Study area

This research was conducted in lots 3 and 4 of the experimental field at the AUGURA Tulenapa Research Center (Carepa-Antioquia), located at 7°46'46" N and 76°40'20" W, with an average altitude of 20 m a.s.l., an average temperature of 27°C, a solar brightness of 1,800 h/year, 87% relative humidity, and rainfall of 844.9 mm distributed in the first production cycle and 2.088 mm in the second cycle.

The soils included fine Fluventic Eutrudepts, fine loam over clay Fluvaquentic Eutrudepts and fine loam Vertic Endoaquepts. These soils were developed from deep, poorly drained, and poorly filtrated alluvial sediments with a moderately slow hydraulic conductivity, high cation exchange capacity (CEC) and low carbon content (^{Gutiérrez, 2007}). Table 1 shows the data of the chemical analysis of the soils used in this study.

TABLE 1 Chemical analysis of the soil. 

CEC - cation exchange capacity.

Plant material

The crop belonged to the AAA group, giant Cavendish subgroup, Williams clone, sixth generation, with a planting density of 1600 plants ha^-1.

Experimental management

For the development of the experiment, 20 experimental units were selected, corresponding to sections called "botalon" with 1563 m² (250 plants). Fertilization cycles began with the selection of the ratoon plant (September), which coincided with the periods of maximum flowering of mother plants. The experimental lots had the same agronomic management as the other productive units, except for fertilization through the application of the treatments.

Samplings were carried out in the fertilizer application area at 30 cm on the pseudostem, at depths between 5 and 15 cm and in the nutrient absorption zone at a space between 5 cm and 15 cm from the region of the soil bordering the corm that had the highest proportion of roots (^{López and Espinosa, 1998}). Samples were collected at different phenological stages (vegetative phase, floral differentiation, flowering, fruit filling, and harvesting). Fertilization was carried out every three weeks, and sampling was performed one week before the fertilization cycles, 17 applications per year, spaced out at approximately three weeks between applications. The commercial fertilizer source used was urea (46%).

Evaluated variables

Total nitrogen and carbon were determined with the methodology of Dumas (^{Batjes, 1996}), with dry combustion and 0.2 g of soil. Measurements were taken with a Tru Spec elemental analyzer (Laboratory Equipment Company LECO, Otario, Canada) calibrated with standardized samples in the Soil analysis laboratory at the Universidad Nacional de Colombia. Productivity was measured by subtracting the weight of the rachis before taking the samples from the fresh weight of the cluster. This value was presented in kg per plant and transformed to ha^-1.

Experimental design and statistical analysis

A randomized complete block design was used with five treatments and four replicates. The treatments were adjusted from soil analyses performed before the start of the research and from the balanced dose proposed by CENI-BANANO (321.8 kg ha^-1 of N, 87.1 kg ha^-1 of P₂O₅, 678.8 kg ha^-1 of K₂O, 50.5 kg ha^-1 of CaO, 117.5 kg ha^-1 of MgO, 64.2 kg ha^-1 of S, 1.4 of kg ha^-1 of B, and 9.3 kg ha^-1 of Zn). Treatments included an absolute control, a control without nitrogen, and doses of 161, 321 and 483 kg ha^-1 of N.

Multivariate analyses of variance that met the criteria of homogeneity and normality of matrices were used, and the interactions between the study areas (application and fertilization) and treatments were evaluated by applying the Tukey test. The analyses were complemented with a correlation analysis to determine the best functional relationship between the production and C/N ratio (IBM SPSS Program).

Total soil carbon and nitrogen content

The total nitrogen content of the soil showed significant statistical differences in the interaction zone of the study between the treatments. The treatment with 483 kg ha^-1 of N had the highest contents in the fertilization zone with an average of 0.33%. However, there were no significant differences with the other treatments (Tab. 2). In the absorption zone, the treatments did not show significant differences; the total nitrogen contents oscillating in the soil were between 0.25% and 0.27% (Tab. 2).

TABLE 2 Contents of total nitrogen (%) in soil of the study area and treatment interaction in two production cycles. 

Values with the same letter within the column do not statistically differ according to the Tukey test (P<0.05). ns: not significant.

Nitrogen fertilizers can be lost through volatilization (44.1°%), leaching (14.8%) and denitrification (4.4%). Fertilizers, along with the decomposition of roots, leaves or other contributions of organic matter that favor nitrogen fixation by microorganisms cause significant increases in the total nitrogen content of soil (^{Ju et al., 2009}) that was (vermiculite), which corroborates the proposal by Fuertes not observed in the present study. and González (2007).

A significant result for the fertilization zone was observed with the treatment of 483 kg ha^-1 of N, which had an increase of 27% in the N content, as compared to the control without nitrogen, in which the total nitrogen was 0.24%. This increase was due to the application of urea, whose dissolution allows nitrogen to take different routes. This could have possibly favored the incorporation of ammonium into the soil, along with other action pathways, such as being absorbed by the plants or by microorganisms, oxidized to a nitrate form, leached or denitrified (^{Gasser et al, 1967}) (Tab. 2).

