Nutritional balance and recovery rate of macronutrients by ‘Gigante’ cactus pear under different fertilizations

ABSTRACT The objective of this study was to evaluate nutrient export, nutritional balance and macronutrient recovery rate in ‘Gigante cactus pear under different fertilization doses, spacings and configurations during three production cycles. The 72 treatments were arranged in a randomized block design with three repetitions, two configurations in the plots (triple and quadruple row), three spacings in the subplots (3.00 x 1.00 x 0.25 m; 3.00 x 1.00 x 0.16 m and 3.00 x 1.00 x 0.125 m); and 12 types of fertilization in the sub-subplots (with organic, organomineral and mineral sources). Exports of nutrients are higher under fertilizers with higher nutrient supplies, regardless of source. The order of export was K > Ca > N > Mg > S > P for mineral fertilization, organic fertilization with 30 Mg ha-1 year-1 and organomineral fertilization with 30 and 60 Mg ha-1 year-1 plus NPK and PK, and it changed to K > Ca > N > Mg > P > S under organic fertilization with 60 and 90 Mg ha-1 year-1 and organomineral fertilization with 90 Mg ha-1 year-1 plus K. The nutrient balance after three production cycles was positive for N, P and S, negative for Ca and Mg in all fertilizations, and positive for K under mineral fertilization with 600 kg ha-1 of K2O and organomineral fertilization with 30, 60 and 90 Mg ha-1 year-1 plus 600 kg ha-1 of K2O. The highest recovery rates of K and the lowest recovery rates of N occur under organic fertilization.


Introduction
Knowledge about the nutritional requirements of cactus pear is fundamental for the sustainability of its cultivation, since the great extraction and export of nutrients during successive cycles (Silva et al., 2016a;Donato et al., 2017c) can lead to depletion of soil nutrients and result in a reduction of yield in the absence of replacement of the minerals exported by harvest (Dubeux Júnior et al., 2010).
Quantitative changes in nutrient extraction in response to fertilization reveal a high nutrient export potential in cactus pear cladodes (Dubeux Júnior et al., 2010;Silva et al., 2012). Thus, evaluating the accumulation and export of nutrients represents a relevant starting point in the definition of fertilizer doses to be used and ensure the production sustainability of soils.
Fertilization with mineral, organic or organomineral fertilizer sources, in appropriate proportions, optimizes the production of green matter, accumulation of nutrients in the cladodes and dry matter yield of cactus pear. In addition, they can determine differences in the amounts of nutrients extracted and exported by the cladodes, which becomes more important when considering several harvest cycles. Additionally, the balance of nutrient inflows and outflows in the system and the recovery rate by the plant of the applied nutrients may differ according to the sources and/or doses. The objective of this study was to evaluate for three production cycles the export, nutritional balance and apparent nutrient recovery rate by 'Gigante' cactus pear (Opuntia ficus-indica) cultivated with different planting configurations, spacings and fertilizations.

Material and Methods
The experiment was conducted for three production cycles at the Federal Institute of Bahia, Campus of Guanambi, BA, Brazil, located at coordinates of 14º 17' 44" S latitude, 42º 41' 39" W longitude and altitude of 545 m. The climate of the region is semiarid hot and dry and the soil is an Oxisol. In the first production cycle, lasting 650 days between planting and harvesting, conducted between October 2011 and July 2013, precipitation was equal to 912.29 mm, corresponding to two rainy seasons, between October 2011 and February 2012 (442.67 mm) and between October 2012 and April 2013 (469.62 mm). In the second cycle, lasting 330 days between two successive harvests, from August 2013 to July 2014, precipitation was equal to 853.96 mm. In the third cycle, also lasting 330 days, between August 2014 and July 2015, precipitation was equal to 879.60 mm. Data were recorded by an automatic weather station installed in the experimental area.
The preparation of the area, removal of seedlings and planting followed the recommendations for the crop (Silva et al., 2012;Donato et al., 2014). The collection of samples for the analysis of macronutrient contents in plant tissues was performed at the time of harvest. Within the usable plot of 2 m, cladodes from the median position of the plant were collected, in a total of 20 samples per treatment with approximately 25 g of green matter each. For this procedure, a hole saw with 5 cm diameter and 4 cm depth, adapted to a drill, was turned on over the cladodes to remove a representative circular slice, which constituted the sample. After collection, the material was sliced and dried in a forced circulation oven at 60 °C for 72 h. Subsequently, the samples were ground in a Wiley-type mill with a 1.00-mm-mesh sieve, identified, placed in plastic pots and then sent to the Soil Laboratory of the Agricultural Research Company of Minas Gerais (Epamig Norte), for the analytical procedures.
