Seasonal and long-term effects of nutrient additions and liming on the nifH gene in cerrado soils under native vegetation.

Biological nitrogen fixation (BNF) represents the main input source of N in tropical savannas. BNF could be particularly important for Brazilian savannas (known as Cerrado) that show a highly conservative N cycle. We evaluated the effects of seasonal precipitation and nutrient additions on the nifH gene abundance in soils from a long-term fertilization experiment in a Cerrado's native area. The experiment consists of five treatments: (1) control, (2) liming, (3) nitrogen (N), (4) nitrogen + phosphorus (NP), and (5) phosphorus (P) additions. The nifH gene sequence was related to Bradyrhizobium members. Seasonal effects on N-fixing potential were observed by a decrease in the nifH relative abundance from rainy to dry season in control, N, and NP treatments. A significant reduction in nifH abundance was found in the liming treatment in both seasons. The findings evidenced the multiple factors controlling the potential N-fixing by free-living diazotrophs in these nutrient-limited and seasonally dry ecosystems.

Introduction

Over the last decades, the scientific community's attention has increased toward the substantial role of tropical regions in many biogeochemical cycles, especially on the N budget. The nitrogen cycle is a complex set of processes (Box 1, Figure 1) vastly mediated by soil microorganisms (Swift et al., 1998; Galloway et al., 2004; Reed et al., 2010; Hietz et al., 2011; Pajares and Bohannan, 2016). The majority of studies focus onpan> forest ecosystems (Hietz et al., 2011; Homeier et al., 2017; Figueiredo et al., 2019; Brookshire et al., 2019). However, tropical savannas could also play a significant role inpan> critical processes of the N cycle (Bustamante et al., 2006), such as inpan> the biological nitrogen fixation (BNF), considered as the main source of N in these areas with estimated inputs ranging from 16 to 44 kg N ha−1 year−1 (Cleveland et al., 1999).

Figure 1

Nitrogen cycle

Schematic representation of major pathways in the N cycle occurring in the atmosphere-soil interface. The microbial genes involved in each step are indicated. The values in the figure represent the fluxes in the N budget measured for the cerrado sensu stricto described in a review by Bustamante et al. (2006). Anammox, anaerobic ammonia oxidation; DNRA, dissimilative nitrate reduction to ammonium; Norg, organic nitrogen. Assimilative processes are indicated by gray arrows. Dissimilative processes are indicated by solid dark arrows. Decomposition is indicated by colored solid arrows. Reduction and oxidation reactions are represented by solid and dashed arrows, respectively.

Nitrogen is a highly dynamic element inpan> nature, occurrinpan>g inpan> different chemical forms strongly influenced by environmental oxidative and reductive conditions. The N biogeochemical cycle (Figure 1) involves a high number of chemical transformations catalyzed by a wide variety of microbial enzymes (Correa-Galeote et al., 2014). The N cycle integrates dissimilative, assimilative, and decomposition processes (Levy-Booth et al., 2014). Briefly, N-fixing microbial communities that express the nitrogenase enzyme, encoded by the nifH gene, reduce the atmospheric N2 to NH4+ in a dissimilative process. The available NH4+ can pass through two metabolic routes: (1) the dissimilatory pathway in which NH4+ is transformed into hydroxylamine (NH2OH) by the activity of the enzyme ammonia monooxygenase (encoded by the amoA gene) present in ammonium oxidizing bacteria and archaea or (2) the immobilization by plants and microorganisms in an assimilative process of N-NH4+. In the dissimilative processes, the NH2OH is quickly oxidized to NO2- by microorganisms that express the hao gene, which encodes a hydrazine oxidase associated with nitrification (includes the nitritation and nitratation) and also with the anammox process. Following the nitrification's metabolic pathway, the NO2- can be oxidized to NO3- by a nitrate oxidoreductase synthesized by microbial communities that express nxr genes. Plants or microorganisms assimilate the NO3- or it can be directly reduced to NH4+ by the dissimilative nitrate reduction to ammonium (DNRA) or even return to NO2- in another dissimilative reduction process when the napA, narG, or nasA genes are expressed. These genes encode nitrate reductases involved in the initial stage of denitrification. The NO2- (from the dissimilative reduction process or generated in the nitritation stage) can return to NH4+, be directly reduced to gaseous forms, or pass through the anammox process before being reduced to N2. Nitrite reductases encoded by the nirA gene catalyze the reduction of NO2- to NH4+. The reduction of NO2- to the gaseous form NO involves nitrite reductases encoded by genes such as nirS and nirK. The norB genes encode nitric oxide reductases that catalyze the NO reduction to N2O, and the nosZ gene encodes nitrous oxide reductases that catalyze the N2O conversion to N2. In the anammox process, either NH4+ or NO2- which have undergone the partial nitrification process or resulting from the dissimilative reduction of NO3- can be converted to hydrazine (N2H4) and then to N2 from the activity of a hydrazine oxidoreductase, encoded by hzo gene. In assimilative pathways, the N-NH4+ or N-NO3- are immobilized by microbes and plants. The element N can return to the cycle through decomposition. The organic nitrogen compounds (Norg) are degraded during the litter decomposition and can be immobilized again by plants or microorganisms. The Norg could also be mineralized into NH4+, continuing the cycle. The Norg mineralization or ammonification can be catalyzed by several other enzymes, such as chitinase, encoded by the chiA gene.

