database available at website mentioned above). Groups of sequences were clustered at 97% nucleotide similarity to be inclusive of OTUs at fine levels of resolution for phylogenetic and statistical analyses that otherwise may be missed at lower identity thresholds. For AOA, we excluded one replicate each in the interplant control and N addition samples because of their poor quality of the pyrosequencing data. The minimum number of high-quality sequences after filtering was 525 for AOA and 950 for AOB, with sufficient rarefied analysis producing 179 OTUs for AOA and 325 OTUs for AOB (total number of sequences >200 bp prior to quality filtering: AOA, 9830; AOB, 29 569). Phylogenetic analyses were carried out on a single alignment file (separately for AOA and AOB) that included sequences from our Qiime pipeline, as well as the reference sequences described above. All sequences were combined and realigned using default parameters for muscle and analyzed by the tree-building module of the MEGA 5 software with the following parameters: Neighbor-joining method, Jukes–Cantor nucleotide substitution model, 100 bootstrap replicates, uniform rates among sites and pairwise gap-data deletion (Tamura et al., 2011). Raw sequence reads for the entire project have been deposited in the Sequence Read Archive at NCBI with accession number SRX738968 for the AOA data and SRX739281 for the AOB data.

Statistics

Statistical tests were carried out using Qiime for α and β diversity measures on processed pyrosequencing data, while all other analyses were in SPSS (v20.0 Windows). All soil properties, AO rates and amoA abundance data were tested for linear model assumptions in SPSS using normal probability plots (for normality) and Levene's test (for equal variance), and transformed (natural log) when necessary. Individual two-way analysis of variance (ANOVA) tests were used to evaluate the effects of plants (‘patch’) and N addition (‘treatment’) on the following dependent variables: amoA gene abundance (per g soil and per ng extractable DNA, separately for AOA and AOB), AOA to AOB ratio, slurry V_max AO rates, static V_max AO rates, amoA-copy-specific AO rates, net NH₄⁺ change (averaged across supplemented NH₄⁺ concentrations) and each of the soil properties. Significant interactions between patch and treatment were evaluated further using one-way ANOVA (α = 0.025). The copy-specific AO rates were calculated using V_max AO rate (slurry and static) divided by the number of amoA gene copies per g soil. We used bivariate Pearson correlations to assess relationships between soil properties vs community parameters (amoA data and AO rates) across all samples. We used linear regression analyses to assess relationships between amoA gene abundance at the domain level vs V_max AO rates and also analyzed amoA gene abundance of individual OTUs vs soil properties and AO rates. In Qiime, we tested for the effect of N addition on OTU-based communities separately for AOA and AOB per patch type, using only strictly relevant diversity metrics (Lozupone and Knight 2005; Caporaso et al., 2010b): α diversity [Shannon's diversity, observed richness and phylogenetic diversity (PD)] and β diversity [weighted and unweighted Unifrac, the multivariate group dispersion analogue of Levene's test (PERMDISP) and analysis of similarity].

RESULTS

Effects of N addition on the abundance of amoA genes and the kinetics of AO

To aid in interpretation of long-term N addition effects on amoA gene abundance and AO rates, we considered the influence of soil properties and the relative importance of fertilizer N vs ammonification as a possible NH₄⁺ source. Long-term N addition clearly resulted in an accumulation of NO₂⁻, NO₃⁻ and NH₄⁺, regardless of patch type (Table 1). Also as expected, SOM was higher in soil under plants than between plants. N addition slightly acidified these alkaline soils—an effect known to worsen conditions for AO (Arp and Stein 2003)—and yet AO rates still increase in these N-amended desert plots (Table S1, Supporting Informartion). Both types of AO rates we measured (maximum static and slurry AO rates) were strongly predicted by pH (negative correlation) and SOM (positive correlation). Background AO rates measured in unamended incubations, however, were most strongly and positively related to NH₄⁺ concentration across all patch types and N treatments. Additionally, N addition significantly increased net rates of NH₄⁺ loss in both interplant and under plant patch types (patch, P = 0.56; treatment, P < 0.01). These data suggest that N addition stimulated NH₄⁺ loss from consumption processes (e.g. NH₃/NH₄⁺ oxidation, microbial immobilization) relatively more than it increased NH₄⁺ concentrations from organic N mineralization.

Table 1.

