EVALUATION OF INTEGRATED NITROGEN AND PHOSPHOROUS MANAGEMENT USING THE TT BIPLOT METHOD IN SOYBEAN

: To investigate the effects of integrated nutrient management on oil, protein, grain yield and some traits of soybean, we conducted a factorial experiment involving 4 bio-fertilizer (no inoculation, inoculation with Barvar-2, inoculation with Biosoy and dual inoculation with Biosoy and Barvar-2) and 3 chemical fertilizer levels (no chemical fertilizer, 66 kg ha -1 diammonium phosphates + 50 kg ha -1 urea, 132 kg ha -1 diammonium phosphates + 100 kg ha -1 urea) with 100% chemical fertilizer (200 kg ha -1 diammonium phosphates + 150 kg ha -1 urea) as control base in a randomized complete block with four replications at the research farm of the University of Mohaghegh Ardabili, Iran. Analysis of variance showed that treatment combinations affected most of the traits (P>0.01). We used the biplot analysis as the treatment × trait (TT) biplot to determine the best treatment combinations and traits. The first two principal components (PC1 and PC2) explained 94 and 96% of the total variant of the standardized data in 2017 and 2018, respectively. Accordingly, application of Biosoy and 150 kg ha -1 urea + 200 kg ha -1 diammonium phosphate significantly increased stem height at harvest, number of grains per plant, biomass, hundred-grain weight, oil and protein yield, protein percent and grain yield compared to the other treatment combinations. The results showed that there were positive correlations between these traits. Also, non-inoculated plants and no chemical fertilizer treatments significantly increased oil percent. The results indicate that higher-yielding treatment combinations had lower oil percent. The biplot was used for ranking of treatment combinations based on a single trait. These study results suggest that biofertilizers had a positive influence on soybean and that they could diminish the use of chemical fertilizers. The study reveals that the TT biplot was able to graphically show the interrelationships between traits and support visual comparison of treatments.


Introduction
Soybean (Glycine max (L.) Merr.) is one of the most important oilseed crops in the globe (Ali, 2010;SoyStats, 2019) since soybean grains are used for the production of oil and protein for humans, feed for livestock and biofuel for cars. Global production of soybean is expected to be increased to 371.3 million tons until 2030 (Masuda and Goldsmith, 2009). Approximately 76,000 hectares of the soil is under soybean cultivation in Iran with an average yield of 2.44 tons per hectare (FAOSTAT, 2013). There is an important scope to increase the area under soybean and grain yield of soybean in Iran (Shahraeen et al., 2005). Among the crucial nutrients, macronutrients such as nitrogen and phosphorus play a crucial role in improving crop growth and yield. Soybean nitrogen demands are met by biological N fixation and soil nitrogen (Milić et al., 2002;McCoy et al., 2018). Nitrogen is the most important element that crops need in the supreme amount and often restricts crop growth and development (Zhao and Shen, 2018) as well as crop yield. Nitrogen affects a lot of physiological processes in higher plants (Cechin and Fumis, 2004). Phosphorus (P) is also another important element for plants. Tsvetkova and Georgiev (2003) reported that P deficiency decreased soybean dry mass and nodule weight. Dry bean grain yield and yield components were significantly increased with P fertilization and associations of grain yield with shoot dry weight and pods per plant were significantly positive (Fageria et al., 2010).
Generally, nitrogen and phosphorus fertilizers were applied to replace soil N and P with the high price and ecological risk. Because of avoidance of environmental problems, people health, and further crop production to cover the growing demand of the global population, combined nutrient management using biological and mineral fertilizers is a proper method to meet crop needs. An ecological way to increase soil fertility and plant productivity might be the use of biofertilizers such as N2-fixing bacteria and phosphate solubilizing bacteria (Sessitsch et al., 2002). Biofertilizers are needed for sustainable farming systems (Schütz et al., 2018). Because of an increase in N and P uptake, nitrogen-fixing and phosphorus solubilizing bacteria are of importance for crops (Zahir et al., 2004;Zaidi and Mohammad, 2006). Therefore, application of these microbes as environmentally friendly bio-fertilizers can reduce the use of expensive nitrogen and phosphorous fertilizers which cause many environmental problems (Serpil, 2012). Also, phosphorus bio-fertilizers produce plant growth-promoting substances that increase P uptake, the efficiency of nitrogen fixation and availability of some microelements to plants (Kucey et al., 1989). Seed inoculation with Bradyrhizobium in soils with low N content without N fertilizer application increased biological nitrogen fixation (Smith, 1992). Co-inoculation of soybean and alfalfa by Pseudomonas strains and Rhizobia produced more nodules and nodule dry weight (Rosas et al., 2006).
The experimental dataset must be analyzed and interpreted effectively. Diverse methods typically cause a similar outcome for a given dataset (Sabaghnia, 2012). The main effect of genotype and genotype by environment (GGE) biplot graphical tool was developed by Yan et al. (2000) for the graphical analysis of multi-environment trials.
Influences of genotypes, environments or testers are shown by the biplot. The GGE effects of multi-environment trials are exhibited by the GGE biplot. It is constructed by plotting the first two principal components (PCs) which are commonly referred to as models that decompose the environment-centered data. (Yan et al., 2007). However, it can also be equally used for all types of two-way data that assume a two-way structure (Yan and Kang, 2003). The treatment combinations can be generalized as rows and the multiple traits, for example, morphologic and physiologic properties as columns. Yan and Rajcan (2002) used a genotype × trait (GT) biplot, which is an application of the GGE biplot technique to investigate the genotype by trait data and is an excellent tool for visualizing the genotype by trait data. The TT biplot was an excellent tool for visualizing treatment combinations by trait data.
The aim of the study was to examine the possibility of reducing the use of N and P fertilizers by using bio-fertilizers by the TT-biplot technique.