n class="Chemical">Nitrogen cycle

Schematic representation of major pathways in the N cycle occurring in the atmosphere-soil interface. The microbial genes involved in each step are indicated. The values in the figure represent the fluxes in the N budget measured for the cerrado sensu stricto described in a review by Bustamante et al. (2006). Anammox, anaerobic ammonia oxidationpan>; DNRA, dissimilative nitrate reduction to ammonium; Norg, organic nitrogen. Assimilative processes are indicated by gray arrows. Dissimilative processes are indicated by solid dark arrows. Decomposition is indicated by colored solid arrows. Reduction and oxidation reactions are represented by solid and dashed arrows, respectively.

The Brazilian Cerrado is the second largest biome in South America, covering about 24% of the national territory and predominates in central Brazil (Ribeiro and Walter, 2008). It is one of the most heterogeneous and biodiverse tropical savannas worldwide, known to support a highly complex vegetation structure that evolved under interactions between water and nutrient availability (Myers et al., 2000; Reatto et al., 2008; Silva et al., 2008; Haridasan 2008). The soil microbial communpan>ities inpan> this biome have been studied inpan> terms of compositionpan>, diversity, and richnpan>ess (Quirinpan>o et al., 2009; Viana et al., 2011; Araujo et al., 2012; Rampelotto et al., 2013; Pereira de Castro et al., 2016). However, inpan>formationpan> onpan> the role of microorganisms inpan> nitrogen cycling remains scarce. In particular, information on the impact of nutrient enrichment in soil microbiome from natural areas has been poorly discussed.

Nitrogen is the mainpan> componpan>ent of proteinpan>s and nucleic acids, and therefore, it is conpan>sidered as onpan>e of the most critical elements for the existence of life on Earth. The N turnover comprises a series of complex and dynamic reactions strongly dependent on environmental oxidizing or reducing conditions. These reactions (Box 1, Figure 1) are typically divided into BNF, nitrification, denitrification, anaerobic ammonia oxidation (anammox), and ammonification processes (Canfield et al., 2010; Stein and Klotz, 2016) catalyzed by specific microbial enzymes (Correa-Galeote et al., 2014). In Cerrado soils, the most substantial fraction of inorganic N in the soil is N-NH4+, and the N cycle stages that occur under anaerobic conditions little contribute to the N balance (Nardoto and Bustamante, 2003). For example, net nitrification in native Cerrado soils is often undetectable (Nardoto and Bustamante, 2003), and emissions of NOx and N2O are very low, suggesting a highly conservative N cycle (Pinto 2002; Fernandes Cruvinel et al., 2011). Thus, BNF can represent the most critical N metabolism process in Cerrado soils, as observed for many terrestrial ecosystems, where this process represents about 97% of natural N inputs (Vitousek et al., 2002; Galloway et al., 2004).

