Agronomic evaluation of Herbaspirillum seropedicae strain ZAE94 as an inoculant to improve maize yield in Brazil

Gabriela C.ALVES,Carlos L.R.DOS SANTOS,Jerri E.ZILLI,Fabio B.DOS REIS JUNIOR,Ivanildo E.MARRIEL,Farley A.da F.BREDA,Robert M.BODDEY and Veronica M.REIS

1Federal Rural University of Rio de Janeiro,Department of Soils,km 7 BR 465,Seropédica RJ 23900-000(Brazil)

2Federal University of Mato Grosso,2367 Fernando Corrêa Costa Avenue,Boa Esperança M 78060-900(Brazil)

3Embrapa Agrobiology,km 07 BR 465,Seropédica RJ 23891-000(Brazil)

4Embrapa Cerrados,BR-020 km 18,Planaltina DF 73310-970(Brazil)

5Embrapa Maize and Sorghum,MG 424 km 45,Sete Lagoas MG 35701-970(Brazil)

ABSTRACT Diazotrophic bacteria applied as a seed inoculant can improve the grain yield of several crops including maize.The current study aimed to test the agronomic efficiency and contribution of biological nitrogen fixation(BNF)of the endophytic diazotroph Herbaspirillum seropedicae strain ZAE94 to maize under field conditions.Eighteen field assays were conducted in four different locations during consecutive years on two hybrids and two varieties of maize in a random block design with four replicates using a peat-based inoculant.The inoculant containing the ZAE94 strain was applied without nitrogen(N)fertilization or with 40 kg N ha-1 and was compared to the application of 40 and 80 kg N ha-1 without inoculation.Crop productivity and N accumulation in the grain were evaluated in addition to 15N natural abundance(δ15N)to evaluate BNF in the treatments without N fertilization.Fertilization at 40 kg N ha-1 plus bacterial inoculation produced crop yields similar to the treatment with 80 kg N ha-1 and increased grain N content,especially in the off-season with 40 kg N ha-1.The inoculation treatments showed lower δ15N values than the non-inoculated treatments,which was most evident in the off-season.The BNF contributed about 30%of N accumulated in plants inoculated with ZAE94.On average,64%of the N fertilized plots showed an increase of the parameters evaluated in the inoculated treatments,compared with the control.Inoculation also increased root length,root volume,and leaf area,and these parameters were positively correlated with plant weight using a hydroponic assay.This study revealed that the application of H.seropedicae inoculant increased the amount of N in plants owing to BNF,and there is a better chance of yield response to inoculation under low N fertilizer application in the off-season.

Key Words:agronomic efficiency,biological nitrogen fixation,endophytic diazotroph,inoculation,15N natural abundance,plant growth-promoting bacteria

INTRODUCTION

Maize(Zea maysL.)is widely used in food and feed,and it is incorporated in the energy matrix for ethanol production.Its production in the 2020/2021 harvest was approximately 1.07×108t in Brazil and makes the country the world’s third largest producer(Fiesp,2020).The amount of agricultural inputs used for planting this crop corresponds to approximately 3.8×106t of fertilizers per year(Fowleret al.,2013),and the country has experienced an increase of approximately 300%in the past 50 years from less than 2 to over 6 t ha-1today.However,grain yield can reach 12 t ha-1in plantations with high-input management,and this crop is currently known to be more dependent on chemical fertilizers than other crops in Brazil(Rochaet al.,2014).However,only a part of the applied nitrogen(N)is utilized for maize growth,and there are substantial environmental losses due to leaching and volatilization(Hirelet al.,2011).

There are approaches that may help increase crop productivity and mitigate some of the adverse effects of N application,much of which is lost from the soil/plant system in different ways;these approaches include inoculation with selected diazotrophic bacteria that can promote growth,particularly on poor or degraded soils in Brazil(Baldani and Baldani,2005).Results obtained by different researchers have demonstrated that inoculation with N-fixing bacteria can promote growth of the host plant by mechanisms which are not yet fully understood.Such mechanisms are probably associated with increased root mass and improved use of nutrients including N,in addition to potential benefits derived from the contribution of biological N fixation(BNF)by bacteria(Dobbelaereet al.,2003;Alegria Terrazaset al.,2016).This group of bacteria is known as plant growth-promoting rhizobacteria(PGPR)and may improve the use of fertilizers applied to crops such as maize;for this reason,PGPR are used as inoculants in agriculture(Adesemoye and Kloepper,2009;Hungriaet al.,2010).

Comparing the effects of the inoculation of rhizobium in the legume/rhizobium symbioses to those of associative bacteria,the direct contribution related to BNF is less efficient.However,even if only a fraction of the required N can be provided by the association with the diazotrophic bacteria,the gain would be equal to or higher than that observed in legumes(Döbereiner,1992).In tropical areas,N requirement of soybean can be entirely met by diazotroph inoculation.The main difficulty of this approach is the inconsistency of responses to inoculation,which is linked to factors such as plant species,climatic conditions,interaction with the soil biota,strain selection,and quality of the inoculant(Sumner,1990;Okon and Labandera-González,1994).

