Liming, fertilization, and rhizobia inoculation on cowpea yield in a Brazilian Amazon upland forest environment

– The objective of this work was to evaluate the response of the BRS Tumucumaque cowpea ( Vigna unguiculata ) cultivar to liming and to phosphorus and potassium fertilization in the absence of inoculation or nitrogen fertilization, as well as to inoculation with selected Bradyrhizobium strains, in an upland forest area in the state of Amapá, eastern Amazon, Brazil. Seven experiments were conducted in field conditions, in order to evaluate six limestone, four phosphorus, and four potassium rates, besides the inoculation with four Bradyrhizobium strains. Leaf nitrogen concentrations, pod and grain dry mass per plant, grain yield, and soil pH, Al +3 , and Ca +2 + Mg +2 were evaluated. The obtained data were subjected to the analysis of variance, to the regression analysis, or to test to compare means. The soil presented a high buffer power, but liming reduced its chemical limitations. Leaf nitrogen concentrations were correlated to cowpea grain yield. However, there is no effect of the interaction between potassium and phosphorus rates on grain yield. In addition, grain yield is not influenced by liming and potassium fertilization, but is increased by phosphorus fertilization. Inoculation with the UFLA 3-84 and INPA 03-11B strains promotes a greater grain yield than that with BR 3262 and BR 3267, as well as a similar grain yield to that of the control without inoculation.

To increase yield under different edaphoclimatic conditions, improvements have been made, although not yet widely adopted in Brazil, using: efficient rhizobia strains for inoculation (Soares et al., 2006;Leite et al., 2018;Ferreira et al., 2019), cowpea cultivars with superior performance (Bastos et al., 2011;Santana et al., 2019), inputs to correct soil acidity and supply nutrients (Farias et al., 2016;Manzeke et al., 2017), and cropping practices. There is also still a need for scientifically-based technical information for the recommendation of inoculants, liming, and fertilizers for soils from the state of Amapá, which commonly follow the recommendations for soils from other states. These limitations make the differences between the real average yield (obtained by farmers) and potential yield (productivity in experimental conditions) of cowpea genotypes high.
The difference between actual and potential cowpea yield can be reduced by the optimization of an inexpensive nitrogen source: biological nitrogen fixation (BNF). However, although cowpea forms nodules with several rhizobia species, the presence of a high and diverse rhizobia population in the soil hampers the introduction of efficient strains. The rhizobia population is particularly large in regions with a high precipitation (Thies et al., 1995) and in soils with high organic matter contents, such as those in areas that have been forested and with successive cropping with inoculant application (Zilli et al., 2013).
Rhizobia symbiosis may be improved by liming, which reduces soil chemical limitations, causing a systemic effect and resulting in an increase in cowpea root nodulation (Farias et al., 2016). Several authors, for example, have reported an increase in the grain yield of cowpea with phosphorus application (Silva et al., 2010;Melo et al., 2018). In addition, physiological feedback mechanisms in phosphorus, nitrogen, and carbon metabolisms, as well as phenological alterations in the plant, may also affect the contribution of BNF, and practices that raise BNF rates at the start of the crop cycle and/or keep them high for longer periods during the cycle can assure high leaf nitrogen concentrations, increasing BNF contribution to cowpea biomass production and grain yield (Silva Júnior et al., 2018).
The objective of this work was to evaluate the response of the BRS Tumucumaque cowpea cultivar to liming and to phosphorus and potassium fertilization in the absence of inoculation or nitrogen fertilization, as well as to inoculation with selected Bradyrhizobium strains, in an upland forest area in the state of Amapá, eastern Amazon, Brazil.

