Biofertilizers are becoming increasingly popular in many countries and for many crops. They are defined as products containing active or latent strains of soil microorganisms, either bacteria alone or in combination with algae or fungi that increase the plant availability and uptake of mineral nutrients [1]. In general, they contain free-living organisms associated with root surfaces but they may also include endophytes, microorganisms that are able to colonize the intercellular or even intracellular spaces of plant tissues without causing apparent damage to the host plant. The concept of biofertilizers was developed based on the observation that these microorganisms can have a beneficial effect on plant and crop growth (e.g., [2]). Consequently, a range of plant growth-promoting rhizobacteria (PGPR) has been identified and well characterized. Direct beneficial effects can occur when the microorganisms provide the plants with useful products. The best known case of this are microorganisms that can directly obtain N from the atmosphere and convert this into organic forms usable by plants. Such biological nitrogen fixers (BNF) include members of the genus Rhizobium, Azospirillum, and blue-green algae. Rhizobia are symbiotically associated with legumes and nitrogen fixation occurs within root or stem nodules where the bacterium resides [3]. The genus Azospirillum also has several N-fixing species, which are rhizobacteria associated with monocots and dicots such as grasses, wheat, maize and Brassica chinensis L. [4,5]. Azospirillum strains have been isolated from rice repeatedly, and recently the strain Azospirillum sp. B510 has been sequenced [6,7]. Considerable N fixation by Azotobacter spp. and Azospirillum spp. in the rice crop rhizosphere was reported repeatedly [6,8], but others [9] questioned such high amounts of non-symbiotic N fixation in agriculture. Instead, it was hypothesized that the beneficial effect of Azospirillum inoculums may not derive from its N-fixing properties but from its stimulating effect on root development [2], probably often triggered by phytohormones [10]. This view was confirmed by [11], who concluded that the main effect of Azospirillum spp. is the stimulation of the density and length of root hairs, the rate of appearance of lateral roots, and the root surface area. Phytohormone production and a beneficial effect on plant growth were also shown for a range of other microorganisms [12,13].

Another important genus for biofertilizer producers is Trichoderma, a fungus present in nearly all soils. Trichoderma spp. thrive in the rhizosphere and can also attack and parasitize other fungi. Trichoderma spp. have been known for decades to increase plant growth and crop yield [14,15,16], to improve crop nutrition and fertilizer uptake [16,17], to speed up plant growth and enhance plant greenness [18], as well as to control numerous plant pathogens [19,20,21]. A part of these effects may also be related to the fact that some Trichoderma spp. seem to hasten the mineralization of organic materials [22], thus probably releasing nutrients from soil organic matter. Positive effects on plant nutrition were also described for other organisms, and many soil bacteria may enhance the mineral uptake of the plant, as for example by the increased solubility of phosphate in the soil solution [23].

There is a wide range of reports on the effect of biofertilizer application in crops grown in non-flooded soils (unlike lowland rice), and the technology for Rhizobium inoculation of leguminose plants is well established. A review on results from Azospirillum inoculation experiments across the world and covering 20 years was conducted by [11]. They found a success rate of 60–70% with statistically significant yield increases on the order of 5–30%. However, the vast majority of these trials were on wheat, maize, sorghum, or millet, and only one of the experiments included in the analysis was on rice. Consequently, results from biofertilizer use in rice are still rare. Some reports from groups promoting the use of biofertilizers indicated considerable yield increases upon their use. Trichoderma harzianum, used as a coating agent for rice seed, was reported to result in a 15–20% yield increase compared with rice plants receiving full inorganic fertilizer rates only [22]. As already mentioned above [8], reported enhanced growth and development of rice and maize after the use of biofertilizer containing Azospirillum spp, and asserted the biofertilizer would provide 30–50% of the crop’s N requirement. Similarly, [6] claimed that the inoculation of rice seedlings with Azotobacter spp. and Azospirillum spp. was able to substitute for the application of inorganic N fertilizer, and that this technology enabled rice yields of 3.9 to 6.4 t·ha⁻¹ (yield increases in comparison with the control were about 2–3 t·ha⁻¹). Another study tested the effect of rice root inoculation with Azospirillum spp. under different N fertility levels, and found a more pronounced yield response at lower levels of inorganic N fertilization [24]. Generally, rice yield increases in this study were lower, and ranged around 0.5 t·ha⁻¹. A yield-increasing effect on rice by inoculation with Azospirillum sp. strain B510 was also shown by [25] but the experiment was conducted in pots only.