Bacteria of the genusHerbaspirillumwere first described in 1986 by Baldaniet al.(1986),and the first species,H.seropedicae,was named after the locality where it was isolated.This new bacterium was considered a true endophytic diazotroph and was thus assumed to be important for plant growth promotion(Chebotaret al.,2015).This opened a new perspective for studies on the association of microorganisms and plants because this species requires a living plant to survive.This finding was confirmed by inoculating the soil without vegetation and after several days it was not possible to detect the bacterium.However,when sterilized seeds were planted,the bacterium was detected again.This discovery was made using the same semi-solid culture medium without N addition after modifying an old formula of a N-free broth(NFb)medium traditionally used forAzospirillumgrowth by lowering the final pH to 5.8(Baldaniet al.,1992).Thus,H.seropedicaeis considered a species adapted to the Brazilian tropical soils.

Studies have shown that this species invades the intercellular spaces of roots,penetrates the endoderm,colonizes the conductive vessels,and may reach the aerial part of plants after inoculation(Olivareset al.,1996;James and Olivares,1998).In addition,the bacterium also expresses thenifgenes(coding nitrogenase)in roots,stems,and leaves of rice,sorghum,maize,and wheat inoculated under controlled conditions(Jameset al.,2002;Roncato-Maccariet al.,2003a).Monteiroet al.(2012)described the genetic potential ofHerbaspirillumto elicit beneficial effects in host plants,and studies on inoculations with species of this genus,includingH.seropedicae,showed several direct and indirect effects on plants.Roncato-Maccariet al.(2003a,b)used this species on rice and suggested that the production of phytohormones may affect plant growth.Herbaspirillumproduces gibberellins and indole 3-acetic acidin vitro(Bastiánet al.,1998)as well as siderophores(Pedrosaet al.,2011).

Based on this knowledge,a previously selectedHerbaspirillumstrain(Alveset al.,2015),classified asH.seropedicaeZAE94(also referred to as BR11417),was applied in field experiments to evaluate its agronomic contribution to maize under field conditions.

MATERIALS AND METHODS

Strain and growth conditions

TheH.seropedicaestrain ZAE94(provided by the Biological Resource Center Johanna Döbereiner Culture Collection,Brazil)showed the best performance out of 21Herbaspirillumstrains used on two maize genotypes,which were grown in a greenhouse with sterile and non-sterile substrate and in field experiments(Alveset al.,2015).This strain was isolated from disinfested roots of rice grown at the Embrapa Agrobiology Experimental Station located in Seropédica,Rio de Janeiro State,Brazil(Baldaniet al.,1986).The ZAE94 strain was initially grown in the solid culture medium JNFb(Olivareset al.,1996),and one colony was inoculated in liquid DYGS medium(Baldaniet al.,2014)at 30°C for 24 h using a rotary shaker at 175 r min-1.Subsequently,100 μL of the bacterial suspension was transferred to new DYGS medium(75 mL)for cultivation for 24 h under the same conditions.Optical density at 600 nm was adjusted to 1 in order to obtain an inoculum containing approximately 109cells mL-1,and this final suspension was applied to 175 g milled dry peat,which had previously been homogenized,autoclaved,and pH-adjusted to 6.0.The inoculated peat was kept at 30°C for 24 h before the final application.After this,the peat inoculum was applied to the seeds at a ratio of 250 g peat per 10 kg maize seeds.The same inoculation treatment was used for all field experiments at Embrapa Agrobiology,which was the source of the strain.The counting showed a population of 106–107cells per seed.This strain was not sensitive to the seed treatment using pirimiphos-methyl,deltamethrin,captan,and metalexil.Bacterial abundance was quantified using the most probable number method(McCrady table using three replicates of JNFb semi-solid medium without N,as described by Baldaniet al.(2014)).The control plots were not treated with the peat inoculant(absolute control).

Field experiments

The evaluation of agronomic effects of ZAE94 strain was performed at four locations:Seropédica in Rio de Janeiro State(RJ)(22°44′38′′S,43°42′28′′W,26 m above sea level(a.s.l.)),Sete Lagoas in Minas Gerais(MG)(19°28′S,44°15′08′′W,732 m a.s.l.),Planaltina in Distrito Federal(DF)(15°39′84′′S,47°44′41′′W,1 000 m a.s.l.),and Boa Vista in Roraima(RR)(2°40′60′′N,60°50′31′′W,88 m a.s.l.).The climate at RJ,MG,DF,and RR is Aw(tropical with hot and humid summer and dry winter),Aw,Cwa(humid subtropical climate),and Awi(tropical climate with alternative wet and dry periods and a well-defined dry season),respectively,according to Köppen classification.The mean monthly temperatures,total monthly rainfall,and dates of planting at the four locations of the maize experiments are presented in Fig.1.The soil is Alfisol at RJ and Oxisol at the other three sites.A total of 18 field experiments were conducted,nine of which were conducted during the rainy season with summer crops growing(first harvest)and nine during the dry season in the late summer/early winter(off-season,also termed second harvest,except for Boa Vista,where the rainy season is in winter).Details of the field experiments are shown in Table I.