Field experiments
We conducted the experiment in a clay loam soil during 2017 and 2018 growing seasons. The previous crop was wheat. After harvesting wheat, the soil was plowed to a depth of 20 cm, and disked. The soybean seeds were sown manually on June 8, 2017 and June 15, 2018, in both sides of ridges of 50 cm apart, at a density of 50 seeds m 2 . Each plot consisted of five 4-m long rows, with 50 cm between rows. Weeds were controlled regularly by hand. The farm was irrigated on the basis of the conventional crop system. We arranged a factorial set of treatments on the base of a randomized complete block design with four replications. The biological fertilizer consisted of no-inoculation, inoculation with Barvar-2, inoculation with Biosoy and dual inoculation with Biosoy and Barvar-2. Different levels of the chemical fertilizer were applied: 0%, 33% (66 kg ha -1 diammonium phosphate + 50 kg ha -1 urea), 66% (132 kg ha -1 diammonium phosphate + 100 kg ha -1 urea) and 100% (200 kg ha -1 diammonium phosphate + 150 kg ha -1 urea) of chemical fertilizers. Combination of all 4 bio-fertilizer levels and 3 chemical fertilizer levels including 0%, 33% and 66% was arranged in a factorial experiment. The one-hundred chemical fertilizer level was used as control. Biosoy bio-fertilizer including effective Bradyrhizobium japonicum populations was purchased from the Mehr Asia Biotechnology Company and Barvar-2 bio-fertilizer, including two phosphate solubilizing bacteria Pseudomonas putida s-P13 and Pantoea agglomerans, was provided from Green Biotech Co. The population of bacteria in both bio-fertilizers was 10 8 per gram.
Soybean seeds (cv. Williams) were separated into four samples, and one sample was control (no inoculation). The other samples were inoculated with Biosoy, Barvar-2 and Biosoy + Barvar-2 according to the instructions on their packages, respectively. Chemical fertilizers were applied in bands on one side of rows after planting. Ten plants from central rows were randomly harvested to measure yield and yield components.

Results and Discussion
Analysis of variance Combined analysis of variance over two years exposed significant differences (P>0.01) among treatment combinations for all the traits but the grain number per pod (Table 1). The effect of the year on the studied traits was insignificant. Except for grain yield, treatment combinations × year interactions were insignificant for all other traits. The results indicated that increasing application of chemical fertilizers had a noteworthy influence on all traits and except for oil percentage, increased the value of all other traits. Combined inoculation with Biosoy and Barvar-2 was more effective than inoculation with a single microorganism. Seeds inoculated with Biosoy and Barvar-2 + Biosoy had greater values than noninoculated and inoculated seeds by Barvar-2 in most traits, but had lower values for oil percentage. It is notable that, in most traits such as grain and oil yields, application of Biosoy and Barvar-2 + Biosoy inoculums without any use of chemical fertilizers achieved higher values than 100% of chemical fertilizers. Usually, in inoculated plants with Biosoy and Barvar-2 + Biosoy, application of 33% of chemical fertilizers improved most of the traits but higher doses of chemical fertilizers did not improve the traits.