The cladodes were analyzed for the contents of the macronutrients N, P, K, S, Ca and Mg, expressed in g kg -1 , according to Malavolta et al. (1997): N, by sulfuric acid digestion with the Kjeldahl method; P, K, S, Ca and Mg by nitric-perchloric digestion. The dry matter content (DM) was also determined by dividing the weight of the dry sample by the weight of the fresh sample, and the value obtained was multiplied by 10 to be transformed to base g kg -1 of natural matter (NM) (Silva & Queiroz, 2009).
At harvest, in the three production cycles, the mother cladode and three primary cladodes were preserved. To calculate green matter yield (GMY) (Mg ha -1 ), the cladodes of the usable plot were harvested and weighed. Dry matter yield (DMY) (Mg ha -1 cycle -1 ) was calculated according to the dry matter content of the treatment multiplied by GMY. Nutrient export was calculated according to DMY and nutrient content in cladodes and expressed in kg ha -1 cycle -1 and also in kg Mg -1 cycle -1 .
As used by Silva et al. (2016a) and Donato et al. (2017b), the nutritional balance was calculated for the macronutrients (N, P, K, Ca Mg and S) in each cycle, considering the difference between the amounts of nutrients added to the soil in each fertilization treatment and the amounts exported by the collection in the cladodes, without considering the supply of nutrients through the soil. Likewise, the final balance was also calculated after three production cycles.
In addition, in each production cycle, the apparent recovery rate of nutrients by the plant (APRR) was also estimated in a similar way to the procedure performed by Maluf et al. (2015), considering the amounts of nutrients applied by organic, organomineral and mineral fertilizations and the nutrient content in the cladodes. APRR (%) = (C(i) in the treatment with fertilization -C(i) in the treatment without fertilization)×100/ Q(i) applied in the treatment with fertilization), where: C(i) in the treatment with fertilization is the amount of nutrient i (kg ha -1 ) for the treatment with fertilization; C(i) is the amount of nutrient i (kg ha -1 ) in treatment without fertilization (0-000-000-000); Q(i) is the amount of nutrient applied (kg ha -1 ) in the treatment with fertilization; C(i), amount of nutrient i (kg ha -1 ), was calculated by the expression C(i) = CN(i) x DMY, in which: CN (i) is the content of nutrient i in the cladode (kg kg -1 ) and DMY is the dry matter yield of the treatment (Mg ha -1 ).
The data of nutrient exports in the cladodes were subjected to analysis of variance to verify the occurrence of interactions and the means of fertilization were grouped by the Scott-Knott criterion (p ≤ 0.05).

Results and Discussion
The export of nutrients by the cladodes of 'Gigante' cactus pear in the three production cycles varied with fertilization, regardless of planting configurations and spacings. No interactions were recorded between the factors studied.
The export of N by cactus pear cladodes was lower in the absence of fertilization in the three production cycles, followed by organic fertilization with 30 Mg ha -1 year -1 of bovine manure and 60 Mg ha -1 year -1 in cycle II (Table 1). The highest exports of N were obtained with the mineral fertilization 300-300-600 kg ha -1 year -1 of N-P 2 O 5 -K 2 O in cycles I and III, and with fertilization with organic source, except 30 and 60 Mg ha -1 year -1 of manure, mineral or organomineral in cycle II.
Mineral sources, such as the ammonium sulfate used in the present study, for being more soluble, promote higher N contents in cactus pear cladodes (Silva et al., 2012) and, consequently, greater export of nutrients (Silva et al., 2016a). In addition, the increase in N dose promoted an increase in the concentration and availability of this nutrient in the soil and, consequently, greater export of N (Donato et al., 2017b;Lopes et al., 2018).
The export of P and K by cactus pear cladodes was lower in the absence of fertilization in the three production cycles, followed by organic fertilization with 30 Mg ha -1 year -1 of manure in cycle III for P and in cycle I for K. In general, the highest doses of P and K applied to the soil resulted in higher export of this nutrient in the three production cycles.