Fig.1 Mean monthly temperature,total monthly precipitation,and planting time at the four locations of the maize field experiments in Brazil:Seropédica,Rio de Janeiro State(RJ),Sete Lagoas,Minas Gerais(MG),Planaltina,Distrito Federal(DF),and Boa Vista,Roraima(RR).Solid arrows indicate the planting time in rainy reason,and dash arrows indicate the planting time in dry season.

At the beginning of the experiments,soil samples were taken from the 0–20 cm layer randomly from 20 dispersed points at each site.In addition,soil samples(0–20 cm)were taken from each treatment each year after harvesting,with six subsamples for each replicate.Soil samples were dried at 60°C for 48 h,pulverized to pass through a 2-mm sieve,and analyzed for their chemical and physical properties according to the methods described by Nogueira and de Souza(2005).The evaluated chemical parameters were soil pHwater,exchangeable aluminum(Al),exchangeable calcium and magnesium(Ca+Mg),and phosphorus(P),potassium(K),and carbon(C)contents.Soil basic properties at the time of the experiments are shown in Table II.

TABLE IDetails of the field experiments conducted at Seropédica,Rio de Janeiro State(RJ),Sete Lagoas,Minas Gerais(MG),Planaltina,Distrito Federal(DF),and Boa Vista,Roraima(RR),Brazil for four consecutive agricultural years from 2005 to 2008

TABLE IISoil properties at the four experimental sites,Seropédica,Rio de Janeiro State(RJ),Sete Lagoas,Minas Gerais(MG),Planaltina,Distrito Federal(DF),and Boa Vista,Roraima(RR),Brazil

Each year,50 d before the start of each experiment,soil acidity was adjusted by lime application if necessary,and basic fertilization was applied annually at the time of sowing,as shown in Table I.Two maize genotypes were tested at each site;one genotype was a variety and the other was a simple hybrid developed at the Embrapa Maize and Sorghum Research Centre in Sete Lagoas.At Seropédica,Sete Lagoas,and Planaltina,the variety BRS106 was used,and at Boa Vista,BRS4157 was planted.The simple hybrid BRS1030 was used at Seropédica,Sete Lagoas,and Planaltina,and BRS1010 was used at Boa Vista.Each genotype was subjected to five treatments:not inoculated and not fertilized with N as a control(CK),not inoculated but fertilized at 40 kg N ha-1(N40),not inoculated but fertilized at 80 kg N ha-1(N80),not fertilized with N but inoculated with theH.seropedicaestrain ZAE94(I),and inoculated and fertilized at 40 kg N ha-1(IN40).Nitrogen was applied twice;half of the total amount was applied 15 d after planting(stage V3)and the remainder was applied 45 d after planting(stage V8).The experiments were laid out in randomized blocks with four replicates.The plot sizes are shown in Table I.Herbicides and insecticides were used when necessary in all treatments according to the technical recommendations for each region.None of the plots were sufficiently irrigated;thus,growth depended on rainfall.

Agronomic evaluations were performed to determine grain yield and total N in the grain at the end of the crop cycle.Yield was estimated based on grain weight determined in the field and on the dry matter percentage.After weighing,subsamples of the grain were dried to constant weight at 65°C using a forced circulation air oven.The subsamples were initially passed through a Wiley mill(2 mm)and were then finely ground using a roller mill similar to that described by Arnold and Schepers(2004).Nitrogen was determined according to the semi-micro Kjeldahl method(Rowland,1938).In Seropédica,plant shoots were analyzed for natural15N abundance at the grain filling stage 12 weeks after emergence in four of the six experiments.This analysis was performed on leaf and stem samples from three plants per plot.After analyzing the N content,samples containing 30–60 μg N were analyzed for15N abundance using a continuous-flow isotope-ratio mass spectrometer(Finnigan MAT,Bremen,Germany)in the John Day Stable Isotope Laboratory of Embrapa Agrobiology(Ramoset al.,2001).To assess BNF,three weed plants were collected at the same time from the respective plots as non-N-fixing reference plants and analyzed for N and15N abundance in the same manner.

