This research was conducted between May and November 2015 in the EEP-INIAP, located at km. 13, Portoviejo-Santa Ana route, in El Cady, Columbus Parish, Canton Portoviejo, with the following geographic coordinates: latitude 0.1° S, and longitude 80° 23' W. EEP is located at 47.40 m.a.s.l.. has an annual average temperature of

Figure 1

Solid line represents average precipitation while the dashed line symbolizes evaporation rate; the hatched area shows the water deficit in the study area from 2008-2015.

In this research, 112 tomato cultivars were assessed for paste production from different tomato seed distributors in the U.S.A. These cultivars were distributed as follows: 1.67 % Harrys Moran, 8.33 % Orsetti Seed, 13.33 % United Genetics, 16.67 % Heiz Seed, and 60 % Ohio University. The basic statistics for the parameters evaluated are summarized in Table 1.

Table 1

Crop management and production data collection.

In May 2015, the research started with the germination of 112 tomato cultivars, using germinating trays of 162 cavities, peat as a substrate and one (1) seed per well; transplantation was carried out 20 days after sowing with 1.5 x 0.4 m spacing, in order to obtain a population of at least 16,666 plants per hectare. Twenty g.plant^-1 of fertilizer was applied with a mix of urea, 8-2-20 and 0-0-60, in a 2-1-1 relation five days after the first application. The second fertilization was applied with ammonium sulfate, potash muriate and calcium nitrate in a 1-1-1 relation after 35 days; finally, a third fertilization using magnesium and YaraMila(r) was applied after 50 days. Fertilization was supplemented with foliar fertilizers, according to the recommendations of the Department of Soil and Water Management of EEP.

Additionally, a drip irrigation system was installed applying irrigation three hours per plant three times a week. A plastic cover (mulch) was used, and the weed control requirements were minimal. Harvesting was carried out manually in a single pass, when fruits were 75% mature.

Sixteen plants from all the cultivars assessed were planted until at least six plants per net plot were obtained. Twenty-five (25) ripe fruits of each cultivar were weighed with a precision balance in order to compute average fruit weight. Total number of fruits from these plots were counted and averaged. In addition, number of affected fruits by pests, diseases, those that were green and had imperfections were also registered. During harvesting, total fruits per each net plot were weighed, and production was established in t.ha^-1.

Tomato samples were analyzed according to total soluble solids (TSS), percentage of acidity, pH, and ascorbic acid according to procedures described by AOAC (2007). TSS of tomato paste was estimated using a refractometer (ATAGO 3T) and readings were corrected to 20°C, while pH was measured using a pHmeter (Orion 420 A +) (AOAC, 2007).

A principal component analysis (PCA) was used to reduce the number of variables (10) and to find those with greater weight. A type of varimax rotation axe was performed and correlations were identified among component variables (Bortz, 2010).

A cluster analysis was performed for the group of industrial tomato varieties with similar characteristics. Distances between each pair of observations were sized using the Gower methodology, which allows working with different kind of variables and standardizing them according to their rank (Bortz, 2010). The cluster analysis was assessed by the Wald minimum variance methodology, pseudo _^t2 test and the distance matrix in order to choose the appropriate number of variables. In addition, a canonical discriminant analysis was executed to find the responsible variables of group formation. Distance between combination pairs was computed in order to observe similarities using the Wilks Lambda test, in order to address the equality hypothesis between clusters. These analyses were carried out with the SPSS program version 21.0 for Windows.

A correspondence analysis was performed in order to associate production and quality characteristics of resulting clusters from the discriminant analysis. Its advantage lies in the generation of two-way tables, based on the Chi-square test associated with orthogonal components (Bortz, 2010); moreover, these analyses were also carried out with the SPSS Statistics 21 software (analytical software, New York, USA).

According to Bortz (2010), the variables with coincidences greater than 0.70, are acceptable to the PCA interpretation. From 10 variables proposed in the methodology, 30% were deleted, including pest resistance variables. Table 2 shows the commonalities of selected variables.