The NH₄+ from urea can easily adhere to the cation exchange sites in inorganic and organic colloids in the soil, and its proportion depends on the soil's edaphogenetic characteristics (^{Fuentes and González, 2007}). Specifically, the treatment with 483 kg ha^-1 of N showed a tendency to increase the total nitrogen in the fertilization zone and establish significant statistical differences in the second production cycle. These effects were attributed to the precipitation conditions (near the normal ranges for the period) and to the type of clay present in the studied soils

The response to the nitrogen doses may have been related to the ammonium released by the urea in the soil that can be directly associated with the interaction between the clay present in the studied soil (vermiculite, illite, and montmorillonite). This interaction causes ammonium fixation through expandable clays and loss through runoff, an effect attributed to increases in the total nitrogen for the high doses (483 kg ha^-1 of N), as observed in the second production cycle with increases in precipitation, compared to the first cycle.

Total soil carbon showed significant differences between the treatments for the fertilization zone, where the treatment of 483 kg ha^-1 of N had the higher values (2.47 and 2.45%). No significant differences were found in the absorption zone although the observed values tended to be lower than those of the fertilization zone; however, the treatment with 483 kg ha^-1 showed the best response (Tab. 3).

TABLE 3 Contents of total carbon (%) in soil of the study area and treatment interaction in two production cycles. 

Values with the same letter within the column do not statistically differ according to the Tukey test (P<0.05). ns: not significant.

Increase in total carbon was associated with the residual acid effect generated by urea that could have acted on the calcium carbonate applied to the crop and the possible release of mineral carbon present in the parental materials (carbonates) of the soil (^{Guerrero, 2004}; ^{Zapata, 2004}).

Urea reacts with moisture, CO(NH₂)₂ (urea) + H⁺ + 2H₂O; this reaction releases 2 NH₄+ (ammonium) and HCO₃ ^- (bicarbonate) (^{Guerrero, 2004}). The latter was able to react with elements of the soil since the morphogenesis of the soils where the research was developed indicated that there was a high load of suspended sediments, sand, and some silt with calcium and magnesium contents. The calcium had a saturation over 70%, allowing carbonate and calcium to form calcium bicarbonate that affected the increases in the total carbon of the soil causing an increase in the total carbon in the fertilization zone. ^{Liu et al. (2008)} evaluated four treatments with fertilization and found that an increase in the inorganic carbon of the soil resulted from fertilization because of carbonate stability. Therefore, further studies in which organic carbon is dissociated from inorganic components are needed.

Carbon-nitrogen ratio (C/N ratio)

The C/N ratio displayed an interaction between the treatments and study areas. The treatments resulted in an increase in the C/N ratio values for the fertilization zone. The control and 0 kg ha^-1 of N showed values in the fertilization area similar to the dose with high N. These results are associated with the nitrogen and carbon present in the soil that was released from organic matter. This can also be influenced by the recycling process between a mother plant and its successors due to nitrogen-free fertilization (^{Robinson and Galán 2012}). In the absorption zone, the behavior was similar, but with small variations between the treatments and values lower than in the fertilization zone (Tab. 4).

The highest carbon contents were observed for the fertilization zone with the treatment with 483 kg ha^-1 of N. These values are the product of the acidifying effect of urea accompanied by conditions of high humidity in the second production cycle. This condition helped the movement of these elements to the absorption zone and impacted the C/N ratio (^{Jones and Jacobsen, 2005}). For this reason, the values seen in the present study, all less than or equal to 8, were below the levels considered normal or ideal for agricultural land (^{Moreno and Moral, 2008}; ^{Osorio, 2014}) (Tab. 4).

TABLE 4 Carbon-nitrogen ratio in soil for the treatment and study area interaction. 

Values with the same letter within the column do not statistically differ according to the Tukey test (P<0.05). ns: not significant.

^{Torres et al. (2017)} stated that, for the doses 161 and 321.8 kg ha^-1 of N, there was low agronomic efficiency, corroborating the low response in the nitrogen application as reflected in the C/N ratio with values oscillating between 6 and 6.4 for the absorption zone (Tab. 3). ^{Moreno and Moral (2008)} confirmed that low C/N ratios can be associated with an excess of nitrogen that is lost, a situation that can be associated with the low agronomic efficiency observed by ^{Torres et al. (2017)}.

The C/N ratio values during the stages of development varied between 6.0 and 8.2, indicating a high nitrogen content compared to the carbon in the soil, but the values can also be interpreted as a rapid release of the mineral and organic carbon (^{Osorio, 2014}) that occurred in the present study.