Greenhouse experiment for root measurement

Maize seedlings were inoculated in a hydroponic system over 21 d using an automated greenhouse with controlled humidity and temperature at Embrapa Agrobiology.The maize genotype SHS5050 was used,which is a triple hybrid of the Santa Helena Company previously studied by Alveset al.(2015)and is considered a promising hybrid for studying the interaction with diazotrophic bacteria.The seeds were washed using tap water and surface-sterilized using a solution of 0.5%sodium hypochlorite and 0.01%Tween 20 under agitation using a rotary shaker at 65 r min-1for 5 min.After that,the seeds were washed three times using 50 mmol L-1sterile phosphate buffer at pH 7.0 under the same conditions as described above at 5-min intervals.The seeds were then pre-germinated on sterile Germitex™paper at 30°C for 72 h under a 12-h light cycle in a controlled plant growth chamber.After germination,the seedlings with root length of 3–7 cm were selected and inoculated with the bacterial inoculant.For inoculation,the strain ZAE94 was used after it had been grown in DYGS medium until reaching an optical density of 1 at 560 nm,corresponding to 109cells mL-1.The suspension was diluted 100-fold using a phosphate buffer at pH 6.0 to a density of 107cells mL-1,and roots were immersed in the suspension for 15 min.The control was treated the same way after heat-killing the bacteria at 100°C for 15 min.

After inoculation,the seedlings were transferred to a hydroponic system.Each vessel received 6 L modified halfstrength Hoagland nutrient solution(Hoagland and Arnon,1950)with N concentration of 3.0 mmol L-1for initial adaptation of the plants.The pH of the medium was adjusted to 5.8 every two days using 0.5 mol L-1KOH or 0.5 mol L-1H2SO4.Greenhouse conditions were kept stableviaautomated control of the air temperature and aeration.The experiment adopted a randomized block design with three replications.Low and high concentrations of N(0.3 and 3.0 mmol L-1)were set up.The plants were harvested 14 d after transplanting.The evaluated variables were dry shoot and root biomass,total N,and root morphology using WinRHIZO Pro 2016(Régent Instrument Inc.,Canada)coupled to a professional flatbed scanner(EPSON Expression 11000XL,Epson,Japan).A resolution of 400 dpi was used for root morphology measurements as described by Bauhus and Messier(1999)and Boumaet al.(2000).The evaluated characteristics were root length,volume,diameter,projected area,surface area,number of tips,ratio between root length and volume,and number of bifurcations.

Statistical analyses

The field experiments adopted a randomized block design with four replicates.A regression analysis of the treatments CK,N40,and N80 was performed to test the effect of N on the production variables.After that,the least significant difference(LSD)test was used(P＜0.05)in a factorial scheme 5×2×2(treatment×genotype×season)except for the site in RR,for which a factorial design of 5×2(treatment×genotype)was used because all experiments at this location were carried out during the same harvest season.Next,a third analysis was performed considering all sites including RR in the same design and factorial scheme 5×2×2(treatment×genotype×season).

Pearson’s correlation coefficient(r)value was determined to identify the highly correlated variables for the greenhouse experiment.This analysis was also performed on data from the field experiments regarding the yield of each treatment(I,N40,IN40,and N80)compared to that of the control.

Redundancy analysis(RDA)was used to analyze the productivity of each treatment,in which productivity was used as response variable and the chemical and covariate variables(fertility and climate variables)were used as explanatory variables.Statistical analyses were performed using Canoco software version 4.5(Ter Braak and Smilauer,2002).

RESULTS

Analysis of variance(ANOVA)results of the 18 inoculation experiments are display in Table III.As expected,grain yield responded to the application of 40 or 80 kg N ha-1(Tables IV and V,Fig.2)at all four sites in almost every case.The strongest responses to N fertilizer occurred at the DF site with N applied at 80 kg N ha-1.The grain yield of the simple hybrid BRS1030 increased from 8.4 to 9.5 t ha-1and its total grain N increased from 131 to 146 kg N ha-1(Table V).The variety BRS106 at this site was even more responsive to the addition of 80 kg N ha-1.Though at a lower level,its grain yield increased from 6.5 to 8.5 t ha-1and total grain N from 100 to 132 kg N ha-1.Dry season crops produced lower yields at this site but responded well to N fertilizer application(Fig.2).