Soil characteristics from plots used in this study.

		pH		WHC (%)a		SOM (%)b		NO2– (μg N·g−1)		NO3– (μg N·g−1)		NH4+ (μg N·g−1)
Soil patch typec	Treatmentd	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Interplant	Control	8.39	0.07	32.0	4.5	2.02	0.57	0.13	0.08	1.40	0.41	0.42	0.17
Interplant	N addition	8.21	0.08	35.1	2.8	2.45	0.44	1.58	0.71	36.10	21.37	22.47	21.31
Under plant	Control	8.25	0.11	42.1	3.3	3.26	0.37	0.03	0.01	4.50	4.63	1.06	0.04
Under plant	N addition	8.11	0.06	45.2	6.9	3.59	0.55	0.16	0.19	29.12	5.11	12.77	9.39
		Two-way ANOVA results, P value
Patch x Treatment		0.651		0.992		0.868		0.171		0.459		0.464
Treatment		0.007*		0.284		0.214		0.002*		0.002*		0.036*
Patch		0.034*		0.006*		0.003*		0.002*		0.772		0.519

^a WHC = water holding capacity. ^b SOM = soil organic matter. ^c Soils were collected from spaces between plants or under the canopy of L. tridentata shrubs. ^d N addition plots were treated with 60 kg of N (as NH₄NO₃) ha^–1·yr^–1 during 2005–2012. Significance at α = 0.05 indicated by bold and *. SD = standard deviation, n = 3.

Surprisingly, N addition increased abundance of archaeal amoA genes compared to controls, regardless of the measure used (copies per g soil, Fig. 1; or copies per total community DNA, 904 vs 548 archaeal amoA copies per ng DNA, in soils between plants). As expected from many other studies, long-term N addition also increased the abundance of bacterial amoA gene copies. Fertilization decreased the AOA to AOB ratio in the relatively fertile soils under plant canopies (Fig. 1; Table S2, Supporting Information; N addition, 3.6 AOA/AOB; Control, 4.9 AOA/AOB) but generally increased it in the interplant soils (N addition, 6.2 AOA/AOB; Control, 4.3 AOA/AOB).

Figure 1.

Quantification of amoA gene copy numbers for AOB and AOA from Sonoran Desert soil in N addition and control plots. Error bars are standard errors of independent field triplicates.

Slurry maximum AO rates (i.e. at V_max) were significantly higher after long-term N addition compared to those of the control soils (Fig. 2; P < 0.05 in both cases). This trend was also supported by measurements of static maximum AO rates (Fig. S1, Supporting Information; P < 0.05 in both cases). NH₄⁺ supplementation only enhanced AO rates in the static incubations of unfertilized soils but not in fertilized soil (Fig. S1, Supporting Information), nor in any slurried incubations. These patterns show that rates of AO under undisturbed, unamended conditions are NH₄⁺-limited and may be influenced by anthropogenic N additions. Taken together with the microbial abundance data, these results suggest that at least some of the AOA and AOB are likely contributors to AO (Fig S2, Supporting Information) and—as shown by the significant increases of AO rates as well as the abundance of archaeal and bacterial amoA genes after N addition—both archaeal and bacterial ammonia oxidizers are responsive to environmental change.

Figure 2.

Concentration-response kinetics of ammonia oxidation using the shaken-slurry assay for net potential rates. To test the effect of long-term N addition on ammonia oxidation rates, soils were supplemented with a range of NH₄⁺ concentrations in the short-term laboratory methods to measure kinetics of ammonia oxidation. NO₂⁻ accumulation is measured after sodium chlorate inhibition as a proxy for ammonia oxidation. Bi-directional error bars are standard deviations of independent field triplicates to show variation in supplemented NH₄⁺ and measured ammonia oxidation rates.

To investigate the functional capacity of ammonia-oxidizing communities in bulk soils, we evaluated the affinity (K_m) parameter from the AO kinetics plots (Fig. 2; Fig. S1, Supporting Information). K_m was not possible to estimate formally in the shaken-slurry assays since rates were always close to maximum regardless of supplemental N addition, highlighting the low ammonia demand of ammonia oxidizers in desert soils. Residual NH₄⁺ as low as 17 μM was measured in these assays (Fig. 2), implying that the community K_m is likely at or below this low value, which is significantly lower than typical K_m values for known AOB cultures (see rates compiled in Martens-Habbena et al., 2009). In the intact incubations that were not continuously aerated (static AO rates; Fig. S1, Supporting Information), the mean K_m was 2.6 mM for control soils under plants and 1.2 mM for the control soils between plants. Effects of N addition could not be evaluated, given that the long-term N addition itself prevented incubations at low enough ammonium.