Biplot analysis
The treatment combination evaluation based on multiple characters is a vital feature of biplot analysis. An ideal treatment combination must meet numerous requirements. For example, not only yield but also end-use quality must be acceptable in a high-yielding treatment combination. Also, yield and quality can be divided into many components. For the target trait of some components, it is essential to define the most useful components to avoid unintended selection. Figures 1a and 1b show the records of 13 treatment combinations with 12 measured traits. The polygon view is based on PC axes (PC1 and PC2) in which the traits are the tester and the 13 treatment combinations are entries. The first two biplot axes in Figure 1 explained 94 and 96% of the entire deviation of the standardized data in 2017 and 2018, respectively.
There are five and seven vertex treatments in Figure 1a (The first year) and Figure 1b (The second year), four of them (B3f1, B1f0, B2f1 and B0f0) were good for several traits, but the others were not good or were unfavorable. The B3f1 was the best in terms of plant height at harvest (PH), number of grains per plant (NG), biomass (BM), hundred-grain weight (SW), protein percent (P%), protein content (PY), oil percent (OY), chlorophyll index (CH), and grain yield (GY), and therefore, it seems that dual inoculation with Biosoy and Barvar-2 × 66 kg ha -1 diammonium phosphate + 50 kg ha -1 urea could increase the value of many characteristics of soybean. In contrast, the B1f0 treatment combination (Barvar-2 + no chemical fertilizer) was the best for oil percent (0 %), and therefore, it seems that this trait is sensitive to the chemical fertilizer and Biosoy (Figures 1a and 1b).
The polygon view of biplot shows that B0f0 (no inoculation + no chemical fertilizer) was the best in terms of grain number per pod (NGP). The data indicated that the grain number per pod was not correlated with other traits (Figure 1). These results are in agreement with the Pearson correlation coefficients in Table 2. Dual inoculation with Biosoy + Barvar-2 significantly achieved the uppermost number of grains per pod with 2.51 and 2.45 grains, respectively. Our outcome is in contrast with Argaw (2012) who showed that the use of bio-fertilizers decreased the use of the chemical fertilizers. Each year, the traits fell into the same groups and the treatment combination × trait biplots were similar in the two years.
In Figures 1c and 1d  The interrelationships among all the measured traits are shown by the vector view of the TT biplot (Figure 2a). Yan et al. (2007) show that the cosine of the angle between the vectors of any two traits approximates the correlation coefficient between them. Figure 2 shows that PH, NG, BM, GY, PY, P%, HGW, OY and CH traits were extremely and positively associated. These traits were also negatively associated with oil percent and the increase of these traits caused the decrease of oil percent. Vollmann et al. (2011) reported that nodulating of soybean by B. japonicum increased leaf dimensions, time to maturity, stem height, numbers of pods and weight of 1000 grains. Grain protein content was severely amplified in soybeans with nodules in comparison with non-nodulated lines, whereas, on the other hand, non-nodulated plants had greater oil percent than nodulated plants. Zerihun et al. (2015) have noted that when soybeans planted in soils with neutral pH, rhizobium inoculation and application of 50 kg ha -1 of diammonium phosphate treatment combination are recommended. Piccinin et al. (2011) also stated that the application of a half dose of N for plants inoculated with Azospirillum brasilense increased the performance and yield of wheat.

