EFFECT OF FERTILIZER-N APPLICATION AND SEED COATING WITH RHIZOBIAL INOCULANTS ON SOYBEAN YIELD IN EASTERN PARAGUAY ( 1 )

A remocao do N em graos de soja na colheita pode exceder a fixacao biologica do N2, particularmente se os rendimentos dos graos forem tao elevados quanto os obtidos tipicamente nos solos "Terra Roxa" da regiao oriental do Paraguai. A aplicacao do fertilizante nitrogenado ou a inoculacao das sementes com rizobios que estimulem a nodulacao podem representar uma maneira de equilibrar o balanco de N, porem o efeito de tais tratamentos parece ser altamente especifico do local. O objetivo deste estudo foi examinar os efeitos da aplicacao do nitrogenio (N) e da inoculacao com rizobios (I) na nodulacao, na acumulacao de N e nos rendimentos de soja sob condicoes da regiao oriental do Paraguai. A pesquisa em campo foi conduzida durante duas safras consecutivas de soja. As condicoes de seca ou estiagem no primeiro ano atrasaram a semeadura e reduziram o numero de plantas por m2 e o numero de vagens/planta. Os rendimentos de graos foram geralmente inferiores a 2 t ha-1; no entanto, o tratamento + N + I aumentou o rendimento em aproximadamente 75 %. Nas condicoes favoraveis do segundo ano, os rendimentos foram em torno de 4 t ha-1 e os tratamentos nao tiveram nenhum efeito. A acumulacao de N foi mais elevada no primeiro ano e, portanto, nao pode explicar as diferencas observadas no rendimento entre anos e combinacoes de tratamentos. O efeito positivo do tratamento + N + I no primeiro ano foi associado com o crescimento mais rapido da raiz, que poderia ter reduzido a suscetibilidade ao estresse hidrico intermitente. A biomasa dos nodulos diminuiu entre a floracao e a maturacao das vagens no tratamento +I, enquanto aumentos adicionais na biomasa dos nodulos no tratamento -I podem ter conduzido a competicao pela assimilacao entre nodulos e desenvolvimento das vagens. Com base nesses resultados preliminares, conclui-se que a aplicacao de N e a inoculacao da semente podem oferecer beneficios a curto prazo em anos desfavoraveis, sem apresentar efeitos negativos no rendimento em anos favoraveis.


INTRODUCTION
Soybean (Glycine max.[L.] Merr.) is the most important cash crop grown on the red clayey "Terra Rossa" soils of Eastern Paraguay and, due to a combination of favorable climate and high soil fertility, yields are among the highest in the world (FAO, 2004).Typically, soybean is grown in the summer intercropped with wheat in the winter.A no-tillage farming system was used for both crops.Under continuous no-till system, the bulk density and penetration resistance in "Terra Rossa" soils tend to increase (Kubota et al., 2005a).With high soil compaction it is possible that N fixation is reduced by insufficient oxygen supply to soybean roots and nodules (Voorhees et al., 1976).This poses the threat that soybean plants become increasingly dependent on soil-N rather in detriment of N 2 fixation to satisfy their high N demand.
Several studies of N budgets in soybean-based systems reported that N removal at harvest may exceed biological fixation (Harper et al., 1989;Peoples & Craswell, 1992;Vanotti & Bundy, 1995).Vanotti & Bundy (1995) further showed that soybean can stimulate microorganisms that enhance soil N mineralization.As a result soil N fertility can decline over years, particularly if yields are high and if a second winter crop such as wheat also exports a considerable amount of N (Kubota et al., 2005b).The combination of high rates of N removal with a potential reduction of biological N 2 fixation due to soil compaction may make it necessary to balance N budgets, either through N fertilizer application or by enhancing nodulation and N 2 fixation.However, no study has reported on the effect of N fertilization on soybean in Eastern Paraguay.
Most studies on the effect of fertilizer-N on soybean growth and N 2 fixation by rhizobium conducted elsewhere concluded that fertilization reduces N fixation through a reduction in the number, weight and activity of nodules (Starling et al., 1998;Chen et al., 1992).However, this effect depended on a variety of factors such as amount of N applied and type of soil, climate, and farming system.A few studies showed that small amounts of starter-N applied at sowing could improve root growth and N uptake by soybean prior to full nodulation (Hardarson & Zepata, 1984;Wani et al., 1995;Gulden & Vessey, 1998).Furthermore, positive responses to N fertilization have been reported when soybean yields are affected by drought (Lyons & Earley, 1952;Purcell & King, 1996).
Coating soybean seeds with rhizobial inoculants has been suggested as a way to improve N 2 fixation (Dobereiner et al., 1995;Seneviratne et al., 2000) but the success of inoculation was found to be highly variable (Peoples & Craswell, 1992).In Brazil, effects of starter N and rhizobial inoculation have been tested repeatedly.Starter N increased N accumulation in plants but grain yields usually remained unchanged.Inoculation increased soybean grain yields in the Cerrado region of Brazil by up to 750 kg ha -1 (Coutinho et al., 1999).This positive effect has been confirmed in other studies and seed coating with inoculants is now being recommended as a standard practice for soybean growers in Brazil (Alves et al., 2003).
However, so far no study on the effect of N fertilization or inoculation on soybean has been reported for Eastern Paraguay and rhizobial inoculation is not yet a common practice among soybean growers of that region.The objective of this study was to examine the effect of starter-N and rhizobium inoculation on nodulation, N accumulation and soybean yield under the conditions of Eastern Paraguay.