An alternative way of looking at differential efficiency in ammonium utilization is to normalize the maximum AO rate by the size of the community (Fig. 3; Fig. S3, Supporting Information). Here, AO was more efficient (higher rates per copy of amoA gene) under plants than between plants (P < 0.05 in all cases), and long-term N addition led to more efficient rates of AO compared to control soils in the spaces between plants (significant for the shaken-slurry assay, Fig. 3; P < 0.001, Fig. S3, Supporting Information). These results suggest a change in community function (NH₃ processed per amoA), given that the ammonia-oxidizing community adapted favorably to higher nutrient soils (e.g. under plants and N addition).

Figure 3.

Effects of N addition on the function of the amoA gene-containing community using estimates of copy-specific ammonia oxidation rates. Specific rates were calculated as maximum ammonia oxidation rate (V_max) from the shaken-slurry assay, divided by amoA gene copy number per g soil. Error bars are standard errors of V_max and amoA calculations for independent field triplicates.

Composition of the ammonia-oxidizing community

All phylotypes detected were related to either Nitrososphaera (Thaumarchaeota) or Nitrosomonas and Nitrosospira (both β-Proteobacteria), at a ratio of about 45:10:1, respectively (Figs 4 and 5). Even though this ratio is subject to potential primer biases, choosing primers that are unlikely to miss abundant AOM groups (Junier et al., 2008) helps to combine qPCR and sequencing data for approximate abundance comparisons of phylotypes across domains. The most abundant phylotype, belonging to the Nitrososphaera subcluster 1.1, accounted for 60% of all the amoA sequences. The community composition was minimally influenced by patch type or N addition treatment, when assessed at the level of OTUs, and we could not detect any significant differences in the relative abundance of the dominant members (Fig. 4). In soil under plants, N addition decreased AOA PD (P < 0.001) but increased the within-group variance for AOB (PERMDISP analysis, P = 0.037), suggesting that N addition has distinct effects on community relatedness of the AOA than of the AOB. However, all other α and β diversity metrics revealed that the structure of AOA and AOB was not influenced by N addition or patch type (P > 0.1 in all cases). Together—at least as far as one can detect based on the amoA gene sequences—these data suggest that long-term N addition had a minor effect, if at all, on AOA and AOB community structure.

Figure 4.

Community composition of OTUs (clustered at 97% nucleotide similarity) based on bioinformatics of amoA gene pyrosequences. Diversity measures were analyzed separately for AOA and AOB. Brackets include taxonomic classifications and percentage of phylotype out of AOA or AOB as an average across all treatment and patch replicates. The remaining 2% of AOB were unclassified to the species level. The remaining 2% of AOA are identified under Nitrososphaera subclusters 2, 8 and 9. Despite biases with qPCR and pyrosequencing technologies, the AOB and AOA bars are drawn at scale (about 25 and 75% of total amoA, respectively). Each bar is the average of independent field and pyrosequencing triplicates (excluding one replicate each in the interplant control and N addition samples for the AOA).

Figure 5.

Neighbor-joining phylogenetic tree of archaeal and bacterial amoA gene sequences. Sequences for this study were obtained from pyrosequencing of the amoA gene. Sequences of known strains and subclusters are used as reference groups. OTUs were clustered at 97% nucleotide similarity and taxonomy classified at multiple phylogenetic levels within the Thaumarchaeota (formerly named Crenarchaeota) as proposed by Pester et al. (2012) and within the AOB (Purkhold et al., 2000). The relative abundance of phylogenetic groups within the AOA or AOB detected in this study is shown in parentheses next to the clade name. Height of clades is proportional to the OTU richness. Bootstrap support is represented by full (75–99%) and empty (50–75%) markers at the nodes.