The variation of total carbon contents during the development of the banana crop tended to decrease; this can be attributed to the soil and climatic conditions and the agronomic management that implied a gradual loss of soil quality (^{Rosales et al., 2006}). It is necessary to establish a management plan for organic matter that maintains and improves carbon content and, therefore, nitrogen in order to proportionally reduce nitrogen doses for the benefit of producers and the environment (^{Torres et al., 2017}).

Carbon-nitrogen ratio and production

Production during the two cultivation cycles showed positive responses to the treatments of 321.8 and 483 kg ha^-1 of N, with values of 41.3 and 42.0 t ha^-1 for the first cycle and 40.9 and 46.4 t ha^-1 for the second cycle (Fig. 1), values within the production range reported by ^{Robinson and Galán (2012)} for cv. Williams (24 to 64 t ha^-1). The effect of low production of absolute control in cycle two was associated with the loss of the nutritional recycling effect of the mother-ratoon relationship (^{Robinson and Galàn, 2012}).

FIGURE 1 Production in t ha¹ in the treatments and productive cycle. 

Responses in production to high doses of nitrogen were associated with low levels of total carbon (<2%) in the study areas (^{Gutiérrez, 2007}). It should be remembered that no applications of organic matter were performed on any treatment in the production cycles that accentuated the effect of the nitrogen doses, an effect similar to that found by ^{Srikul and Turner (1995)}, ^{Orozco and Pérez (2006)} and ^{Nyomby et al. (2010)}.

To evaluate the association between the production variable and the C/N ratio, a correlation analysis was performed that obtained coefficients of 0.44 and 0.42 in the absorption and fertilization zones, respectively. These coefficients were positive, implying that, as the C/N ratio increased, production increased.

The regression analysis for the study areas between the production variable and the C/N ratio was significant (R² = 0.2). This led to the inference that, although the C/N ratio values were low, the best response in production was associated with the high C/N ratio. The low correlation coefficient that can be attributed to production did not depend on the C/N ratio alone.

The correlation analysis for the study areas during the two production cycles did not show significant results between the production and C/N ratio values. However, for the first production cycle, the best fit was seen for the treatment with 321 kg ha^-1 of N (r = 0.95). In the second cycle, the treatment with 483 kg ha¹ of N (r = 0.86) had the best fit, leading to the inference that, even though the C/N ratio values were low, the treatments with the best response in production were associated with the higher C/N ratios. ^{Robinson and Galán (2012)} confirmed that the behavior of a cultivar is associated with its genetic characteristics, agronomic management, and climatic and soil characteristics (Tab. 5).

TABLE 5 Correlation analysis of the carbon-nitrogen (C/N) ratio in soil and production by treatment. 

** Correlation is significant at the 0.01 level (2-tailed).

* Correlation is significant at the 0.05 level (2-tailed).

When analyzing the response by treatment, the control in the correlation between production and C/N ratio showed correlation coefficients of -0.49 and -0.37 for the absorption and fertilization zones, respectively (Tab. 5). This negative value means that, as production increased, the C/N ratio decreased; this could be attributed to the fact that the nitrogen extracted by the plants for production was available in the soil and carbon decreased before the application of organic matter. ^{Torres et al. (2017)} confirm that the low productivity of the absolute control during the second production cycle is associated with the loss of the effect of nutrient recycling from the mother and ratoon relationship and the impoverishment of the soil without fertilizer, as corroborated by ^{Robinson and Galán (2012)}.

The correlation values in the nutrient absorption zone for the two cycles reached coefficients higher than 0.7, reflecting the influence of nitrogen fertilization. ^{Nave et al. (2009)} stated that nitrogen inputs help meet the requirements of crops, improving productivity, and increasing carbon storage in the soil, as observed in the treatment with 483 kg ha^-1 of N with the best response in production and its interaction with the C/N ratio (Tab. 5).

For cycle two, the treatment with 483 kg ha^-1 of N obtained a correlation of 0.86 between production and the C/N ratio and 0.75 for the absorption zone. By taking into account the low correlation coefficient from one area to another, it could be inferred that the plants absorbed the nitrogen and the fertilizer was not immobilized; it passed from the fertilization zone to the absorption zone (Tab. 5) and was not lost through leaching or runoff processes. It is worth mentioning that for the second cycle the treatment with 321.8 kg ha^-1 of N showed a significant adjustment in the absorption zone for the production correlation and C/N ratio (r = 0.95). This demonstrates that the contributions of nitrogen through fertilizer had an impact on the production and, in turn, on the C/N ratio.

The high doses of nitrogen in the form of urea had an important impact on the carbon and nitrogen contents despite the possible losses of the latter through leaching and volatilization (^{Mengel and Kirkby, 2000}).

^{Torres et al. (2017)} concluded that production can be increased by improving the efficiency of fertilization that includes organic matter. This requires the development and promotion of technological models that take into account the observations of ^{Robinson and Galán (2012)} for interactions among soil, plants, climate, and agronomic management.