Materials and Methods
Seven experiments were carried out to evaluate the effects of liming, fertilization, and rhizobia inoculation on soil and cowpea yield, in 2012, 2013, and 2014, in an upland forest environment, in the experimental area of Embrapa Amapá, located in the municipality of Mazagão, in the state of Amapá, Brazil (0º07'19"S, 51º17'57"W, at 7 m altitude). According to the Köppen-Geiger classification, the climate is of the Am type, with an average annual temperature of 27.3ºC and an average annual rainfall of 2,410 mm. The soil in the experimental area is a Latossolo Amarelo, according to the Brazilian soil classification system (Santos et al., 2018), i.e., an Oxisol. The physicochemical properties of the soil are shown in Table 1.
To evaluate the effects of phosphorus and potassium fertilization on the soil and cowpea yield, three field experiments were also carried out in 2012, 2013, and 2014, in an upland forest environment soil (sample 2) in an area that had previously left fallow (Table 1). In 2012, at 60 DBS, the soil received 2,700 kg ha -1 limestone to increase base saturation. The treatments consisted of: four rates of P 2 O 5 (0, 40, 80, and 120 kg ha -1 P 2 O 5 as triple superphosphate) applied at sowing; and four rates of K 2 O (0, 35, 70, and 105 kg ha -1 K 2 O as potassium chloride), 35 kg ha -1 applied at sowing and 70 and 105 kg ha -1 as topdressing 25 DAS. Each plot received 25 kg ha -1 FTE BR12 micronutrient fertilizer at sowing. The experiment had a 4×4 factorial arrangement in a randomized complete block design, with three blocks. The experimental plots measured 15.75 m 2 , consisting of six rows with 0.5 m spacing between rows and five seeds per linear meter, also of the BRS Tumucumaque cultivar, totaling 100,000 plants per hectare.
To assess the effects of inoculation with the strains recommended by Ministério da Agricultura, Pecuária e Abastecimento for cowpea under the edaphoclimatic conditions of the state of Amapá, a field experiment was carried out in 2013, in a soil of an upland forest environment (sample 3) ( Table 1). The area had been frequently used for cowpea cultivar selection. At 60 DBS, the soil received 2,700 kg ha -1 limestone to increase base saturation. The treatments, carried out in a randomized complete block design with three blocks, consisted of: inoculation with the UFLA 3-84, BR 3267, INPA 03-11B, and BR 3262 Bradyrhizobium strains; and a control without inoculation and without nitrogen application. The used strains were obtained from Embrapa Agrobiologia, and a peat-based inoculant was added at a rate of 250 g per 35 kg seed with sucrose solution (10% w/v) to increase adherence. The experimental plots measured 15 m 2 , composed of six rows with 0.5 m spacing between rows and five seeds per linear meter of the BRS Tumucumaque cultivar. Each plot received 80 kg ha -1 P 2 O 5 as triple superphosphate, 25 kg ha -1 FTE BR 12 at sowing, and 70 kg ha -1 K 2 O as potassium chloride -50% at sowing and 50% as topdressing at 25 DAS.
The following parameters, measured in response to limestone rates, were evaluated: grain yield at 60 DAS; leaf nitrogen concentrations at 30 DAS; and pH, Al +3 , and Ca +2 + Mg +2 from compound samples collected after pod harvesting (ten single samples were used to obtain compound samples from each experimental plot). The parameters assessed in response to phosphorus and potassium rates were grain yield and leaf nitrogen concentrations at 30 DAS, and in response to inoculation, pod and grain dry mass per plant, as well as grain yield. The data were subjected to

