Treatment 1

Recirculation without chemical analysis of the draining solution. In a 5 000-liter water tank, the nutrient solution was prepared at 100% concentration. It was applied in drip irrigation, providing 20% drainage at each irrigation. Each time the vat was emptied, 4 000 L were prepared with a new nutrient solution at 100% concentration plus 1 000 L of the drained, collected and disinfected solution, without making corrections of pH, CE or any of its nutrients.

Treatment 2

With chemical analysis of the drained solution. In a vat of 5 000 liters capacity, starting from a nutritive solution at 100% concentration, it was applied in drip irrigation providing 20% drainage in each irrigation. Each time the vat was emptied, 4 000 L were prepared with a nutrient solution selectively adjusted by nutritional element (N, P, K, Ca, Mg and S) basedon a chemical analysis of the drainage that was done every 15 days, the remaining 1 000 L were contributed with the drained nutrient solution, collected and disinfected.

Treatment 3

With solution adjustment according to theoretical absorption. In a 5 000 L water tank, starting with a nutrient solution at 100% concentration, it was applied in drip irrigation, providing 20% drainage in each irrigation. Each time the vat was emptied, 4 000 L were prepared with a selectively adjusted nutrient solution considering an estimate of the absorption rates of tomato plants for each nutritional element (90% for N, P and K and 50% for Ca, Mg and S), compared to the initial nutrient solution (^{Adams, 2004}; ^{Velazco et al., 2012}; ^{Urrestarazu, 2015}), the missing 1 000 L were supplemented with the drained, collected and disinfected solution.

Treatment 4

Witness, with open system. In a vat of 5 000 L capacity, the nutrient solution was prepared at 100% concentration and applied in drip irrigation, seeking 20% drainage and each time the tank was emptied, 100% of the initial volume (5 000 L) was prepared again with the complete nutrient solution.

The frequency of irrigation and the amount of nutrient solution to be applied daily seeking 20% drainage of nutrient solution in each treatment, it was achieved by applying between 3 and 6 irrigations daily depending on the climatic conditions and phenological stage of the crop.

The drainage of each treatment was collected and temporarily stored in 1 000 L vats before recirculating it, returning it every fifth to seventh day to the 5 000 L vat that corresponded to it according to the treatment. Oxygen water (50% hydrogen peroxide) was applied prior to recirculation to disinfect the drained solution at the rate of 50 ml of commercial product per 1 000 L vat (60 ppm H₂O₂ in solution).

For the chemical analysis of the drained solution required to define the concentration of salts in the preparation of the nutrient solution in treatment 2, a sample of 500 ml was taken from the 1 000 L tank collected from this treatment, placing it in amber bottles for laboratory analysis. To determine nitrogen the micro-kjeldahl method was used, for phosphorus the vanadato-molibdate method was used, reading in a GENESYS 10 UV spectrophotometer with an absorbance of 420 nm and potassium was determined by means of a JENWAY flame photometer (^{Chapman and Pratt, 1973}).

To measure calcium and magnesium, a Phillips Pye Unicam SP 9 atomic absorption spectrophotometer was used and finally sulfur was determined by the turbidimetric method with barium chloride (^{Chesnin and Yien, 1951}). The plants were blunted (removal of the terminal bud from the main stem) at 81 dds, leaving two leaves above the third formed inflorescence. All the lateral shoots were eliminated as they appeared, leading the crop to a single stem per plant, from where the fruits of the three clusters were harvested.

A random complete block design was used with four treatments and seven repetitions. The experimental unit was 20 m² (240 plants). The data obtained were subjected to variance analysis and average comparisons with the Tukey test (p= 0.05), using the ^{SAS statistical program (2002)} version 9.0. Morphological variables (plant height, stem diameter and foliar area), dry weight (total plant and fruit), and yield and its components (average weight and number of fruits) were evaluated.

The plant height (cm) was measured from the base of the stem to the height where the apex pruning was made. The stem diameter (mm) was measured with an electronic vernier ‘digimatic caliper’ (Mitutoyo, model CD-6 CS) at the height of between the first and second inflorescence. The foliar area per plant (cm²) was determined with the support of a foliar area integrator brand LI- 3000A, Lincoln, Nebraska and with this data the foliar area index (m² leaf m^-2 of covered area) was estimated.

The total dry weight per plant (g) was measured with the support of a balance and stove for drying at 60 °C to constant weight. The dry weight of fruit (g) was measured with a balance drying it at the same time as the rest of the plant, with this data and with the total dry weight per plant, the harvest index (dry weight of fruit/total dry weight of plant) was estimated. The number of fruits (fruits m^-2) was obtained from the sum of the commercial quality fruits obtained in each cut.

The total yield was calculated per unit of surface area (kg m^-2), adding the weight of the fruits in each of the cuts harvested in each m². The average fruit weight (g) was calculated by dividing the yield per unit surface (g m^-2) by the number of total fruits harvested. Morphological variables were measured at 102 dds and dry weight variables were measured at 143 dds. In addition, water and macronutrients expenditure were calculated and plant tissue analysis of the plants managed was performed in each of the tested treatments.