At RJ,the mean yields of genotype BRS1030 showed a linear response(six experiments)to the N application of 80 kg N ha-1(Fig.2),and yield increased from 4.041 to 4.946 t ha-1,resulting in a gain of 0.9 t ha-1(Table V).The BRS106 at this site responded less strongly to N fertilization(Fig.2);however,in the rainy season(three experiments)grain yield increased from 3.683 to 4.477 t ha-1.The heterogeneity and poor physical structure of the soil at the field station of Embrapa Agrobiology in Seropédica(RJ)is a challenge with regard to the uniformity of experiments,particularlyfor experiments using maize.At MG,the mean grain yield increased in response to 80 kg N ha-1in all five experiments on both maize genotypes(BRS1030 and BRS106),and yields ranged from 5.776 to 7.663 t ha-1and the increase ranged from 0.7 to 1.1 t ha-1(Tables II and III).In contrast to at RJ and DF,at this site the variety BRS106 showed a tendency to produce higher grain yields than BRS1030.At RR,the simple hybrid BRS1010 showed a strong response to 80 kgN ha-1with mean yields increasing from 4.408 to 5.694 t ha-1,whereas the response of the variety BRS4157 was considerably weaker(Table V).However,with the addition of 40 kg N ha-1of fertilizer,there were significant(P＜0.05)increases in grain yield and grain N content of the hybrid BRS1030 for the means of five experiments at MG and for the N accumulated in the grain by this same genotype in the six experiments performed at RJ.However,no significant responses to inoculation with or without the application of 40 kg N ha-1were observed in the variety BRS106 at any site(Table V).

Fig.2 Grain yields of maize in response to N fertilizer application at the four experimental sites in Brazil:Seropédica,Rio de Janeiro State(RJ),Sete Lagoas,Minas Gerais(MG),Planaltina,Distrito Federal(DF),and Boa Vista,Roraima(RR).At each site,two maize genotypes were grown:a variety(V)and a hybrid(H).Maize was grown in both rainy season(RS)and dry season(DS)at RJ,MG,and DF while only in RS at RR.

TABLE IIIAnalysis of variance(ANOVA)results of the 18 inoculation experiments with the endophytic diazotrophs Herbaspirillum seropedicae strain ZAE94 using two maize genotypes(G)under different treatments(T,N level and inoculation of ZAE94 or not)conducted during rainy and dry seasons at Seropédica,Rio de Janeiro State(RJ),Sete Lagoas,Minas Gerais(MG),Planaltina,Distrito Federal(DF),and Boa Vista,Roraima(RR),Brazil

TABLE IVRegression models showing the responses of grain yield(y,t ha-1)of two maize genotypes to the application level of N fertilizer(x,kg N ha-1)at the four experimental sites,Seropédica,Rio de Janeiro State(RJ),Sete Lagoas,Minas Gerais(MG),Planaltina,Distrito Federal(DF),and Boa Vista,Roraima(RR),in Brazil

TABLE VMaize grain yields and grain N contents in the different treatmentsa)a at the four experimental sites,Seropédica,Rio de Janeiro State(RJ),Sete Lagoas,Minas Gerais(MG),Planaltina,Distrito Federal(DF),and Boa Vista,Roraima(RR),in Brazil

The combined results of all 16 experiments conducted at RJ,MG,and DF in the rainy and dry seasons showed that BRS1030 and BRS106 produced a significantly higher(P＜0.05)grain yield when inoculated withH.seropedicaestrain ZAE94 and fertilized with 40 kg N ha-1than noninoculated and non-N fertilized maize(Fig.2).The genotype BRS1030 showed a trend of higher grain yields when inoculated and receiving 40 kg N ha-1,but this trend was negative for the BRS106 variety.Considering only the means of the results of all dry season experiments(n=9),the grain yields were lower,but when the BRS1030 was fertilized with 40 kg N ha-1,the grain yields were significantly higher in the inoculation treatments(Table V).The yield of this genotype with inoculation plus 40 kg N ha-1was not significantly lower than at 80 kg N ha-1without inoculation.The grain yield of BRS106(means of nine dry season experiments)responded positively to 40 kg N ha-1with almost no further increase in the 80 kg N ha-1treatment,but the inoculation of ZAE94 with 40 kg N ha-1of fertilizer showed tendency to decrease yields.

In the combined analysis of the two experiments performed at RR,the BRS1010 hybrid produced approximately 50%more grain than the variety BRS4157(Table V).Application of N or inoculation of the variety BRS4157 did not increase crop yield at this location.BRS1010 showed positive response in the absence of N fertilization after a single application ofHerbaspirillumand produced an amount of grain equal to that of the treatment receiving 80 kg N ha-1.Nitrogen fertilization was not required at this location in order to increase grain yield,as inoculation seemed to be sufficient.

Crop productivity related to edaphoclimatic conditions showed substantial variability and formed three groups of dispersion data of all experiments performed over four years(Fig.3).Group I comprised the experiments performed at RJ and RR.Organic C content and rainfall regime differed between these locations,but both sites showed the lowest soil fertility and inoculation plus N application produced the largest increases in plant yield.This is reflected by the largest number of points above the regression line between grain yield in CK(x)and that in the other four treatments(y).The N80 and IN40 treatments showed the highest production averages(Fig.3).Group II comprising the crop productivity of MG and group III of DF showed the smallest yield increments but the highest crop productivity.