Based on these reports, it can be assumed that biofertilizers could offer an opportunity for rice farmers to increase yields, productivity, and resource use efficiency. And, the increasing availability of biofertilizers in many countries and regions and the sometimes aggressive marketing brings ever more farmers into contact with this technology. However, rice farmers get little advice on biofertilizers and their use from research or extension because so little is known on their usefulness in rice. Necessary would be recommendations describing under which conditions biofertilizers are effective, what their effect on the crop is, and how they should best be used. To start addressing these issues, we conducted this study, testing different biofertilizers in an irrigated lowland rice system in the Philippines during four seasons. The objectives of the study were (1) to evaluate the effects of different biofertilizers on irrigated rice grain yield, (2) to investigate possible interactions of the effect of these biofertilizers with different inorganic fertilizer rates, and (3) to determine, based on the results, whether biofertilizers are a possible option to improve the productivity of rice production and under which conditions they give good results.

In all four seasons and across the biofertilizer treatments, grain yield increased with increasing amounts of applied fertilizer (Table 4, Figure 1). However, this increase was not always statistically significant and the yield increase varied considerably between seasons. Overall, the lowest grain yields occurred in the 2009 WS, ranging only from 1.9 to 2.7 t·ha⁻¹. Generally, low yields in that season were due to a typhoon that caused considerable damage through flooding of the experimental field and lodging of the crop. For this reason, the crop was harvested prematurely by about 1 week, which further reduced attainable yields.

Grain yields in the other three experimental seasons were similar and ranged from 4.0 to 5.2 t·ha⁻¹ in the 2009 DS, from 3.4 to 5.1 t·ha⁻¹ in the 2010 WS, and from 3.8 to 5.6 t·ha⁻¹ in the 2010/11 DS. These ranges already indicate a relatively low yield increase due to fertilizer application in the 2008/09 DS (up to 1.2 t·ha⁻¹ for the full fertilizer rate of 120-60-60 kg N-P₂O₅-K₂O ha⁻¹) and the 2009 WS (up to 0.8 t·ha⁻¹ for the full fertilizer rate of 90-30-30 kg N-P₂O₅-K₂O ha⁻¹). A higher response to inorganic fertilizer was achieved in the 2010 WS (up to 1.7 t·ha⁻¹ for the full fertilizer rate of 90-30-30 kg N-P₂O₅-K₂O ha⁻¹) and the 2010/11 DS (up to 1.8 t·ha⁻¹ for the full fertilizer rate of 120-60-60 kg N-P₂O₅-K₂O ha⁻¹).

The effects of biofertilizer treatments on grain yield, depending on the inorganic fertilizer treatment, are shown in Table 4 and Figure 1. Significant yield increases due to biofertilizer use were observed in all experimental seasons with the exception of the 2008/09 DS. In the 2010/11 DS, no significant difference between the three biofertilizers tested was detected, but all three achieved better yields than the Control. The biofertilizer achieving the highest average grain yields across all four inorganic fertilizer treatments and in all four seasons was BN. Statistically significant interactions between biofertilizer treatment and inorganic fertilizer treatment could not be detected in any season (at p ≤ 0.05), suggesting that the effect of the biofertilizer was independent of the inorganic fertilizer rate. However, there was a trend of higher yield increases due to biofertilizer use at low to medium inorganic fertilizer rates (Table 4, Figure 1). This trend was most obvious for the BN biofertilizer whereas the performance of the BS and BG biofertilizers was less consistent.

The grain yield increase due to biofertilizer only (0% RR inorganic fertilizer treatment) usually ranged from 200 to 300 kg grain ha⁻¹ for the best biofertilizers with the exception of the 2010 WS, when the BN treatment had an almost 800 kg·ha⁻¹ better grain yield than the Control. In relative terms (Table 5), the seasonal yield increase across fertilizer treatments was between 5% and 18% for the BN biofertilizer (up to 24% for individual treatment combinations), between 3% and 13% for the BS biofertilizer (up to 24% for individual treatment combinations), and between 1% and 9% for the BG biofertilizer (up to 28% for individual treatment combinations). For the calculation of the relative yield increase, only average values could be compared and no statistical analysis could be conducted.

The effect of biofertilizer on the agronomic efficiency of N fertilizer (AEN) is shown in Table 6. For these calculations, the yield of each treatment was compared with the grain yield baseline (the Control treatment in which no biofertilizer and no inorganic fertilizer were used) and the yield increase was divided by the N rate applied. Again, only average values could be compared and no statistical analysis was possible. The results (Table 6) indicate considerably higher overall AEN values in the 2010 WS and the 2010/11 DS. Also, the AEN values are generally higher at low N rates and decrease with higher N application rates. The biggest AEN increase caused by biofertilizer occurred at the lowest N fertilizer rate (25% RR treatment), and, among the different biofertilizers tested, the BN biofertilizer resulted in the highest and most consistent AENs.