DISCUSSION

Source of N for ammonia oxidizers

Many studies from various non-arid ecosystems have shown that inorganic N addition either does not affect AOA or allows AOB to outcompete AOA (e.g. Di et al., 2009; Jia and Conrad 2009; Stopnisek et al., 2010; Xia et al., 2011; Levicnik-Hofferle et al., 2012 and reviewed in Hatzenpichler 2012). A few studies have shown that AOA may react favorably to NH₃ originating from organic N sources or N mineralization (Chen et al., 2008; Schauss et al., 2009; Kelly et al., 2011; Daebeler et al., 2012; Levicnik-Hofferle et al., 2012; Lu et al., 2012). Studies showing an increase in AOA abundance after inorganic N additions are rare (Verhamme, Prosser and Nicol 2011; Daebeler et al., 2014; current study). The positive response by AOA may be explained by NH₃ availability from organic N or mineralization from organic sources (He et al., 2012). However, organic N inputs are relatively low in ecosystems such as deserts and other extreme environments (Schimel and Bennett 2004; Booth, Stark and Rastetter 2005). Although N inputs to arid lands significantly increase productivity and N content of seasonal herbaceous annual plants, net potential N mineralization in soil does not appear to be consistently augmented by N addition—perhaps due to the frequency of water limitation, the patchiness of plant growth and organic matter loss pathways such as photodegradation and aeolian/hydrologic transport (Hall et al., 2009, 2011; Rao et al., 2009). Regardless of the role of organic N, our results highlight the unique, positive response of AOA to long-term inorganic N addition in the low organic matter plant interspaces of desert soils.

The use of inorganic N fertilizers by AOA may be plausible in arid systems. Since heterotrophs are likely the first to consume organic N upon metabolic activation after drought, the typical pulses of resource availability imparted by fast drying/wetting cycles may force AOA to utilize NH₃ from inorganic N sources (Placella and Firestone 2013). Additionally, alkaline and hot environments may enhance NH₄⁺ deprotonation, leading to NH₃ gas diffusion throughout the soil matrix (McCalley and Sparks 2009; Geisseler et al., 2010). The same strains of AOA may be capable of using either NH₃ from organic N or inorganic N sources depending on environmental conditions (He et al., 2012), as shown in vitro for the only pure AOA isolate from soil, Nitrososphaera viennensis (e.g. urea; Tourna et al., 2011), and predicted in silico based on the genome of a recent enrichment culture, Candidatus N. gargensis (Spang et al., 2012).

Size, structure and function of ammonia-oxidizing communities in arid land soils

We hypothesized that long-term N addition selects for ammonia oxidizers that are more copiotrophic (lower substrate affinity, higher activity per amoA gene copy) than those in unfertilized soils (Martens-Habbena et al., 2009; Prosser and Nicol 2012). Indeed, N addition elevated the AO rate per amoA copy, but this effect was significant for only the least fertile parts of the landscape (in soil between plants; Fig. 3) where the desert-adapted ammonia oxidizers may be functioning differently after fluctuations in the environment. Differences in organic compounds between soils under and away from vegetation may affect function of ammonia oxidizers as it does in cultures (Lehtovirta-Morley et al., 2014). However, the relative abundance of dominant amoA OTUs was constant across treatments, with small changes only in the minor members (Fig. 4). Of course, we cannot fully discount the idea that perhaps the minor OTUs represent those that are ecologically relevant, while the numerically dominant groups are less efficient or inactive (Lennon and Jones 2011). This scenario has yet to be proven experimentally and is unlikely to be the case here since archaeal and bacterial amoA gene abundance—largely determined by the common OTUs—was positively correlated with AO rates (Fig. S2, Supporting Information).

Evidence of unique AO patterns in deserts

Arid land soils face extreme environmental conditions that may select for unique phylogeny and niche separation. Terrestrial studies worldwide have revealed that the ‘marine’ clade AOA (Group I.1a) are often the main contributors to AO and responders to changes in conditions from soil incubations (Hatzenpichler 2012), despite being outnumbered by the ‘soil’ clade (Group I.1b; Verhamme et al., 2011; Isobe et al., 2012; Long et al., 2012; Zhang et al., 2012; Lu and Jia 2013). Here, we show that AOA within the ‘soil’ clade responded significantly to N addition, and the abundance of this group was positively related to AO rates in desert soil. We also found that Nitrosomonas sequences outnumbered Nitrosospira, a rarity pattern for soil systems. Wastewater discharge in a desert environment was found to harbor Nitrosomonas-like strains (Angel et al., 2010), but is an unlikely scenario for the rural location of our soils in a protected state park. Alternatively, dominance of many Nitrosomonas spp. appears to be limited to alkaline, high-salt and sometimes high-NH₄⁺ conditions (Webster et al., 2005; Cantera, Jordan and Stein 2006; Koops et al., 2006; Ke and Lu 2012). Pulsed resource availability—a characteristic of arid lands—may also drive this distribution, since Nitrosomonas strains have advantages over Nitrosospira such as faster growth responses after starvation (Bollmann et al., 2002). Additionally, in most soils studied previously, AOA outnumber AOB to a greater extent than found in our study (Leininger et al., 2006). In the occasional cases where AOB outnumber AOA, typically up to 10-fold in terrestrial systems (e.g. Di et al., 2009), other arid lands also have a novel distribution as AOB outnumber AOA by 100-fold in cold desert biocrusts (Marusenko et al., 2013a). Overall, atypical ammonia-oxidizing communities appear to occupy desert soils.