Fig.3 Responses of maize grain yield to N fertilizer and/or inoculation with the endophytic diazotroph Herbaspirillum seropedicae strain ZAE94 in the experiments conducted at the four sites,Seropédica,Rio de Janeiro State(group I),Sete Lagoas,Minas Gerais(group II),Planaltina,Distrito Federal(group III),and Boa Vista,Roraima(group I),in Brazil.CK=not inoculated and not fertilized as a control;N40=not inoculated but fertilized at 40 kg N ha-1;N80=not inoculated but fertilized at 80 kg N ha-1;I=not fertilized but inoculated;IN40=inoculated and fertilized at 40 kg N ha-1.The dash line is the regression line.See Table I for details of the experiments.

An RDA was used to test the relationships between five production variables and 21 environmental variables in a matrix(r=0.95,P=0.002)in the soil system(Table VI).Not considering the environmental variables,RDA 1 explained 88.2% of the variability of the data(Fig.4).The environmental variables explained 89.1%of the variability of the production data.Of this,98.9% was explained by RDA 1 and 0.6%by RDA 2.In addition,the biplot showed that all variables of productivity were correlated.However,the highest yields were influenced by soil organic C content,altitude,and water accumulation(upper right quadrant).Next to these variables were the experiments in DF and MG,which were associated with the highest averages of crop productivity.The variables temperature and pH,at the other end of the production variables,were negatively influenced and were positioned lower in the lower right quadrant(Fig.4).Close to these variables were the experiments conducted in RR and RJ.

Fig.4 Redundancy analysis(RDA)of the responses of maize yield to N fertilization and/or inoculation with the endophytic diazotroph Herbaspirillum seropedicae strain ZAE94 and other environmental variables.See Table VI for the RDA variables.

In the experiments at RJ with both maize varieties,the meanδ15N of the three weed species was significantly higher than that of leaves or stems of maize plants inoculated withHerbaspirillum;moreover,in non-inoculated maize plants,the stems showed significantly lowerδ15N values than the mean value of the weed species,which was also true for four out of the eight maize leaf samples.The results for maize shoots were based on the arithmetic mean of the leaf and stalk data,and not on the weighted mean,as no whole plants were harvested.The most straightforward interpretation for these lowerδ15N values in the maize plants compared to(presumably)non-N-fixing weed species would be that the maize plants obtained N by BNF,and inoculation usingHerbaspirillumenhanced this contribution.Using the overall means of the four experiments,an interpretation of the results in this manner would suggest that without inoculation,maize plants obtained 14.2%and 20.4%of their N from BNF during the rainy and dry seasons,respectively,and when treated with theHerbaspirillum,these values increased to 28.4%and 31.2%,respectively(Table VII).The interpretation of theseδ15N results is discussed below.

TABLE VISymbols and descriptions of variables included in the redundancy analysis(RDA)

To better understand the effects of inoculation withH.seropedicaeZAE94,a hydroponic experiment was conducted to measure root growth in addition to mass gain.Of the eight variables evaluated using root scan software,four were modified by inoculation and N addition,apart from the increase of leaf area(Table VIII).Particular attention was paid to three variables that reflected the increase in plant productivity:leaf area with a 19% increase,root volume increasing by 25%,and fine root(diameter of 0–1.5 mm)number increase,which are responsible for the greater absorption(9% increase)of water and nutrients when inoculated.These results directly reflected the dry matter gain of aerial parts(27%)and roots(8%),as observed by the positive correlations of these variables(Table IX).

DISCUSSION

Although the yields obtained in all experiments were low compared with the potential yield of maize,they werecomparable with Brazilian mean yields.The experiments were applied in order to test the agronomic efficiency ofH.seropedicaeZAE94 inoculation on maize grown on four different soil types.Crop productivity withoutHerbaspirilluminoculation was the highest in the field trial at MG,reaching values above 7 t ha-1with 80 kg N ha-1(Table V,Fig.2).Similar grain yield was achieved at DF(6–7 t ha-1),particularly in the non-inoculated treatments.Crop productivity at RJ during several years of field trials in both rainy seasons in the summer and dry seasons in the winter(second harvest)in a sandy soil commonly classified as Alfisol reached 4–5 t ha-1,as yield was limited by the edaphoclimatic conditions at this site.Alveset al.(2015)used two maize genotypes in the same soil and reported yields of approximately 4 t ha-1with 80 kg ha-1of N application.These differences between sites reflect the edaphoclimatic conditions and may be explained by the physicochemical properties and genealogy of the soils(Table II).For example,the Alfisol at RJ inhibits plants growth and development due to its physicochemical characteristics such as low pH(5.6),less than 10 g kg-1organic matter,shallow(40–50 cm)depth,and high sand/silt content(approximately 58%)(de Macedoet al.,1998).Oxisols have excellent physical characteristics,which,coupled with the flat or gently rolling terrain,facilitate high-input cropping systems in these areas.The cultivation of maize that is adapted to these regions typically produces high yields in comparison to crops planted on Alfisols.Oxisols are typically acidic and dystrophic and therefore require liming to prevent Al toxicity and fertilizing due to the high degree of weathering(Santoset al.,2006).