In BNF, N2 is assimilated and transformed onpan>ly by a select group of microorganisms that can be plant symbionpan>ts or free-livinpan>g diazotrophs making the understanding of the microbial ecology involved essential (Reed et al., 2010; Pajares and Bohannan, 2016). These microorganisms are capable of expressing the nitrogenase enzyme codified by nif genes. The nifH gene has been accessed with molecular techniques for studies on microbial communities' potential to fix atmospheric N (Gaby and Buckley, 2011, 2012; Pajares and Bohannan, 2016). This gene encodes the iron-protein subunit of the nitrogenase enzyme complex in bacteria and Archaea (Zehr et al., 2003), whose role is to catalyze the reduction of N2 to NH4+ in the BNF, a high energy-demanding process (Zehr et al., 2003; Shridhar, 2012). From this process, reactive N forms required for macromolecule biosynthesis are made available for plant uptake. Since the nifH gene is widely distributed between bacteria and Archaea with conserved sequences in both domains (Zehr et al., 2003), it may be considered as a useful biological marker to infer the ecological role and potential of the microbial community in N-fixing (Gaby and Buckley, 2012). Proteobacteria members and some other taxonomic lineages within Actinobacteria, Cyanobacteria, and Firmicutes contain the nifH gene (Mirza et al., 2014; Dahal et al., 2017). These bacterial groups and specially Proteobacteria are highly abundant in Cerrado soils (Quirino et al., 2009; Souza et al., 2016; Pereira de Castro et al., 2016).

Cerrado's soil microbiome is sensitive to soil chemical characteristics and management (Souza et al., 2016; Silva et al., 2019). It also responpan>ds to the marked seasonpan>ality that regulates the water availability in the biome (Pereira de Castro et al., 2016). Likewise, BNF could be affected by environmental changes promoted by liming and nutrient additions into soils (both standard practices in agricultural areas in the Cerrado) since the content as well as the ratios between nutrients can affect the function of these organisms (e.g., Vitousek et al., 2013; Weisany et al., 2013). A previous study showed that fertilization with N and N combined with P altered other N metabolism process in Cerrado soils under native vegetation (Jacobson et al., 2011). There was an increase in NOx emissions when only the N was added into the soils. On the other hand, NOx emissions were lower when the N fertilizer was added together with P (Jacobson et al., 2011). These results indicate a greater immobilization of N in the biomass and reinforce the co-limitation of the N cycling in Cerrado soils by N and P (Jacobson et al., 2011). Also, the seasonal distribution of rainfall determines the microbial activity in Cerrado soils with an N-NO3- accumulation for short periods after the first rain events and subsequent immobilization (Nardoto and Bustamante, 2003). However, evidence of ecological determinants of microbial groups' potential fixation activity in the Cerrado soils under native vegetation is quite scarce. To understand the potential impacts of nutrient additions and liming under N2 fixation in Cerrado soils, we investigated how it could affect the abundance of nifH genes in soils from a long-term fertilization experiment in a natural area of Cerrado. We expected that increase in mineral nitrogen availability would affect N-fixing free-living microorganisms as it occurs in symbiotic associations. Thus, we hypothesized that the addition of nitrogen and nitrogen combined with phosphorus would reduce the abundance of the nifH gene of free-living N-fixers. The effect would be more evident with N and P combined supplementation due to a reduction of phosphorus limitation which would favor nitrogen assimilation. Additionally, we hypothesized that increased soil pH in response to liming could increase nifH gene abundance.

Results

Soil parameters

Soil moisture (37.6% and 24.3%) and temperature (22.2°C and 18.4°C) decreased from the rainy to the dry season (Table 1). The inorganic N conpan>tent (N-NO3- and N-NH4+) also showed marked seasonal variation, decreasing concentrations from rainy to the dry season in all treatments (Table 1). The soil pH showed less seasonal variation but was strongly affected by the treatments. In comparison with the control treatment, there was an increase in soil pH in the liming treatment (control vs. liming—rainy: 4.67 vs. 6.11; dry: 4.00 vs. 6.27), N (rainy: 3.76; dry: 3.64), NP (rainy: 3.73; dry: 3.95), and P treatments (rainy: 3.73; dry: 3.95) (Table 1).