Growth and activity characteristics derived from culture experiments can be combined with environmental data to explore relationships between AOA and AOB at the physiological and ecosystem scale (Stark and Firestone 1996; Schauss et al., 2009; Prosser and Nicol 2012). For example, since maximum AO activity per cell is higher for Nitrosospira and Nitrosomonas strains than for AOA (10- and 35-fold, respectively), the contribution of AOB to our AO rates must be much more important than could be predicted from their abundance. With a 10:1 ratio of abundance between Nitrosomonas and Nitrosospira in our soils, the weighted average maximum cell activity for Nitrosomonas plus Nitrosospira should be 33-fold higher than that of AOA. Based on the assumption that 1 amoA copy exists per AOA cell, and that a weighted average of 2.1 amoA copies are found per AOB cell (2 and 3 amoA copies per cell for Nitrosomonas and Nitrosospira, respectively; Norton et al., 2002), we can estimate that the maximum AO activity per amoA copy is 16-fold higher for AOB than AOA in our soils. Assuming equal number of genomes and level of transcription/translation of amoA, the fact that AOA amoA copies outnumber AOB in our soils by 4.3-fold still means that AOB contribute 3.7-fold more than AOA to overall AO rates. These calculations are consistent with our data, which show that AOB contribute on average 4.5 times more to AO rates than AOA (compare slopes in Fig. S2, Supporting information). Even though AOB are the dominant contributors at the ecosystem scale (e.g. total AO), the doubling of AOA abundance in soils between plants of N-fertilized plots means that the relative importance of AOB and AOA to AO may change with N increases. This type of study refines predictions of how environmental conditions affect the link between community dominance and AO rates.

CONCLUSION

N addition affects arid land N cycling primarily through changes in community size, but less so through changes in community composition. This study shows significant and positive effects of inorganic N addition on abundance of Nitrososphaera-related AOA in soils. This pattern has been rarely shown before, especially where N inputs from organic sources are low such as in unique conditions of desert soils. Increased anthropogenic activity resulting in environmental N enrichment may continue to alter ecosystem function through responses by both the AOA and AOB. This work stresses the importance of research in arid lands in that results from mesic systems may not be readily applicable, particularly given that agricultural and pastoral systems in drylands occupy ∼32% of the terrestrial land surface worldwide and often contain alkaline soil that is routinely exposed to high temperatures (Koohafkan and Stewart 2008). These systems may contain AOM communities more similar to hot deserts than to arable lands from more mesic environments.

Our results highlight the effects of N enrichment on AO rates and the community size of ammonia oxidizers. We explored patterns resulting from long-term N enrichment, yet it remains to be seen whether population dominance also shifts during short-term N changes associated with pulsed moisture fluctuations that are characteristic of arid lands. Seasonal changes may occur in AO communities (e.g. AOA abundance may dominate relative to AOB in the summer; Sher et al., 2013), but it is currently unclear whether these changes are related to rates of nitrification in arid and semi-arid soils following long-term N additions (Hall et al., 2011). Future work is essential to investigate how our results compare to those of other arid lands and at scales that were not tested here. Further research is also necessary to predict the AOM contribution to ecologically and atmospherically important gases such as N₂O or NO from nitrifier denitrification and nitrification in these desert soils.

We would like to thank David Huber, Jennifer Learned, Brenda Ramirez, Julea Shaw and Natalie Myers for assistance with lab work and training. Lindsey Pollard provided photograph for the graphical abstract. We are grateful to Jean McLain, Egbert Schwartz, Estelle Couradeau and Elizabeth Cook for manuscript review.

FUNDING

This work is supported by NSF through the CAP LTER program (grant BCS-1026865). Funding was also provided by the NSF Western Alliance to Expand Student Opportunities (WAESO) program and the Graduate & Professional Student Association (GPSA) at ASU.

Conflict of interest statement. None declared.