A simple method to evaluate the number of bradyrhizobia on soybean seeds and its implication on inoculant quality control

Soybeans [Glycine max L. (Merr.)] are one of the major crops in grain-producing countries worldwide. Particularly in Argentina, production for 2010 has been estimated in about 50.0 million tons, which represents 7 percent less than the record of 54.0 million tons covering a production area of 18.0 million hectares (USDA, 2010). Improved crop productivity has been associated with different practices, such as the incorporation of agronomical, physical, chemical or biological technologies. The use of mineral fertilizers has been subordinated to biological fertilization, principally due to its economical and environmental advantages (Adesemoye and Kloepper, 2009). Inoculation with symbiotic nitrogen-fixing Bradyrhizobiaceae bacteria has become a simple and effective way to significantly improve soybean yield and productivity. Within the genus Bradyrhizobium, B. japonicum is the best example of successful symbiotic fixation under very large scale field conditions. Therefore, B. japonicum has been successfully incorporated as the active principle of soybean inoculants in Argentina, Brazil, Paraguay, USA, Canada and other producing countries worldwide in the last 30 years.

Inoculant improvement has been associated to more effective methods to evaluate the quality of bacterial formulations (inoculants) (Tong and Sadowsky, 1994), as well as those related with the isolation, identification and enumeration of bacteria in carriers, seeds, soil and plants. Thus, the enumeration of B. japonicum could be considered the most essential method used for defining the soybean inoculants quality, even before the evaluation of activity under experimental field conditions.

Currently B. japonicum enumeration in liquid formulation currently includes several methods; however, the most probable-number plant infection method (MPN) proposed by Vincent (1970), and the direct colony-forming unit (CFU) counting on agar plates proposed by Somasegaran and Hoben (1994) are still the most frequently used methods in several laboratories worldwide.

Several discrepancies between viable counts and the field efficacy of the inoculants have been reviewed by Catroux et al. (2001). Results as cfu.ml^-1 counting only represent the theoretical maximum provided by the inoculants under field conditions. Some sub-lethally damaged bacteria may survive under optimal growth conditions (e.g. culture medium), but not on seed or soil. On the other hand, the agronomical behavior of soybean inoculants under field conditions show a solid tendency in different reports that lead to assume that "when the number of viable rhizobia by seed increases, nodulation and yield are improved", as proposed by Brockwell et al. (1988). According to Weaver and Frederick (1972) more than 10⁵ B. japonicum cells provided per seed are required to obtain the maximum number of nodules on the tap root. Similarly, Obaton and Rollier (1970) reported an increase in soybean yield in field conditions when the inoculation rate increased to 10⁶ bacteria inoculated per seed. In other experiments, Hume and Blair (1992) reported that the effect on soybean yield could vary by the dilution of the same inoculant or the inoculation with different number of B. japonicum cells.

Therefore, both the cell number and the bacterial physiological state could be responsible for improving inoculant functionality and soybean productivity in agronomic conditions. Despite this fact there are no published reports that correlate a mathematical function with B. japonicum death from inoculated soybean seeds. For that reason, it is mandatory to find a more trustworthy reproducible method, than direct cell counting and simpler than the MPN plant assay, which would allow a more consistent way to evaluate bradyrhizobia cell numbers after inoculation. In this direction, the number of viable bradyrhizobia provided per seed counted on selective media agar plates could be considered the best method. However, its use may be impaired because soybean seeds under actual field growing conditions are obviously non-sterile and the bacterial or fungal seed microbiota normally interferes with the counting, which implies that bradyrhizobia cell enumeration is normally more difficult starting from non-sterile seeds than from inoculants during quality control procedures (Hiltbold et al., 1980).