(Mg ha -1 ) of manure -N (kg ha -1 ) -P 2 O 5 (kg ha -1 ) -K 2 O (kg ha -1 ); Means followed by the same letter in the column belong to the same grouping by the Scott-Knott criterion (p ≤ 0.05) Table 1. Exports of nitrogen (EN), phosphorus (EP), potassium (EK), sulfur (ES), calcium (ECa) and magnesium (EMg) by cladodes of 'Gigante' cactus pear subjected to mineral fertilization, organomineral fertilization and organic fertilization in three production cycles However, in cycle II the export of K was higher under organic or organomineral fertilization with 60 and 90 Mg ha -1 year -1 of manure and, in cycle III, organomineral fertilization with 60 and 90 Mg ha -1 year -1 of manure plus 600 kg ha -1 of K 2 O culminated in higher export of K.
The higher doses of P added by different sources favored greater absorption of this nutrient by the plant, regardless of the source used, and corroborate the results found by Donato et al. (2017b). In relation to K, the highest values of K export were recorded in cycle III and correspond to the highest dry matter yields -DMY (Table 2). The total K added in these fertilizations corresponds respectively to 622.9 and 685.4 kg ha -1 and resulted in DMY of 21.1 and 23.2 Mg ha -1 , with exports of 868.4 and 1,002.6 kg ha -1 of K (Tables 1 and 2). These values of DMY and export are consistent with those found in other studies (Silva et al., 2016a;Donato et al., 2017b) and attest that the export of K varies with DMY, but is influenced to a greater extent by the dose of K added to the soil.
The export of S, Ca and Mg by cactus pear cladodes was lower in the absence of fertilization in the three production cycles, followed by organic fertilization with 30 Mg ha -1 year -1 of manure in cycles II and III for S and in cycles I and III for Ca and Mg.
Mineral fertilization and organomineral fertilization with N and P, due to the higher supply of S because of the use of ammonium sulfate and single superphosphate, and by the greater contribution of Ca due to the use of single superphosphate, a fertilizer that also contains Ca, resulted in higher export of these nutrients and corroborate those of existing research (Silva et al., 2016a;Donato et al., 2017b). The lowest values of Mg exported coincide with the absence of Mg application by mineral fertilization and in the treatment without fertilization and with the lowest dose of Mg in the soil under organic fertilization with 30 Mg ha -1 year -1 of manure, 10 kg ha -1 year -1 of Mg. It is worth pointing out that the synergistic interaction of the applied P with soil Mg and with the Mg present in organic and organomineral fertilizations with 60 and 90 Mg ha -1 year -1 of manure favors the absorption of both (Malavolta et al., 1997) and results in increased Mg exports.
The order of export of macronutrients in 'Gigante' cactus pear cladodes varied with the cycle and fertilizations (Table  2). In the first production cycle, the decreasing order of macronutrient export was K > Ca > N > Mg > S > P under all fertilizations and in the absence of fertilization (Table  1). However, under fertilization with mineral source with N supply, for example 000-300-300-300, 0-300-300-600, and organomineral fertilization with supply of also mineral N 30-150-300-300 and 30-150-300-600, the exported S was approximately twice the P, while under the other fertilizations the exports of S and P were close, with a slightly higher value for S. The use of ammonium sulfate containing S, as a source of inorganic N, justifies these results.
In cycles II and III, the decreasing order of export of macronutrients was Ca > K > N > Mg > S > P in the absence of fertilization, while for fertilizations with mineral source, organic source with application of 30 Mg ha -1 year -1 of manure Table 2. Balance of macronutrients (N, P, K, Ca, Mg and S) as a function of the average export by cladodes of 'Gigante' cactus pear at harvest and the amounts added to the soil by organic, organomineral and mineral fertilizations expressed in (Mg ha -1 ) of manure -N (kg ha -1 ) -P 2 O 5 (kg ha -1 ) -K 2 O (kg ha -1 ) in three production cycles Continues on the next page Continuation of Table 2 DMY -Dry matter yield; 1 The amounts of nutrients added to the soil in the first cycle correspond to the sum of the applications carried out at the time of planting and repeated in the second rainy season, before the first harvest, therefore it represents twice the nutrients supplied to the soil compared to the second and third cycles or organomineral source with 30 Mg ha -1 year -1 of manure plus N-P-K and 60 Mg ha -1 year -1 plus P-K, the order of export was K > Ca > N > Mg > S > P. On the other hand, under fertilizations with organic source with doses of 60 Mg ha -1 year -1 of manure and with 90 Mg ha -1 year -1 of manure and organomineral source with 90 Mg ha -1 year -1 plus K, the order was K > Ca > N > Mg > P > S.