Table 1

Soil properties

Variable	Rainy					Dry
	Control	Ca	N	NP	P	Control	Ca	N	NP	P
Moisture (%)	37.83	38.30	36.20	37.10	38.90	25.83	20.13	24.13	25.17	25.50
	±3.25	±1.53	±1.68	±1.17	±0.15	±0.41	±1.56	±0.96	±1.23	±0.72
Temperature (°C)	22.07	21.25	22.75	21.63	22.63	18.17	19.06	18.37	18.47	18.17
	±0.42	±0.26	±1.30	±0.90	±0.81	±0.38	±0.19	±0.79	±0.65	±0.16
pH (H2O)	4.67	6.27	3.76	3.73	4.08	4.00	6.11	3.64	3.95	3.73
	±0.21	±0.11	±0.07	±0.11	±0.08	±0.09	±0.09	±0.04	±0.30	±0.08
N-NO3- (mg/kg)	2.27	2.22	2.18	1.64	2.60	1.61	1.73	0.88	1.04	1.37
	±0.25	±0.62	±0.61	±0.54	±0.61	±0.14	±0.30	±0.18	±0.72	±0.26
N-NH4+ (mg/kg)	25.13	24.37	43.78	96.43	37.04	9.02	8.32	6.58	8.44	6.68
	±8.95	±7.13	±15.02	±66.24	±25.21	±1.16	±1.35	±0.41	±2.17	±0.19
TN (%)	0.19	0.17	0.18	0.20	0.19	–	–	–	–	–
	±0.02	±0.01	±0.02	±0.02	±0.02
TC (%)	3.66	3.46	3.40	3.64	3.65	–	–	–	–	–
	±0.60	±0.15	±0.49	±0.36	±0.37
C:N	18.85	20.42	18.65	18.54	19.02	–	–	–	–	–
	±0.87	±0.28	±0.38	±0.29	±0.41
P (available, mg/dm3)	1.70	0.65	1.28	6.98	6.60	–	–	–	–	–
	±0.36	±0.33	±0.13	±2.86	±2.67
K (mg/dm3)	28.75	10.75	22.75	28.50	26.75	–	–	–	–	–
	±6.18	±1.89	±3.40	±3.32	±2.50
Mg2+ (cmolc/dm3)	0.11	2.73	0.08	0.13	0.10	–	–	–	–	–
	±0.03	±0.15	±0.01	±0.08	±0.02
Ca2+ (cmolc/dm3)	0.19	4.69	0.15	0.40	0.30	–	–	–	–	–
	±0.05	±0.20	±0.02	±0.17	±0.06
Al3+ (cmolc/dm3)	0.94	0.00	1.38	1.46	1.07	–	–	–	–	–
	±0.17	±0.00	±0.15	±0.14	±0.24

Soils (0-10 cm depth) were collected in the rainy and dry seasons of 2018 in a long-term fertilization experiment in a typical cerrado area located at Reserva Ecológica do IBGE, Brasília, Brazil. Results are expressed as mean values ±standard error. Control, untreated control; Ca, liming; N, nitrogen addition; NP, nitrogen and phosphorus addition; P, phosphorus addition; TC, total carbon; TN, total nitrogen.

Other soil chemical characteristics such as total carbon (TC), total nitrogen (TN), and concentrations of available P, K+, Mg2+, Ca2+, and Al3+ were determined during the rainy season (Table 1). TC varied between 3.66% in control and 3.40% in N treatment; the maximum TN percentage was 0.20% in NP, and the minimum was 0.17% in liming treatment, mean values of the CN ratio varied between 20.42 in liming and 18.54 in N plots. We observed almost three-fold reduction in available P content in liming plots and an increased about four-fold in the NP and P treatment (control: 1.70 vs. liming: 0.65; N: 1.28; NP: 6.98; P: 6.60 mg/dm³). The K content decreased approximately three-fold in the liming treatment compared to control plots (control: 28.75 vs. liming: 10.75 mg/dm³), while Mg2+ (control: 0.11 vs. liming: 4.69 cmolc/dm³) and Ca2+ (control: 0.19 vs. liming: 4.69 cmolc/dm³) increased about 25-fold in the liming treatment (Table 1). The Al3+ contents were zero in the liming treatment (control: 0.94 vs. liming: 0.00 cmolc/dm³) but increased in the NP treatment (1.46 cmolc/dm³) compared with control plots (Table 1).