MATERIALS AND METHODS
Field experiments were conducted in two soybean growing seasons between January 2000 and May 2001 at Centro Tecnologico Agropecuario en Paraguay (CETAPAR), an agricultural research and extension institute located in the Yguazú district of Eastern Paraguay.The soil in the field (S25 ° 27 ' 22.8 '' W0 55 ° 02 ' 42.3 '') is clayey, classified as Rhodic kandiudox according to the U.S. Taxonomy, but locally referred to as "Terra Rossa".Texture and other soil properties are shown in table 1.The plot had been used for soybean -wheat cropping under a notillage farming system for seven years.Before planting the wheat winter crop, a compound fertilizer was applied to the soil surface every year at a rate of 36 kg ha -1 N and 92 kg ha -1 P. No fertilizer was applied until soybean sowing prior to this study.Treatments were arranged in a split-plot design with three replicates with starter N application as the main plot (N) and rhizobial inoculation as subplots (I).Soybean (Glycine max.[L.] Merr.) was planted in all plots after N application.The area of each subplot was 32 m 2 (4 by 8 m).A neighboring field plot was used in the second year.
N fertilizer was applied as urea at a rate of 50 kg ha -1 N just before sowing.No other fertilizers were applied before planting soybean.Two commercial rhizobial inoculants with peat carrier were used in this study: 'Rizoplus Super' in the 1999/2000 growing season (3 g kg -1 seeds), and 'Biomax' in 2000/2001 (2.5 g kg -1 seeds).'Rizoplus Super' contains Bradyrhizobium japonicum while 'Biomax' contains Bradyrhizobium japonicum and Bradyrhizobium elkanni.The inoculants were mixed with seeds before sowing.Different inoculants had to be used due to their varying availability in the country.
In the untilled plots, seeds were mechanically sown (55 kg ha -1 ) at a row spacing of 35 cm.Planting and harvesting were timed according to weather conditions and grain maturity.A soybean cultivar suitable for late sowing (FT2001, indeterminate type) was used because dry conditions delayed sowing by six weeks in the 1999/2000 growing season.Seeds of FT2001 were sown on January 4 and harvested on May 23, 2000.In the second year, cultivar Aurora (determinate type) was sown on November 15, 2000 and harvested on March 29, 2001.
In 1999/2000, shoots and roots were sampled twice at the flowering stage (March 7, 63 days after sowing; DAS) and at the pod-setting stage (April 3, 90 DAS).In 2000/2001, they were sampled three times at the 7-leaf stage (Dec.20, 35 DAS), at flowering (January 17, 63 DAS) and at pod-setting (March 5, 110 DAS).At all sampling dates we sampled five continuous plants from the second row in each plot.Root samples were taken at a depth of 30 cm and nodules were removed by hand from the roots.All plant parts were oven-dried for 48 h at 60 °C and weighed.Dried samples were ground to pass a 2 mm mesh for total N analysis.The ground samples were digested with H 2 SO 4 , and analyzed by the Kjeldahl Table 1.Properties of the "Terra Rossa" soil at the experimental site (CETAPAR, Yguazú District, eastern Paraguay) (1) Classified by USDA system. (2)Identified by the Munsell Soil Color Chart.
method using an auto distillator (MRK-MATIC Auto Vapor-Still).At harvest, all above-ground dry matter was sampled from a 1 m 2 area per plot.
All data were analyzed within each year representing split-plots, with the starter N treatment (N) as main factor and the inoculum treatment (I) as sub-factor.A set of linear contrasts was used to compare the + N + I treatment and means of other treatment combinations.