The inversion in the order of P and S, in the description above, is due to the fact that the addition of organic sources to the soil reduces the adsorption capacity of P, increases the available content and promotes greater mobility in the soil profile of soluble organic forms of P. This is more evident because the driving force of this type of transport is the concentration gradient of the element between the rhizosphere and its surroundings .
In cycles II and III, under fertilizations with mineral source of N and organomineral source with supply of also mineral N, the behavior of S export was similar to that of the first cycle, but 150% higher than P, due to the application of ammonium sulfate (Silva et al., 2016a).
For the macronutrient balance considering the difference between the amount added to the soil and the amount exported at harvest, there is variation with the cycles, fertilizations and dry matter yield (DMY). The N balance was positive in the first cycle for all fertilizations due to the greater supply of the nutrient and lower DMY (Table 2). In the second cycle, it was negative under organic fertilization with 30 Mg ha -1 year -1 of manure and, in the third cycle, due to the higher DMY, therefore higher export for the same dose applied in the second cycle, it was positive under the fertilizations with greater supply of the nutrient, organic and organomineral with 90 Mg ha -1 year -1 of manure.
The balance of P and S was positive in the three production cycles for all fertilizations, being more efficient in cycle I due to the lower DMY and greater amount added of these nutrients, differently in cycle III due to the higher DMY and the same amounts of P and S added compared to the second cycle. Both the inorganic sources used, single superphosphate (18% of P 2 O 5 and 12% of S) and ammonium sulfate (24% of S), and corral manure (1.08% of P 2 O 5 and 0.23% of S) justify these results.
The balance of K was negative in the first production cycle under organic fertilization, with 30, 60 and 90 Mg ha -1 year -1 of manure (Table 2), agreeing with Donato et al. (2017c) because the manure is not enough to supply the K required by the 'Gigante' cactus pear, depending on the chosen harvest age. It was also negative under fertilization with mineral source with supply of 300 kg ha -1 year -1 of K 2 O. In the second production cycle, the balance of K was positive under mineral fertilization with 600 kg ha -1 year -1 of K 2 O and organomineral fertilization with 30 Mg ha -1 year -1 of manure plus 600 kg ha -1 of K 2 O, probably because of the lower export due to the lower DMY, compared to organomineral doses with 60 and 90 Mg ha -1 year -1 of manure plus 600 kg ha -1 of K 2 O, which led to negative balance of K despite the higher supply of K. In the third production cycle, the balance of K was negative for all fertilizations due to the high export of this nutrient in the cladodes harvested (Silva et al., 2016a;Donato et al., 2017b) promoted by the higher DMY.
In the three production cycles, the balance of Ca and Mg was negative, except in cycle I for Ca, in which fertilization with mineral source and organomineral source led to positive balance because the source used to supply P, single superphosphate, also contains Ca (18%), and also the lower DMY of cycle I, hence lower export of Ca, and the greater addition of the nutrient compared to the production cycles II and III. In these cycles, the balance of Ca was more negative in cycle III due to the higher DMY and the same dose of Ca applied.
In relation to Mg, the negative balance under all fertilizations is justifiable, because in fertilizations with mineral source Mg was not added and in fertilizations with organic or organomineral source the added Mg came from the bovine manure, a source with low content of the nutrient (20 mg kg -1 ). However, it became more evident in cycle III due to the greater export for the same dose applied in cycle II and more subtle in cycle I due to the lower DMY and the greater addition of macronutrients. The amount of nutrients added corresponds to the sum of the applications performed at planting and repeated in the second rainy season, before the first harvest, hence representing twice the nutrients supplied to the soil compared to cycles II and III.
The final balance of nutrients after three production cycles (Table 3) was positive for N, P and S in cactus pear subjected to all fertilizations with mineral, organomineral or organic sources; however, it was negative for Ca and Mg under all fertilizations and positive for K under fertilization with mineral source with supply of 600 kg ha -1 year -1 of K 2 O and organomineral source with 30, 60 and 90 Mg ha -1 year -1 of manure plus 600 kg ha -1 of K 2 O.