Taxonomy characterization of nifH gene and 16S rRNA taxonomic links

The BLAST nucleotide alignment result showed that the n class="Chemical">nifH gene sequence recovered from three cloned fragments was classified in the Proteobacteria cluster, comprising sequences most closely related to Bradyrhizobium members (over 93% identical; see Table S1).

From the previously filtered 16S rRNA database (Silveira et al., 2020), we analyzed a total of 612,745 good-quality sequences representing the Proteobacteria group. The result corresponded to 3,863 taxa comprising three classes, 71 orders, 122 families, and 236 genera. The taxonomy tree based on this data set showed a substantial difference inpan> the abunpan>dance of Proteobacteria members between the NP treatment and unpan>treated conpan>trol durinpan>g the rainpan>y seasonpan> (Figure 2A). Proteobacteria abunpan>dance inpan> P treatment also differed from N and liming treatments and in liming compared with NP treatment (Figure 2A). During the dry season, Proteobacteria member abundances differed mainly in the comparisons between liming versus all treatments, including the control plots, with lesser differences in the other pairwise comparisons (Figure 2A). Taxonomic groups inside Proteobacteria phylum comprised the genera Rhodoplanes, Methylovirgula, Rhodomicrobium, Roseiarcus, Pedomicrobium, Rhizobium, Bradyrhizobium, and others affiliated to Alphaproteobacteria, Gammaproteobacteria, and Deltaproteobacteria classes represented in Figure 2A.

Figure 2

Proteobacteria community

Hierarchical taxonomy of Proteobacteria (A) and principal coordinate analysis (PCoA) plot with Bray-Curtis dissimilarity (B). Members of Proteobacteria were recovered from the 16S rRNA gene in the soil samples (0–10 cm depth) collected in the rainy and dry seasons of 2018. Soils were collected from a typical cerrado area where the long-term fertilization experiment was installed, in the Reserva Ecológica do IBGE, Brasília, Brazil. The gray tree on the right represents a key for the unlabeled trees. Each of the smaller trees represents a comparison between the treatments in the columns and rows. Node size represents the relative proportions for that taxon. A taxon colored brown is more abundant in the treatment in the column, and a taxon colored green is more abundant in the treatment of the row. Control, untreated control; Ca, liming; N, nitrogen addition; NP, nitrogen and phosphorus addition; P, phosphorus addition.

Hierarchical taxonomy of Proteobacteria (A) and principal coordinate analysis (PCoA) plot with Bray-Curtis dissimilarity (B). Members of Proteobacteria were recovered from the 16S rRNA gene inpan> the soil samples (0–10 cm depth) collected inpan> the rainpan>y and dry seasonpan>s of 2018. Soils were collected from a typical cerrado area where the lonpan>g-term fertilization experiment was installed, in the Reserva Ecológica do IBGE, Brasília, Brazil. The gray tree on the right represents a key for the unlabeled trees. Each of the smaller trees represents a comparison between the treatments in the columns and rows. Node size represents the relative proportions for that taxon. A taxon colored brown is more abundant in the treatment in the column, and a taxon colored green is more abundant in the treatment of the row. Control, untreated control; Ca, liming; N, nitrogen addition; NP, nitrogen and phosphorus addition; P, phosphorus addition.

The principal coordinate analysis (PCoA) ordination based on Bray-Curtis dissimilarity shown a clear difference in Proteobacteria community distribution among liming samples and all the other treatments, including control (Figure 2B). The N and NP treatments also differed in Proteobacteria community distribution compared with control. The first two PCoA axes explain approximately 60% of data variability (Figure 2B).