The water consumption (L m^-2) was determined by adding up the number of liters of nutrient solution applied during the transplant period at the end of the harvest, and the expenditure of macronutrients (g m^-2) was calculated based on the concentration (mg L^-1) of the nutrients with which the nutrient solution was prepared in each treatment and the amount (liters) applied throughout the crop cycle from transplant to end of harvest.

Since these determinations were made once the crop cycle had ended, comparisons between the different treatments were made only on the basis of the absolute values calculated, without statistical tests being applied. The analysis of plant tissue was also performed at 143 dds, for which three plants were taken per experimental unit in each of the repetitions. With the three plants a composite sample was made, which was dried in a stove at 70 °C.

With a WILEY mill model 4 of 110 VAC, the samples were ground and then stored in transparent glass flasks for later digestion. In a Kjeldahl flask was placed 0.5 g of ground sample, to which 4 ml of a mixture of sulphuric acid and perchloric acid was added in proportion 4:1 plus 1 ml of 30% oxygenated water, giving a black extract. The flask was then placed on a hot plate until the extract boiled and changed to light color.

Then it was allowed to cool for one hour, then it was made up to 50 ml with distilled water. From the solution obtained with this procedure, the determinations of nitrogen, phosphorus, potassium, calcium and magnesium were made (^{Alcántar and Sandoval, 1999}). For sulfur, similar alternating digestion was carried out, but in this case nitric acid was used in the mixture instead of sulfuric acid in 2:1 ratio (^{Alcántar and Sandoval, 1999}). The determination of each of the macronutrients was done with the same methodologies used for the drained nutrient solution mentioned above.

Morphological variables

The variance analysis of this group of variables (data not shown) indicated that there was a highly significant difference for plant height and significant for stem diameter, but not for foliar area index. Mean comparisons (Table 1) show that the plant height in the treatment with 20% recirculation without chemical analysis of the drained solution (T1) was higher with respect to the other two recirculation treatments (T2 and T3) and of the witness (T4).

There is no special explanation for the increased height growth of plants from simple drainage recirculation treatment (T1), possibly it is explained by the influence of differences in microclimatic factors that were presented in plants from the seedbed and that differentially affected their growth in height. For stem diameter, the recirculation treatment with prior analysis of the drains (T2) overcamed the witness without recirculation, with no differences between the other treatments. In foliar area index (IAF), all treatments were statistically equal, with an oscillation between 4.5 and 5.3.

These values are possibly high for optimal net photosynthesis, since according to ^{Sánchez and Moreno (2017)}, the optimal IAF values for a greater accumulation of daily dry matter in tomato handled in greenhouse and high population density should be between 3 and 4. Values greater than 5 cause excessive shading between plants, which could be noticed in the crop, which affects the rate of net photosynthesis per plant and consequently the final yield (^{Taiz et al., 2014}).

Dry weight variables, harvest and yield index and their components

Both the variance analysis (data not shown) and the mean comparison analysis (Table 2) indicate that, for the total dry plant weight variables, harvest index, number of fruits per plant, average fruit weight and yield, there were no significant differences between treatments.

The dry weight and harvest index results obtained suggest that, in tomato handled in a short cycle such as that established in this study, with any of the recirculation systems tested does not affect the growth of the plants or the distribution of the dry matter that is directed towards the fruits, compared to an open system, results that coincide with ^{Sánchez et al. (2014)}, who, in evaluating different recirculation systems of the nutrient solution in tomato, also managed in a short cycle, found no difference between treatments for this variables.

Instead, ^{Tunali et al. (2009)}; ^{Massa et al. (2010)} reported decreases in the growth of tomato plants when using a nutrient solution recirculation system, possibly because they drove the crop with conventional long-cycle management (^{Sánchez et al., 2012}), where over time, there are significant imbalances in the drained nutrient solution (^{Tunali et al., 2009}; ^{Van Os, 2009}).

The yield oscillated between 11.13 and 12.71 kg m^-2 of greenhouse, over a period of 110 days of transplantation at the end of harvest, which gives the opportunity to obtain three growing cycles per year. Taking as an example the average yield of the three systems with recirculation (11.81 kg m²), there would be an annual yield potential of at least 350 t ha^-1, which is similar or even slightly higher than that reported in companies with high technical performance under the long cycle system (Castellanos and Bourbon, 2009).

Surely the short period with which the harvest was finalized (less than four months from transplantation) allowed to escape the nutritional imbalances in the recirculating nutrient solution that occur over time (^{Tunali et al., 2009}; ^{Van Os, 2009}; ^{Sánchez et al., 2014}), so there were no differences in yield between the closed systems and the open system studied. In other work on recirculation of the nutrient solution with tomato, where the growing cycle is much longer, significant decreases in yield have been reported with respect to open systems (^{Tunali et al., 2009}; ^{Van Os, 2009}).