Growing season 1999/2000
The 1999/2000 season was abnormally dry during the regular sowing period (November) and again in March during flowering/early pod development (Figure 1).Due to the lack of rainfall in November 1999, planting was delayed by six weeks and an early pod setting cultivar, FT 2001 (indeterminate type), was used, suitable for late sowing.
At flowering shoot dry weight was not affected by treatments, whereas the combination of starter N with rhizobium inoculation (+ N + I) significantly increased root dry weight (Table 2).Nodule weight responded positively to inoculation but at this early growth stage this did not increase N accumulation.At pod setting no significant difference was detected for root, shoot or nodule dry weight but while nodule weight more than doubled between flowering and pod setting stages in the -I treatment, nodule weight decreased by 28 % in the +I treatment.Total N accumulation did not differ significantly between treatment combinations but in the + N + I treatment significantly more N had been translocated to pods.
The grain yield in 1999/2000 was generally low but N application and seed inoculation had positive effects, and their combined application (+ N + I) increased yield to about 3 t ha -1 , i.e., a 75 % improvement over the other three treatment combinations (Table 3).A yield level of 3 t ha -1 is close to the average yield of the region (Seki, 1999).The + N + I treatment tended to a higher plant density and higher pod number per plant than the other treatments (data not shown), and both factors probably contributed to the significantly higher grain yield.The harvest index (HI) was also highest in the + N + I treatment.

Growing season 2000/2001
The 2000/2001 season can be considered a typical year for the region with evenly distributed rainfall throughout the growing season (Figure 1).Sowing could be completed during the regular period in November and the locally preferred cultivar Aurora (determinate genotype, high yield potential) was used.
In samples taken at the 7-leaf stage the +N treatment significantly increased shoot weight and N accumulation but nodule weight was reduced by more than 80 % compared to the -N treatment (Table 4).
Soybean compensated these initial differences in later growth stages.No significant treatment effects for shoot or root dry weight were detected at flowering or pod setting.Unlike in the 1999/2000 growing season nodule weight increased throughout the season and reached a maximum at pod setting.However, at flowering nodule biomass was significantly higher in the inoculated plants.Grain yield in the 2000/2001 season was much higher than in year one and treatments had no significant effect on grain yield or yield components (Table 5).

DISCUSSION
Benefits of inoculation and N application are generally thought to be higher if N is a limiting factor, either because of low N availability in the soil (Peoples & Craswell, 1992) or because of large N demand of high yielding cultivars (Dashti et al., 1998).However, we could not confirm such a positive effect in the highly productive second year where estimated yields exceeded 4 t ha -1 .Instead, benefits of the + N + I treatment were detected in an unfavorable year when two factors, delayed sowing with an indeterminate genotype and an intermittent dry period during flowering/pod setting coincided, leading to reduced yields.A positive response to starter N application has been reported for each of these factors.Starling et al. (1998) showed that a starter application of 50 kg ha -1 N increased growth and grain yield of an indeterminate soybean type planted late.Sorensen & Penas (1978) found in on-farm trials that sites with severe drought and low yields were most responsive to N fertilizer, while sites with high yield levels did not respond to N fertilizer.Similarly, Purcell & King (1996) observed that N fertilizer application had no effect on irrigated plots but that drought-stressed soybean benefited from fertilizer N due to lower flower and pod abortion rates.They concluded that application of N fertilizer increased drought tolerance of soybean because of the extreme sensitivity of N 2 fixation to drought.
Contrary to these observations N fertilization alone did not increase yields here.A positive effect of N application was only detected when seeds had been coated with rhizobial inoculants.This raises the question about the nature of additional benefits incurred by inoculation.Differences in soybean yields between years could not be explained by differences in N accumulation.On the contrary, high yields in the second year were achieved despite much lower total N accumulation at pod setting (Tables 2 and 4), indicating that N was probably not a yield limiting factor.It appears that soybean was able to satisfy N requirements without N application or inoculation on the "Terra Rossa" soil of the study area.
Based on our limited data set a possible explanation for the positive effect of inoculation in the 1999/2000 season could be linked to the sensitivity of soybean to water deficits during the flowering stage when plants typically react with flower abortion (Raper Jr. & Kramer, 1987;Kokubun et al., 2001).Inoculated plants showed a slight reduction in nodule biomass between flowering and pod setting whereas nodule weight of plants in the -I treatment more than doubled during that period (Table 2).The increase of nodule biomass could have represented an additional stress factor for plants in the -I treatment as developing nodules are strong competitors for assimilates (Marschner, 1999).Negative effects on pod development are likely to increase if this competition between nodules and developing pods coincides with a water deficit.More detailed studies will be needed to confirm this hypothesis.
In addition to short-term and season-specific effects, N application could have some long-term benefits for the fact of adding N to a cropping system characterized by a negative N balance (Vanotti & Bundy, 1995;Zotarelli et al., 2002;Kubota et al., 2005b).

CONCLUSIONS
1.No effect of N application or inoculation on grain yield was detected in a favorable year with high yields.
2. Starter N application together with seed coating with rhizobial inoculants is a viable option for delayed sowing.
3. The + N + I treatment combination appears to offer some protection against intermittent drought in the shortened growth period after delayed sowing.

Figure 1 .
Figure 1.Monthly precipitation during the 1999/2000 and 2000/2001 growing seasons, compared to the 30 year average of CETAPAR.