TABLE VIIAverage values of 15N natural abundance(δ15N)in leaves and stems of maize plants and reference plants at Seropédica,Rio de Janeiro State,Brazil in four experiments,two in rainy season and two in dry season,in consecutive years of 2005 and 2006 using two maize genotypes,a hybrid and a variety

TABLE VIIIVariables evaluated in maize plants after a 14-d hydroponic experiment where the plants were inoculated with the endophytic diazotroph Herbaspirillum seropedicae strain ZAE94(inoculated)or not(control)

TABLE IXPearson’s correlation coefficient matrix of independent variables of the hydroponic experiment where maize plants were inoculated with the endophytic diazotroph Herbaspirillum seropedicae strain ZAE94(inoculated)or not(control)

All experiments reported in the current study used two simple hybrids and two varieties to evaluate the effect ofH.seropedicaeinoculation at four locations(Fig.2).Maize productivity has increased over the last few decades due to the contribution of plant breeding for disease resistance and adaptability to limiting factors commonly found in tropical soils such as Al,and for resistance to environmental stresses such as drought(Fahadet al.,2017).Varieties rather than double-hybrids are usually recommended for small farmers in order to reduce the cost of seeds.The data presented in this study are from varieties and simple hybrids and hence more relevant to smallholders who can rarely afford high N fertilizer input.

The number of experiments testingHerbaspirillumapplication under field conditions is limited.Previous studies indicated thatHerbaspirillumspecies may be promising for agricultural application.The work by Pereiraet al.(1988)was one of the first studies that used anH.seropedicaeinoculant on sorghum and compared it with the inoculation ofAzospirillum lipoferumandNitrospirillum amazonense(formerAzospirillumgenus).Their results showed that the Z95 strain(H.seropedicae)promoted better germination of inoculated plants.This same ZAE94 strain was used as inoculant for rice(original host plant),resulting in a grain yield increase similar to that obtained with the application of 40 kg N ha-1(Baldaniet al.,2000;Guimarãeset al.,2010).Inoculation of sugarcane with a differentH.seropedicaestrain,HRC54,along with another PGPR species,Gluconacetobacter diazotrophicus(strain PAL-5T),stimulated the activity of H+-ATPase and promoted morphological changes in roots thereby resulting in higher nutrient uptake and biomass accumulation at the initial setting(Canellaset al.,2013).Dottoet al.(2010)testedH.seropedicaestrain Sm1 on maize with increased N levels in two hybrids and found that plant response depended on the genotype.

The results of the four experiments at RJ,where weeds and samples of maize leaves and stems were taken for analysis of15N abundance,suggest that the maize plants,even without inoculation withHerbaspirillum,may have obtained significant inputs of atmospheric dinitrogen(N2)viaplant-associated BNF(Table VII).If maize was a nodulated legume,this conclusion would probably be accepted unquestioned.However,it is necessary to consider causalities other than BNF to explain the lowerδ15N values in maize compared to weeds.The weeds almost certainly had been growing in this soil for a shorter time than the maize.In the rainy season,N taken up by the weeds was higher than in the drier and cooler season;however,there is no evidence to suggest that the δ15N of available soil N increased with time during any of the crop growth periods.The weeds almost certainly had more shallow rooting systems than maize plants,therefore the15N abundance of plant-available N decreasing with depth may explain the lower δ15N values in maize.These experiments on maize were conducted on the same soil type and within 200 m of two other studies to quantify the contribution of BNF to sugarcane(Urquiagaet al.,2012;Baptistaet al.,2014)and elephant grass(Pennisetum purpureum)(de Moraiset al.,2012).In the study on sugarcane,soil samples were taken down the profile at 10-cm intervals to 60 cm depth,with a final sample from the 60–75 cm layer.In the study on elephant grass,samples were taken at 10-cm intervals to 30 cm depth,and then at 20-cm intervals down to 70 cm.The soil samples were placed in small pots(400 g soil per pot),fertilized with P,K,and micronutrients,and three or four different weed species were planted.Therefore,the δ15N values of the plants grown in the soil samples taken from the different depths were assumed to represent the15N abundance of the plant-available N in the soil from the respective depth interval(Ledgardet al.,1984).In each case and in all weed species,the δ15N values increased by approximately 2‰with depth to 50 or 60 cm and then stabilized or,in the study on elephant grass(de Moraiset al.,2012),decreased by approximately 0.9‰comparing the 30–50 cm layer to the 50–70 cm layer.Thus,depth of the rooting system does not seem to explain the lower15N abundance in maize compared to weeds.