Treatment combination evaluation based on individual traits
The GGE biplot methodology ranks treatment combinations based on a single trait. Based on seed yield (Figure 2b), B3f1 is the best treatment combination, followed by B2f1, B2f2, B3f2, B2f0 and B3f0. All these treatment combinations had an above-average yield. The other treatment combination had a below-average yield. For plants from non-inoculated seeds, an increase in the chemical fertilizer application enhanced grain yield. When seeds were inoculated by Barvar-2, biofertilizer grain yield was like that of non-inoculated seeds (Figure 2b). Also, at all levels of chemical fertilizers, inoculated plants with Barvar-2 were not superior to non-inoculated ones (data not shown). Seeds inoculated by Biosoy and Biosoy + Barvar-2 produced the highest grain yield at the 33% chemical fertilizer application, and a further rise in fertilizer led to a reduction in grain yield, so that the lowest grain yield was achieved at 200 kg ha -1 diammonium phosphate + 150 kg ha -1 urea (100%) ( Table 2). Some researchers like Argaw (2012), Wasule et al. (2007) and Son et al. (2006) obviously discovered that dual inoculation of seeds by B. japonicum and phosphate solubilizing bacteria expressively had higher growth and yield components than the sole application of both types of bacteria. Similar results stated by Qureshi et al. (2009) showed that seed inoculation with M. ciceri or A. chroococcum significantly enhanced plant biomass and grain yield, but the response was more pronounced with co-inoculation. Tyagi et al. (2003) found that among the various bio-fertilizer treatments, co-inoculation of Rhizobium + PSB produced the highest grain yield even greater than Rhizobium or PSB treatment. An increase in nodulation, grain yield, biological yield and N fixation due to co-inoculation with B. japonicum and PSB was reported by Son et al. (2006) and Abd-alla and Omar (2001). The same result was stated by Wang et al. (2019).
Based on protein content (Figure 2c), the best treatment combinations were B3f1 and B2f2, followed by B2f1, B2f0, B3f0, B3f2, etc. For plants from noninoculated seeds, an increase in chemical fertilizer application enhanced protein content. Ham et al. (1975), Mamatha et al. (2018) and Tahir et al. (2009) observed that protein content was significantly enlarged by nitrogen application. An increase in seed protein due to phosphorus fertilizer application on soybean (Gaydou and Arrivets, 1983) and beans (Zafar et al. 2011) was reported. Seguin and Zheng (2006) observed some increase in protein percent by phosphorus fertilizer. Grain protein content of the inoculated plants with Biosoy (mean = 113.5 g.m 2 ) and Biosoy + Barvar-2 (mean = 119.3 g.m 2 ) was significantly different from the non-inoculated (mean = 65.1 g.m 2 ) and inoculated plants with Barvar-2 (mean = 55.8 g.m 2 ) (data not shown). It seems that inoculated plants with N-fixing bacteria due to higher content of nitrogen produced greater protein content than non-inoculated ones.
However, at all chemical fertilizer levels, there were no differences between inoculated plants with Barvar-2 and non-inoculated plants in terms of seed protein content. Also, the best treatment combination for oil percent was B1f0, followed by B1f1, B0f0,B1f2,B0f1,B0f2,B0f3,etc. (Figure 2d). The results indicated that the high-yielding treatment combinations are the poorest in oil percent. The largest grain oil percent was related to plants inoculated with Barvar-2. A statistically significant difference in seed oil percent between the inoculated plant with Biosoy (mean = 16.4%) and Biosoy + Barvar-2 (mean = 16.2%) with non-inoculated (mean = 18.0%) and inoculated plants with Barvar-2 (mean = 18.1%) was found (data not shown), which is probably because of the adverse effect of nitrogen on Evaluation of integrated N and P management using the TT biplot method in soybean 31 the oil synthesis in soybean. No significant difference between different chemical fertilizer levels was found in seed oil percent. Generally, increasing the chemical fertilizer application enhanced the chlorophyll index for non-inoculated seeds (Figure 2e). This enhancement was insignificant among 0%, 33%, and 66% fertilizer applications, but 100% chemical fertilizer significantly had higher chlorophyll index than 0%. The inoculation of seeds with Biosoy resulted in the highest chlorophyll index when plants received only 33% chemical fertilizer (data not shown). Based on the chlorophyll index (Figure 2e), the best treatment combinations were B3f1 and B2f1, followed by B2f2, B3f2, B2f0, B3f0, etc. Dual inoculation of Biosoy and Barvar-2 increased chlorophyll indices at low rates of the chemical fertilizer, but at a high rate (100%), it was significantly reduced. A single inoculation of Biosoy and dual inoculation with Biosoy and Barvar-2 produced a higher chlorophyll level than no inoculation and Barvar-2 bio-fertilizer at 0, 33, and 66% chemical fertilizer because of a reduction of bio-fertilizer effectiveness. A similar result was reported by Zarei et al. (2011) for soybean.

Conclusion
The current investigation designed to examine the influence of bio-fertilizer treatments on the performance of soybean under different doses of chemical fertilizer. Generally, the highest seed yield was recorded for dual inoculation of Biosoy and Barvar-2 and the application of 33% chemical fertilizer. This could be due to higher nitrogen and phosphorus availability for inoculated plants dually by Biosoy and Barvar-2. However, these bio-fertilizers cannot provide all N and P nutrients needed for plants. When seeds were inoculated with Biosoy and Biosoy + Barvar-2, plants showed the highest performance at 33 percent of the recommended dose of fertilizers. Therefore, nitrogen-fixing bacteria had the potential to reduce chemical fertilizers and increase yield.