In addition, with the management of the crop with long cycles, recirculation promotes the appearance of diseases that gradually spread throughout the crop affecting yield and quality (^{Sánchez et al., 2014}). In similar research with the high population density system and cut to the third cluster, yields of around 15 kg m^-2 have been achieved in four months transplant cycle to harvest (^{Sánchez et al., 2012}; ^{Sánchez et al., 2014}).

However, the yield obtained in this studio was relatively low, possibly due to the high index of foliar area formed (about 5), which caused shading between the plants and a higher rate of maintenance breathing, affecting the daily net photosynthesis and with it the yield (^{Sánchez and Moreno, 2017}).

In any case, with the results obtained it can be noted that, any of the recirculation treatments of the nutrient solution evaluated, is able to be implemented in the tomato production system in short cycles, without affecting the yield, with advantages in saving water and fertilizers with respect to the witness treatment without recirculation as shown below, and that contamination of the water table would also be avoided as mentioned by ^{Nakano et al. (2010)}.

Expenditure of water and fertilizers

Table 3 shows the amounts of water and nutrition provided per m² of greenhouse during the end-of-harvest transplant period. It is observed that the water expenditure was 437.6 L m^-2 of greenhouse for each of the treatments with recirculation of nutrient solution, while for witness treatment without recirculation (T4), the expenditure was 547.2 L.

This difference of 109 l m^-2 was due to the fact that, in the first three treatments, the drained nutrient solution was recirculated, corresponding to approximately 20% of the total applied. Extrapolating these data to a greenhouse hectare and establishing three growing cycles per year, which is possible with the system of three clusters per managed plant, a water saving of approximately 3 300 m³ per year would be obtained by recirculating the drained nutrient solution, which represents a considerable decrease in the use of this resource.

Benoit and Ceustermans (1995) cited by ^{Gul (2011)} mention that, for soilless crops, with 15 to 25% over-irrigation of nutrient solution, approximately 2 900 m³ ha^-1 year^-1 are lost, similar to what is observed in this study. ^{Sánchez et al. (2014)} reported savings of 32.6% in water by comparing a system with recirculation of nutrient solution compared to one without recirculation in the cultivation of tomato grown in beds filled with tezontle substrate.

It should be noted that the saving of water with the recirculation of the nutrient solution is of great agronomic importance, in particular for its implementation in areas where water is an increasingly scarce resource for crop production. Table 3 also it is noted that, based on the absolute values calculated, in the three methods of recirculation of the nutrient solution, the supply of nutrition was lower than the witness.

With the recirculation treatment of the nutrient solution without analysis (T1) less nitrogen, phosphorus and potassium were applied (17.5, 4.8 and 22.4 g m^-2 greenhouse less than the witness, this is approximately 16% for each nutrient), while with the recirculation treatment according to the established theoretical absorption (T3), less calcium, magnesium and sulfur were provided (57, 9.6 and 33.6 g m^-2 of greenhouse less than the witness, corresponding to 39, 37 and 41% for each nutriment, respectively). In T1 treatment, savings of up to 20%of all nutrients would be expected from the witness; however, the savings were only in the order of 16% because this treatment continued to apply the complete nutrient solution until one month after transplantation.

In the preparation of the nutritive solution in the T3 treatment for Ca, Mg and S, only 50% of what was applied in the complete solution was supplied; however, the savings were not of that magnitude because the recirculation of the nutritive solution in this treatment was also initiated one month after transplantation. In this regard ^{Sánchez et al. (2014)} in an experiment with recirculation of nutritive solution in tezontle beds in tomato crop, they exhibited macronutrient contributions of 59.3, 31, 135 and 93.3 g m^-2 for nitrogen, phosphorus, potassium and calcium, respectively.

When extrapolating the data obtained to one ha of greenhouse and three growing cycles per year, the savings with the T1 treatment compared to the witness would be in the order of 525 kg of nitrogen, 144 kg of phosphorus and 672 kg of potassium, while with the T3 treatment the savings from the witness would be 1 734 kg of calcium, 288 kg of magnesium and 1 008 kg of sulfur, which from an economic point of view is very important for the producer.

Nutritional analysis

Analysis of variance (data not shown) and mean comparisons (Table 4) shows that between treatments there were no significant differences for any of the nutrients determined in plant tissue, suggesting that plants absorb nutrients in similar proportions regardless of the concentration to which the nutrient solution has been prepared. With these results it can also be noted that with the management of tomato in short cycles can achieve considerable savings in fertilizers with any of the proven recirculation methods, compared to the witness, without affecting the rate of nutrient absorption by plants (^{Adams, 2004}).

In general, it is also observed (Table 4) that plants absorbed a greater amount of potassium, followed by nitrogen and calcium, and in a lower proportion phosphorus, magnesium and sulfur, results consistent with ^{Quesada and Bertsch (2013)}; ^{Vargas-Canales (2014)} in similar studies with tomato. It should be noted that the percentages of macronutrients absorbed by the plants were within the sufficiency ranges reported by ^{Sánchez (2004)} for the cultivation of tomato, indicating that, with any of the management systems evaluated, the plants had available the elements necessary for their normal growth and development (^{Sánchez, 2004}; ^{Alcántar et al., 2016}).