The15N natural abundance results also show that there was a significant decrease in15N abundance in maize plants when they were inoculated withHerbaspirillum(Table VII).These results suggested that inoculation withHerbaspirillumincreased the contribution of BNF to maize plants when no N fertilizer was added,but this hypothesis was not tested for plants which were N-fertilized.However,there was no significant increase in maize yield or total grain N associated with the apparent increase in the BNF input in the experiments conducted at the RJ site.

The combined results of the 16 experiments where maize was fertilized with 40 kg N ha-1suggest positive effect of inoculation withHerbaspirillumon grain yield of the BRS1030 hybrid,and this effect was statistically significant(P＜0.05)in the nine experiments conducted in the dry or cooler season(Figs.2 and 3).

Nitrogen fertilization is assumed to inhibit BNF and reduce the diazotrophic bacteria population.Medeiroset al.(2006)found a negative effect on populations ofGluconacetobacter diazotrophicusat high N doses applied to sugarcane.Muthukumarasamyet al.(1999)also observed a negative effect of N fertilization on the natural population ofHerbaspirillumassociated with sugarcane planted in India.Kirchhofet al.(1997)found that natural diazotroph populations,includingHerbaspirillum,in potential bio-energy plants such asMiscanthusandSpartinadecreased when plants were N-fertilized.

One explanation for the increased inoculant contribution to the medium level of N application is root structure modification,as observed in other diazotrophs.Riggset al.(2001)found that inoculation withH.seropedicaecaused an increase in dry matter from 49%to 82%when applied in combination with N fertilization,compared with a 16%increase obtained when plants were only inoculated but not fertilized.Dobbelaereet al.(2002)found that under controlled conditions,the effect of inoculation withAzospirillum brasilensestrain Sp 245 andNiveispirillum irakensestrain KBC1(formerAzospirillumgenus)was stronger when associated with lower N levels.Regarding the change in nutrient absorption,particularly N,an improvement in plant N uptake capacity through modification of the root architecture(Table VIII)was previously suggested by Lea and Azevedo(2006),particularly in narrow and deep root systems(such as＞2 m)(Hammeret al.,2009).Inoculation with diazotrophic PGPRs such asAzospirillumandHerbaspirillumstrains improves several parameters of maize root architecture(Dobbelaereet al.,2002),particularly in the presence of humic acids(da Conceiçãoet al.,2008).

Inoculation withA.brasilensestrains has been used in South America for several years,including the recommended Az 36 in Argentina and Abv5,Abv6,and Sp 245 in Brazil(Hungriaet al.,2010).These recommendations are based on the data of Sumner(1990)and Okon and Labandera-González(1994),who evaluated 20 years of data indicating that 60%–70% of the field experiments were successful,with yield increasing by 5%–30%.In our study,the 36 experiments(considering hybrids and varieties separately),42%of the field trials showed a positive effect of inoculation withH.seropedicae(Table V).Considering only N fertilizer treatments compared with the control,positive effects were observed in 64%of all cases(23 experiments).

CONCLUSIONS

The data presented here can be used as an agronomical recommendation for diverse tropical regions of Brazil and other countries with similar soil and climate conditions.There was a better chance of yield responses to inoculation withHerbaspirillumunder low N fertilizer application under adverse climate conditions.Herbaspirilluminoculation is substantially cheaper than an additional amount of 40 kg N ha-1of N fertilizer;therefore this treatment can be recommended for dry season under low N fertilization,and for smallholders who at present use only low-input N fertilization,inoculation withHerbaspirillummay be cost effective.

The evidence from RJ strongly suggested that the application of theHerbaspirilluminoculant increased the contribution of BNF to plant growth.However,there was no observable response to inoculation when the crops were not fertilized with N.In contrast,the best response to inoculation was observed under fertilization at 40 kg N ha-1,which would suggest thatHerbaspirillumstimulated root growth and nutrient uptake,rather than BNF,as observed in the hydroponic experiment.Based on these data,inoculation with this diazotrophic plant growth-promoting bacterial strain can also help increase the efficiency of N fertilizer use and improve grain yields using half the amount of N fertilizer,which also reduces N loss to the environment.Further studies are needed to measure BNF in maize plants under low-N fertilization.

ACKNOWLEDGEMENTS

This work was funded by Brazilian Agriculture Research Corporation–Embrapa,the National Research Council(CNPq),Brazil(No.465133/2014-2),Newton Fund“Understanding and Exploiting Biological Nitrogen Fixation for Improvement of Brazilian Agriculture”(No.B/N012476/1),the Biotechnology and Biological Sciences Research Council(BBSRC),Brazil,the Brazilian National Council for State Funding Agencies(CONFAP),and the Coordination of Improvement of Higher Education Personnel(CAPES),Brazil(No.001).The study is published with the permission of the Embrapa Agrobiology Editorial Committee of Publication.