Seasonal and nutrient addition effects on nifH gene abundance

The relative abundance of the nifH gene was affected by the seasonality of precipitation (p < 0.001) (Figure 3). Relative abundance of the nifH gene counted over four-fold higher during the rainy season in the control plot (p < 0.001) and more than two-fold higher in the N (p < 0.01) and NP (p < 0.001) plots during the same sampling period compared with the dry season.

Figure 3

nifH gene abundance

Relative abundance of nifH gene in the cerrado' soil samples based on calibrated normalized relative quantities (CNRQ values) generated in the qbase + software. The relative abundance of nifH gene was measured from soils (0–10 cm depth) collected in the rainy and dry seasons of 2018 in a long-term fertilization experiment in a typical cerrado area, located at Reserva Ecológica do IBGE, Brasília, Brazil. Control, untreated control; Ca, liming; N, nitrogen addition; NP, nitrogen and phosphorus addition; P, phosphorus addition. Bar plots and error bars represent the average and standard errors between technical replicates, respectively. Red asterisks indicate significant differences in nifH relative abundance between rainy and dry seasons for the same treatment. Black asterisks indicate significant differences in nifH relative abundance among treatments. ∗∗∗ = p < 0.01; ∗ = p < 0.05.

n class="Chemical">nifH gene abundance

Relative abundance of nifH gene inpan> the cerrado' soil samples based onpan> calibrated normalized relative quantities (CNRQ values) generated inpan> the qbase + software. The relative abunpan>dance of nifH gene was measured from soils (0–10 cm depth) collected in the rainy and dry seasons of 2018 in a long-term fertilization experiment in a typical cerrado area, located at Reserva Ecológica do IBGE, Brasília, Brazil. Control, untreated control; Ca, liming; N, nitrogen addition; NP, nitrogen and phosphorus addition; P, phosphorus addition. Bar plots and error bars represent the average and standard errors between technical replicates, respectively. Red asterisks indicate significant differences in nifH relative abundance between rainy and dry seasons for the same treatment. Black asterisks indicate significant differences in nifH relative abundance among treatments. ∗∗∗ = p < 0.01; ∗ = p < 0.05.

During the rainy season, the nifH gene relative abunpan>dance decreased approximately ten- and two-fold inpan> the liminpan>g (p < 0.001) and P (p < 0.05) treatments compared to conpan>trol plots, respectively (Figure 3). Also, the nifH gene abundance was about six-fold lower in the liming treatment compared to N and NP treatments (p < 0.001) and about four-fold lower than the relative counts in P treatment (p < 0.001). There were no differences in nifH relative abundance in the N and NP treatments compared to control (p > 0.05). In the dry season, the relative abundance of nifH decreased about four-fold in liming treatment compared to control (p < 0.001). Again, the liming treatment showed lower nifH abundance than N, NP, and P treatments (between four and five-fold lower; p < 0.001). There were no differences between these last three treatments and control plots (p > 0.05). The nifH gene relative abundance was positively correlated with soil moisture (0.48, p < 0.01) and with the N-NH4+ content (0.38, p < 0.05) (Figure 4).

Figure 4

Soil parameters and nifH relative abundance

Spearman's correlation between soil parameters and nifH gene relative abundance in soils (0–10 cm depth) from a typical cerrado area where the long-term fertilization experiment was installed in the Reserva Ecológica do IBGE, Brasília, Brazil. Significant correlations and their respective p values are highlighted in red. Control, untreated control; Ca, liming; N, nitrogen addition; NP, nitrogen and phosphorus addition; P, phosphorus addition.

Soil parameters and n class="Chemical">nifH relative abundance

Spearman's correlation between soil parameters and nifH gene relative abunpan>dance inpan> soils (0–10 cm depth) from a typical cerrado area where the lonpan>g-term fertilization experiment was installed in the Reserva Ecológica do IBGE, Brasília, Brazil. Significant correlations and their respective p values are highlighted in red. Control, untreated control; Ca, liming; N, nitrogen addition; NP, nitrogen and phosphorus addition; P, phosphorus addition.