In our experiments, the selected biofertilizers were used as recommended by the producers but we could not check the viability or the contents of the products. Thus, we did not verify whether the biofertilizers contained the declared organisms (Table 6; the contents of BG remained unidentified) or the required number of living cells in the inoculate. The importance of quality control and regulation for biofertilizer production was emphasized by [31], who also pointed out that the frequent absence of such mechanisms can cause non-functional products. Maintenance of high standards for Azospirillum inoculants with proven efficient strains and cell numbers on the order of 1 × 10⁹ to 1 × 10¹⁰ colony-forming units (cfu) g⁻¹ or mL⁻¹ was also requested by [11]. But, the fact that the products in our study caused a significant effect on grain yield in three out of four seasons (only two out of four seasons for BG) indicated that the biofertilizers tested had sufficient active ingredients and that the producers maintained a good quality over the four seasons (or 2.5 years). Theoretically, the effect of the biofertilizers could also have been caused by non-living ingredients but the applied amount was so small that even micronutrients could not explain the observed effects. Also, no micronutrient deficiencies are known from either of the two experimental sites.

The general effect of inorganic fertilizer was as expected, and grain yields increased continuously with increasing fertilizer rates (Table 4). However, the response to inorganic fertilizer was low in the 2008/09 DS and the 2009 WS, as also indicated by the low AEN (Table 5). Good and economic values for AEN are usually 15–20 kg grain yield per kg N applied, and, at AEN < 10, inorganic fertilizer use may give negative economic returns depending on the input and output prices [32,33]. Low response in the 2009 WS can be explained by the negative effects of a typhoon and the early harvest. The low response in the 2008/09 DS could be due to the combination of a very fertile soil (high grain yield in the 0% RR treatment) and a limited yield potential in that season (low maximum yields in the 100% RR treatment).

The tested biofertilizers did increase grain yield significantly, and especially the BN biofertilizer did so consistently. Even in seasons in which no significant effect could be detected due to the yield variability between plots, the grain yield with biofertilizer was usually better than without. The seasonal yield increase across fertilizer treatments was between 5% and 18% for the BN biofertilizer (up to 24% for individual treatments; Table 5), which is within the 5–30% range reported for Azospirillum inoculums and non-rice crops by [4,11]. Similarly, the here-observed yield increase for the Trichoderma-based BS (3–13%) was close to the 15–20% rice yield increase described by [22]. The trend of yield increases between the different inorganic fertilizer treatments was not so clear across seasons but yield increases were often lower at higher inorganic fertilizer rates (Figure 1), which was also reported by [24]. Absolute grain yield increases due to biofertilizer were usually below 0.5 t·ha⁻¹ (Table 1, Figure 1), corresponding to an estimated additional N uptake of less than 7.5 kg N ha⁻¹ (based on 0.5% N in straw, 1.0% N in grain, and harvest index 0.5). Both values are far below grain yield increases and additional N uptake reported by [6] and [8], but similar to the rice grain yield increases reported by [24].

The calculated AEN values (Table 6) suggested higher N use efficiency for treatments with biofertilizer use. Increased nutrient uptake and fertilizer use efficiency were also reported for Trichoderma spp. [16,17,34] and for Azospirillum spp. [11]. But, the results could be explained in several ways. One possibility is that the biofertilizer stimulated root growth and thereby increased the uptake of indigenous N from the soil (the higher AEN would then be only an artifact of the calculation method). Second, the increased root growth could reduce N fertilizer losses, and the third option could be biological N fixation (which could explain the superior performance of the BN biofertilizer, supposedly containing organisms capable of biological N fixation). But, our experiment cannot answer the question of which process or combination of processes is at work here, if that is possible at all under field conditions [9].

The study was conducted to evaluate the effect of different biofertilizers on the grain yield of lowland rice, and investigate possible interaction effects with different inorganic fertilizer amounts. The results showed significant yield increases for all products tested in some seasons but the most consistent results were achieved by the Azospirillum-based biofertilizer. In most cases, the observed grain yield increases were not huge (0.2 to 0.5 t·ha⁻¹) but could provide substantial income gains given the relatively low costs of all biofertilizers tested. The positive effect of the tested biofertilizers was not limited to low rates of inorganic fertilizers and some effect was still observed at grain yields up to 5 t·ha⁻¹. However, the trends in our results seem to indicate that the use of biofertilizers might be most helpful in low- to medium-input systems. The results achieved can already be used to develop better advice for farmers on biofertilizer use in lowland rice, but several important questions remain. In particular, biofertilizers need to be evaluated under conditions with abiotic stresses typical for most low- to medium-input systems (e.g., under drought or low soil fertility) and with a range of germplasm because their effect might depend also on the variety used. More upstream-oriented research would be needed to better understand the actual mechanisms involved, which in turn could also contribute to making the best use of biofertilizers in rice-based systems.