Our hypothesis proposes that Bradyrhizobium cells applied on soybean seeds during inoculation, die according to a mathematical function, and this phenomenon plus the selection of specific culture conditions could be useful to establish a trustworthy counting method for bradyrhizobia from seeds, and to design a new protocol for inoculant quality control. The aims of this work were: (1) to improve a selective media for the enumeration of B. japonicum obtained from inoculated soybean seeds; (2) to establish the most representative mathematical function for B. japonicum mortality on soybean seeds after inoculation; (3) to evaluate if some environmental or physiological conditions may modify B. japonicum mortality or survival capacity on soybean seeds; and (4) to propose a reliable method for quality control evaluation of soybean inoculants.

The inoculation of plants with beneficial bacteria can be traced back for centuries (Bashan, 1998). By the end of the 19th century, the practice of mixing "naturally inoculated" soil with seeds became a recommended method of legume inoculation in the USA (Smith, 1992). A decade later, the first patent for plant inoculation with Rhizobium sp was registered. (Nobbe and Hiltner, 1896). The practice of soybean inoculation with Bradyrhizobium sp. became common and economically recommended in many producer countries. In Argentina and many South American countries, soybeans [Glycine max (Merr.) L.] are commonly not fertilized but only inoculated with nitrogen. In 2010, Argentina, Bolivia, Paraguay and Uruguay produced more than 20 million hectares of soybeans, almost 16 million of which (more than 80%) were inoculated with products generated by more than 100 companies with common market.

Bacterial counts on non-selective media are a routine quality control procedure for defining a basic threshold in inoculant quality control. Those inoculants that do not fulfill the requested bacterial numbers are discarded in compliance with different regulations (country dependent). These counts are easily performed, even by small, non-specialized microbiology laboratories when inoculants only contain the desired rhizobial population, that is, when inoculants are formulated in a sterile carrier. However, most inoculants are finally applied on non-sterile material like seeds or even directly on the soil. Once there, rhizobial enumeration is a non-reproducible task, due to the presence of Gram-positive bacteria and fungi in variable numbers that may interfere with their direct growth, as in the case of fast growers, or by the synthesis and secretion of toxic compounds, which make comparisons among different formulations for on-seed survival very difficult to achieve, thus delaying better formulation developments.

Some selective agents have been previously used for selective enumeration of Rhizobium and Bradyrhizobium sp. from soils and non-sterile inoculants. Selective agents included antibiotics, heavy metals, dyes and metabolic inhibitors (Danso et al, 1973; Graham, 1969; Thompson, 1967; Van Schreven, 1970). Pattison and Skinner (1974) reported a formulation that contained pentachloronitrobenzene (PCNB), brilliant green (BG), sodium azide, crystal violet, and penicillin. Gómez et al. (1995) proposed two selective media for the enumeration of B. japonicum from soybean inoculants (at that time mostly using non-sterile peat as carrier) containing tetracycline, rifampicin and chloramphenicol to control bacteria and cycloheximide and pimafucin to control fungal contaminants. In our study, most fungal contamination could be prevented by using previously reported PCNB concentrations on the YEM base medium (P-YEM). However, the presence of large mucoid colonies of some gram-positive bacilli after one week incubation prevented in most cases the differentiation and enumeration of typical rhizobium colonies. Our use of the combined PV-YEM medium allowed a significant improvement in the efficiency of on-seed rhizobial enumeration as it effectively prevented the growth of Gram-positive bacteria and fungi (Figure 1).