Discussion

The soil microbial component constitutes the base for terrestrial ecosystem functioning. Here, we report a comprehensive nifH gene evaluationpan> as a proxy to depict the inpan>teractive effects between precipitation and the long-term nutrient addition on N-fixing activity by diazotrophic soil community. Firstly, the nifH cloned fragment's phylogenetic characterization indicated the sequence taxonomy identity related to Bradyrhizobium members. The genus Bradyrhizobium comprises a group of N-fixing rhizobia in Proteobacteria that could be called generalists with versatile metabolism and ecological relationships. For example, they can include plant-interactive or free-living microorganism species (Kahindi et al., 1997; van Elsas et al., 2019). In the present study, we evaluated nifH DNA sequences from bulk soil and predicted the nifH gene sequence to belong to the putative non-symbiont diazotrophic community. Previous studies in tropical rainforest soils in Costa Rica (Reed et al., 2010) and in Amazon (Mirza et al., 2014) also have reported nifH clone sequences associated to free-living or associative Alphaproteobacteria, including the genera Gluconacetobacter, Azospirillum, Burkholderia, Bradyrhizobium, and others.

The Proteobacteria community taxonomic diversity data set has shown differences inpan> Proteobacteria member abunpan>dance and distributionpan> amonpan>g treatments inpan> the rainpan>y and dry seasonpan>s. It could suggest the potential effects of nutrient addition, promoting shifts in Proteobacteria assembly or in diazotrophic microorganisms affiliated to other taxonomic groups. These changes could also be associated with the decrease in nifH gene abundance (presumably negative effect in N2 fixation) in the treated plots, revealing the potential reduction in the microbial N-fixing process promoted by nutrient additions.

In general, nifH abunpan>dance was relatively smaller inpan> all nutrient additionpan> plots, with more pronpan>ounpan>ced differences in the comparisons with liming treatment. The unexpected significant decrease in nifH gene copy numbers in liming plots could be related to the combined changes in soil chemical parameters (Han et al., 2019) since environmental conditions strictly control the nitrogenase enzyme (Poly et al., 2001; Huergo et al., 2012; Han et al., 2019). The major environmental change promoted by liming into soils was the pH increase. Furthermore, there was also an increase in nutrient contents such as exchangeable Ca2+ and Mg2+ and concomitant decreasing in K+, available P levels, and Al3+ immobilization. A previous soil evaluation indicated a reduction of approximately three-fold in Fe contents in the liming treatment, even after nine years without new additions in our study area (unpublished data). Iron represents an essential element to the nitrogenase proteins complex (Zehr et al., 2003; Gaby and Buckley, 2011), which could be extensively demanded by diazotrophs (Mills et al., 2004; Larson et al., 2018). Although studies including the interactive effects among environmental controls on free-living N-fixing are scarce, similar results were observed in a long-term acidic Ultisol fertilization experiment located in a subtropical monsoon climate region from China (Lin et al., 2018). The researchers found a substantial decrease in nifH gene abundance in plots under NPK plus lime fertilization, associated with the increase in soil pH. The authors also indicated shifts in microbial community composition, replacing the dominant Bradyrhizobium genus by Azohydromonas in the treatment submitted to lime addition (Lin et al., 2018).

The nifH gene abunpan>dance was also reduced inpan> P treatment durinpan>g the rainpan>y seasonpan>. Inpan> this treatment, there was an inpan>crease inpan> the P available conpan>tent and a slight reductionpan> inpan> soil pH. The importance of P availability to N2 fixation is recognized in several studies indicated by Bustamante et al. (2006) in a review of the nitrogen cycle in tropical and temperate savannas. The available P contents are described as a limiting factor to N2 fixation in P-deficient soils in the tropics (Bustamante et al., 2006; Van Langenhove et al., 2019) and could be critical to activate genes for the nitrogenase synthesis (Stock et al., 1990). However, although the increase in P availability is usually described to increase N2 fixation, the free-living diazotrophs could perform BNF over a broader range of phosphorus supply or accessibility than symbionts (Smercina et al., 2019), suggesting other controls to nifH abundance for this group.