Such higher export of K, Ca and Mg by the cladodes harvested, compared to the amount added, is possible due to the buffer capacity of the soil, which replaces these nutrients to the solution. Moreover, the soil does not constitute a sink for these nutrients and, therefore, the plant does not undergo competition in relation to the soil because all these elements retained in the exchange complex will recompose the nutrients absorbed from the solution, as they are absorbed by the plant. This is because the bonds of these elements with the negative charges of the soil are ionic or electrovalent, with very low adsorption energy (Novais & Mello, 2007). Tables 2 and 3 show the need for replacement via fertilization of the exported amounts of K, Ca and Mg. In systems with organic fertilization, K supplementation is required, particularly when the DMY target to be achieved with cultivation is greater than 20 Mg ha -1 cycle -1 (Donato et al., 2017a). Another important point is the incorporation into the production systems of the recommendation of fertilization with fertilizers containing Ca and Mg, regardless of the need for liming of the area (Donato et al., 2017a), because the high export of these nutrients in cladodes can lead to depletion of soils, especially when * -Mg ha -1 of manure -N (kg ha -1 ) -P 2 O 5 (kg ha -1 ) -K 2 O (kg ha -1 ) Table 3. Final balance of macronutrients (N, P, K, Ca, Mg and S) after three production cycles as a function of the average export by the cladodes of 'Gigante' cactus pear at harvest and the amount added to the soil by different treatments with organic, organomineral and mineral fertilizations considering the longevity of cultivation and the achievement of high DMY targets. These measures contribute to avoiding the impoverishment of soil and maintaining their production capacity, which is more worrisome in soils of low fertility and low buffer power for these nutrients. Additionally, there is no contribution of biogeochemical nutrient cycling because cactus pear does not restore nutrients via crop remains after harvest; consequently, soil depletion becomes inevitable if the required amounts of nutrients are not added.
In general, the highest recovery rates in each cycle are associated with the lowest doses of the nutrients applied, and the lowest recovery rates, with the highest amounts added to the soil, whether by mineral, organomineral or organic sources (Table 4). For example, the higher K recovery rates in the three production cycles are associated with organic fertilization with 30, 60 and 90 Mg ha -1 year -1 of manure, precisely the fertilizations that add the lowest doses of K compared to organomineral and mineral fertilizations, while the lowest recovery rates of N and Mg are also associated with these fertilizations that promote the greatest supply.
The lowest P recovery rates in the three production cycles are associated with the greater amounts of this nutrient, either by organic fertilization with 60 and 90 Mg ha -1 year -1 of manure or organomineral fertilization with 60 Mg ha -1 year -1 of manure plus P and K or 90 Mg ha -1 year -1 of manure plus K. For Ca, the lowest recovery rates were also recorded under fertilization with greater supply of this nutrient, that is, mineral or organomineral fertilizations that used single superphosphate as source of P.
The results are consistent and contribute to the nutritional management of 'Gigante' cactus pear under cultivation conditions in the semi-arid region of Bahia, as they show that the fertilizations performed in most cactus pear plantations do not meet the requirement of the crop during the cultivation cycles.

Conclusions
1. Nutrient exports are higher under fertilizations with greater supply of nutrients, regardless of the source.
2. The order of export is K > Ca > N > Mg > S > P for mineral fertilization, organic fertilization with 30 Mg ha -1 year -1 of manure and organomineral fertilization with 30 and 60 Mg ha -1 year -1 of manure plus N-P-K and P-K. For organic fertilizations with 60 and 90 Mg ha -1 year -1 of manure and organomineral fertilization with 90 Mg ha -1 year -1 plus K, the order is K > Ca > N > Mg > P > S.
3. The balance of nutrients after three production cycles is positive for N, P and S, negative for Ca and Mg under all fertilizations, and positive for K under mineral fertilization with 600 kg ha -1 year -1 of K 2 O and organomineral fertilization with 30, 60 and 90 Mg ha -1 year -1 of manure plus 600 kg ha -1 of K 2 O.
4. The highest recovery rates of K and the lowest recovery rates of N occur under organic fertilization, which add the lowest and highest amounts of K and N to the soil compared to mineral and organomineral fertilizations. Table 4. Apparent nutrient recovery rate (APRR) by 'Gigante' cactus pear as a function of mineral, organomineral and organic fertilizations in three production cycles