As already stated, one of the major problems following soybean inoculation is fast bacterial death (Salema et al., 1982; Streeter, 2003). The seed storage temperature after inoculation is empirically considered the most important parameter related to rhizobial survival after seed treatment (Vriezen et al., 2005, 2006; Kremer and Peterson 1983). Moreover, temperature also directly affects the inoculated population desiccation rate generating a second stress condition (Streeter 2003 and 2007). Salinity and seed coat toxicity stress (Deaker et al., 2004) have a strong influence on bacterial survival; however, temperature and the consequent desiccation rate have been the major predicted cause of bacteria injury on seeds. Nevertheless, only few studies have involved the actual analysis of seed survival, perhaps due to contamination problems (Vincent et al., 1962; Bushby and Marshall, 1977; Jansen van Rensburg and Strijdom, 1980; Mary et al., 1986; Cliquet and Catroux, 2001, Kosanke et al., 1992; Vriezen et al., 2006). According to our results, seed storage after inoculation should be recommended at a temperature below 30°C and preferably below 20°C. We propose a survival factor percentage (SFP) defined as a single time parameter resulting from the number of bacteria recovered from seeds after the mortality function reached the plateau for decay, which in some way means when the survival bacterial population stabilized on seeds after inoculation. In our experiments, SFPs were defined 4 hours after inoculation. However, for research and development purposes, we suggest standardizing SFPs at 30°C, as these may reduce the time required for product (or product variation) analysis.

Any plant growth promoting bacteria (PGPB), including rhizobial survival on the seed surface, is usually lower than on solid carriers such as peat, vermiculite or alginate (Bashan et al., 2002) due to the lack of protection against desiccation, and/or toxic compounds present on the seed coat (Burton 1972, 1968). Even if an increase of provided B. japonicum cells improves the nodulation process, plant nitrogen assimilation and grain yield in laboratory conditions (Papakosta, 1992), bacterial physiological state and its resistance to environmental stress may also be critical for its survival on seed and field conditions (Streeter 2007). Roughley et al. (1993) reported a high mortality of Bradyrhizobium sp. after the inoculation of lupine seeds, decreasing by a factor of 10 after one hour, of one hundred after four hours and more than one thousand after one day on the seeds. Similar results were presented by Catroux et al. (1996) in B. japonicum inoculated soybean seeds. The microorganism population recovery from inoculated soybean seeds (as a whole) may be extremely heterogeneous when expressed as cfu.seed^-1, depending on the two components, inoculants and seeds. In the case of inoculants, the major variation sources are the formulation, the presence of some osmoprotectant (e.g. sucrose or carboximethylcellulose), the number of viable microorganisms per volume unit and the recommended doses. In the case of seeds, normally the genotype, the size and the total surface of each grain, storage conditions, a_w, integrity of teguments and mechanical damage can be easily mentioned. Due to all these complex variables, comparison of inoculation treatments (that is, a stated seed A with a stated inoculant A vs another seed B with an inoculant B combination) is frequently misleading.

Our use of relative values like RFPs or SFPs helps dissect the simplest questions, keeping others under the control state, and may even help compare experiments separated in time (Table 3, Figure 2).

The primary benefit of bacterial biological nitrogen (N₂) fixation in grain and forage legumes is the enhancement of yield potentials decreasing the use of exogenous fertilizers. This ability had resulted in commercial legume inoculant production facilities worldwide (Brockwell and Bottomley, 1995). Unfortunately, the widespread availability of legume and other related inoculants had not resulted efficiently equal at level of development of quality control standards protocols. Many countries, including Australia, the Netherlands, Rwanda, Thailand and Russia (Smith, 1992; Marufu et al., 1995) have legislated requirements for minimum populations of target organisms and contaminants per weight unit of inoculants. Canada and France have legislated on product quality by defining the number of viable Rhizobium/Bradyrhizobium required (provided) per seed, or a rate of application, for a full range of crops. The United States and United Kingdom leave the issue on product quality and rates of application at the manufacturer's discretion (Smith, 1992). In Argentina, Brazil and other South American countries actual regulations are insufficient to put pressure on local companies to improve the quality of their products. The lack of internationally accepted regulations for legume inoculant quality and usage parameters has led, on many occasions to inadequate inoculant performance and to the subsequent abandonment of their use. To reverse this trend, a complementary method for soybean inoculant quality control was herein proposed. We improved a selective media for the enumeration of B. japonicum from inoculated soybean seeds, established the most representative mathematical function for B. japonicum mortality on soybean seeds after inoculation and generated Recovery and Survival Factor Percentages as novel tools for inoculant quality control.