In a review about the pivotal players controlling nitrogenase activity, Huergo et al. (2012) highlight the evolved mechanisms inpan> diazotrophs to shut down N2 fixation when N-NH4+ is available in the environment to avoid energy waste during the N2 reduction in BNF process. Similarly, the N-NH4+ was indicated as one of the main drivers of the nifH abundance variation across different agricultural soils, where lower gene copy numbers were related to increased N-NH4+ levels (Pereira e Silva et al., 2013). Here, the nifH relative abundance was negatively affected in plots with higher N-NH4+ levels and also in the liming treatment, where N-NH4+ showed a minor variation compared with control. Thus, the contradictory low positive correlation found between nifH abundance and N-NH4+ in our study could not reflect a linear relationship because the variability in N-NH4+ contents in the soil was more pronounced between seasons due to changes in soil moisture. This result reinforces the understanding that gene abundance could be determined by seasonal and long-term interactive effects in soil chemistry and biological components, such as interactions between soil moisture and N-NH4+ contents and the pervasive long-term alterations in pH and nutrient availability.

The N cycling in Cerrado ecosystems is conservative (Bustamante et al., 2009). The N conservation mechanisms in these ecosystems could be also associated with the low abundance and activity of genes involved in the nitrification process such as amoA, in native Cerrado soils (Catão et al., 2017). However, a previous study in our experimental area showed that the N addition into soils resulted in a rise in NOx emissions to the atmosphere (Jacobson et al., 2011). Our results point to a reduction in potential BNF, indicating that eutrophication can induce long-term changes.

The seasonal dynamic of non-symbiotic N-fixing in Cerrado soils is poorly understood. Our analysis indicated a strong seasonal influence on nifH abunpan>dances. Reducinpan>g nifH gene abundance in control, N, and NP plots was observed in the dry season. The seasonal modulation in nifH abundance may be associated with a high decrease in the soil moisture and temperature during the dry season. It could promote greater oxygen gas diffusion across soil aggregates (Tipping, 2004), increasing aerobic conditions representing a critical inhibitor of nitrogenase activity (Fay, 1992; Norman and Friesen, 2017). Also, aerobic conditions demand substantial energetic resource investments to nitrogenase protection in free-living soil diazotrophs (Norman and Friesen, 2017). Some diazotroph bacteria could even cease nitrogenase production in the presence of high oxygen levels (Bruijn, 2015; Hill, 1988; Reed et al., 2010), which could contribute to explain the decrease in trend in nifH abundance during the dry season.

Limitations of the study

It is essential to highlight that the observed nifH gene abunpan>dance does not inpan>dicate the N2 fixation rate level but represents a proxy to explore the potential N2 fixation by free-living diazotrophs. Thus, further evaluations on nifH sequencing, nifH gene expression based on RNA reverse transcription, and isotope technology will be necessary to infer N-fixing rates to specific diazotrophs taxa and changes in the microbial functions in response to environmental changes promoted by precipitation seasonality and by nutrient additions and liming.

Conclusions

Overall, the increase in nutrient availability and soil pH negatively affected the nifH gene abundance. In contrast, the higher moisture levels in the rainy period had seemed to contribute to the highest nifH relative proportions. Changes in Proteobacteria community distribution among treatments could also be related to variation in the observed nifH abundance since Proteobacteria harbor some important N-fixer groups. Our study is a pioneer for providing the first assessment of the effects of nutrient addition on nifH gene abundance in typical acid soils of Cerrado. Cerrado is a critical Brazilian biome both for agriculture and biodiversity conservation. It is currently strongly threatened by rapid and intensive land use changes and associated impacts. Understanding the potential functionality of soil microbial community in nutrient cycles, such as N cycle, represents a crucial advance to support future studies that could implement new approaches to soil preservation and sustainable use.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Rafaella Silveira (rafaella_silveira@hotmail.com).

Materials availability

This study did not generate unique reagent.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its supplemental files, as well as at the Mendeley Data: https://data.mendeley.com/datasets/4hjj6sprcg/draft?a=860f0447-0868-4f5c-9e3a-f7af2c76de0f. The accession number for the raw 16S rRNA sequence data reported in this paper is NCBI Sequence Read Archive Bioproject: https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA647807.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.