Results and Discussion
Limestone application did not significantly affect cowpea grain yield, except in 2014, the third year of cultivation, when there was a slight decline in productivity with increasing rates (Table 2). Therefore, cowpea responds weakly to liming. In two field experiments carried out in the state of Pará in areas with a Latossolo Amarelo (Oxisol), a response to limestone application was only observed during the third and fourth years in the municipality of Terra Alta, with a soil with 120 g kg -1 clay, and from the third to sixth years in the municipality of Tracuateua, with a soil with 90 g kg -1 clay (Cravo et al., 2012). In these studies, maximum grain yield was obtained with the application of low limestone rates of approximately 1,000 kg ha -1 (Cravo et al., 2012). It should be noted that liming has a systemic effect that results in an increase in cowpea root nodulation, as previously mentioned. The BRS Guariba cultivar grown in an Oxisol had a higher yield of 201 kg ha -1 when receiving liming and rhizobia inoculation (1,160 kg ha -1 ), instead of the treatment without liming but with inoculation (959 kg ha -1 ) (Farias et al., 2016).
Limestone application statistically attenuated soil limitations related to parameters pH, Al +3 , and Ca +2 + Mg +2 in the first two years of cultivation (Table 3). However, the observed effects were limited. For example, the application of 5,490 kg ha -1 raised soil pH level from 4.48 to 4.88 (only by 0.40 unit), increased Ca +2 + Mg +2 concentration from 1.45 to 3.13 cmol c dm -3 , and reduced Al +3 from 1.5 to 0.68 (only by 0.82 cmol c dm -3 ). Furthermore, the values observed between rates after the third harvest were no longer statistically different and were nearly similar to those measured at the start of the experiment. The obtained results are indicative of a high buffering power of the soil, characterized by a resistance to chemical alterations with limestone application, minimizing the residual effect.
No significant effects were related to potassium rates or to the interaction of phosphorus and potassium on grain yield; however, there was a significant effect of phosphorus rates alone during the three experimental years (Table 4). According to Oliveira et al. (2009), cowpea responded to the application of potassium rates in a Regolithic Neosol with loamy texture, resulting in a grain yield of 1,890 kg ha -1 with a rate of 170 kg ha -1 K 2 O. In the present study, grain yield varied from 373 to 984 kg ha -1 , and the highest increments were obtained between the rates of 0-40 and 40-80 kg ha -1 P 2 O 5 , when compared with that of 80-120 kg ha -1 (Table 4). The grain yield of the BRS Guariba and BRS Aracê cowpea cultivars increased with the rates of phosphorus, up to 120 kg ha -1 , and of zinc, up to 3.1 kg ha -1 , in an Oxisol in the state of Maranhão (Melo et al., 2018). In another Oxisol in the state of Roraima, cowpea yield and leaf phosphorus concentrations increased in response to the applied phosphorus rate until 90 kg ha -1 P 2 O 5 (Silva et al., 2010). Phosphorus and nitrogen are the nutrients whose shortage most limits grain yield in tropical environments, and the response to phosphorus fertilization of legume species that are dependent on BNF is high given the energy demand of the fixation process.  Limestone rates did not affect nitrogen leaf concentrations (Table 5). However, phosphorus and potassium rates caused a reduction in these (1) R 2 , coefficient of determination; and CV, coefficient of variation. concentrations, with a significant effect in the first and second years, respectively. These results indicate the occurrence of a nitrogen dilution effect (Jarrell & Beverly, 1981) in cowpea as a function of the increased productivity promoted by phosphorus rates. Moreover, since leaf nitrogen concentration at 30 DAS was closely related to cowpea grain yield, leaf nitrogen could be considered as an indicator for the selection of cowpea genotypes responsive to BNF (Alcantara et al., 2014;Burridge et al., 2016). Inoculation with the BR 3262, BR 3267, UFLA 3-84, and INPA 03-11B strains did not affect pod and grain dry mass per plant of the BRS Tumucumaque cowpea cultivar, compared with the control without inoculation (Table 6). Inoculation with UFLA 3-84 and INPA 03-11B promoted a greater grain yield than that with BR 3262 and BR 3267, besides a similar grain yield to that of the control without inoculation or nitrogen application. In two experiments carried out in the pre-Amazon region in the state of Maranhão, inoculation with the BR 3262, INPA 03-11B, and BR 3299 strains promoted a greater shoot dry mass accumulation and cowpea grain yield, in comparison with that of the control without inoculation, and BR 3299 showed the best result (Gualter et al., 2011).

Conclusions
1. Liming reduces soil chemical limitations but does not influence the grain yield of the BRS Tumucumaque cowpea (Vigna unguiculata) cultivar grown in an upland forest environment in the Brazilian eastern Amazon region.
2. Fertilization with phosphorus increases the grain yield of the BRS Tumucumaque cowpea cultivar, but fertilization with potassium has no effect.
3. Inoculation with the UFLA 3-84 and INPA 03-11B Bradyrhizobium strains promotes a greater grain yield for the BRS Tumucumaque cowpea cultivar than inoculation with the BR 3262 and BR 3267 strains, as well as a similar grain yield to that obtained with the control without inoculation or nitrogen application. Table 6. Pod and grain dry biomass and grain yield of the BRS Tumucumaque cowpea (Vigna unguiculata) cultivar grown in a Latossolo Amarelo (Oxisol), in a upland forest area, in 2013, in function of the inoculation with the INPA 03-11B, UFLA 3-84, BR 3262, and BR 3267 Bradyrhizobium strains, in the municipality of Mazagão, in the state of Amapá, Brazil (1) .