Table 2

Two clusters from the PCA explains 93.98 % of the total variance. The distance matrix for the cluster analysis was generated with seven (7) selected variables. The t² pseudo test suggested creating three clusters (Table 3). The first one grouped 24.11 % of tomato cultivars, the second one 51.78 % and the third one 24.10 %.

Table 3

Three groups of industrial tomato cultivars were classified. Cluster 1 is characterized by high yield and low quality fruit. The second cluster showed average yield and quality. In addition, the third cluster was classified by low production and better quality properties for industrial tomato (Figure 2).

Figure 2

Figure 3 shows the asymmetric map of highest production tomatoes (24.11 %) obtained from the discriminant analysis. The first inertia dimension shifted the 56.70 % of the total variance, separating the number of defective fruits to the left, from yield, average fruit weight, average weight in kg, associating higher production with the seed supplier company (Ohio University).

Figure 3

The discriminant analysis based on the combined covariance matrix of the 10 variables evaluated, according to the Lambda Wilks multivariate test, showed highly significant differences between clusters (P = 0.0001), rejecting the null hypothesis of no difference between clusters. Two canonical variables were obtained from the discriminant analysis; the first function (going from left to right) had a greater weight in characteristics which allow identifying production attributes such as number of fruits per plant, number of defective fruit, disease resistance and yield. The second function was related to quality parameters as ° brix, pH and acidity, from bottom to top of Figure 2.

The second dimension of the asymmetric map clarified the inertia to 26.9 % and contrasts yield, fruits per plant, number of fruits per plant with defective fruit number and quality variables, such as pH, ° brix and acidity. Cultivar SG 07-632 stood out with an average yield of 68.58 t.ha^-1, 72.73 g of average fruit weight, and a production of 72.20 (± 8.39) fruits per plant, followed by cultivar SG 07-627 with an average yield of 64.40 t.ha^-1, 98.18 g (± 8.39) average fruit weight and a production of 50.20 (± 10.45) fruits per plant. Moreover, cultivar SG 07-603 was placed third, showing an average yield of 61.60 t.ha^-1, an average fruit weight of 94.55 g, and a production of 61.78 (± 7.13) fruits per plant.

The asymmetric map of the group with quality features obtained from the discriminant analysis (24.10 %) is shown in Figure 4. The first dimension explains 69% of inertia, where features from average defective fruits oppose with tomato quality properties. The second dimension contributes with 14.70 % of total inertia, contrasting yield with tomato quality characteristics for tomato paste.

The cultivar UG 10109 stood out with a production of 13.00 t.ha^-1, 5 °brix, pH of 4.22, and an acidity value of 10.70. Moreover, cultivar UG 16609 was located in second place with a production of 10.20 t.ha^-1, 5 °brix, a pH of 4.12 and an acidity of 9.80. Finally, cultivar UG 1103913F has a production of 5.10 t.ha^-1, 4 °brix, a pH of 3.97, and an acidity of 10.10. All these materials came from United Genetics.

Figure 4

Fertilization and irrigation is especially important for tomatoes that will be used for paste production, as imbalances from water stress and moisture can lead to reduced production. These conditions were not observed during the experiment carried out in EEP. Industrial tomatoes originated in Ohio University were those that showed highest productions. Cultivar SG 07-632 had a production of 68.58 t.ha^-1 and was ranked in the first place, followed by cultivar SG 07-627 with an average production of 64.40 t.ha^-1 and cultivar SG 07-603 with 61.60 t.ha^-1. Figueiredo, Resende, Faria, Paula, Rizzardi & Meert (2016), when carrying out an agronomic evaluation of industrial tomatoes found that the best yields were obtained from hybrids RVT-07 x RVT-10 (77.05 t.ha^-1) and RVT-02 x RVT-03 (70.65 t.ha^-1). These results contrast with production values reported by Giordano, Fonseca, Silva, Inoue-Nagata & Boiteux (2005), who recorded 107 t.ha^-1 of tomato for industrial purposes. On the other hand, tomato paste yield values assessed in this research were even above the industrial tomato production values reported in India, with an average of 14.16 t.ha^-1 and also compared to a world average of 34.70 t.ha^-1 (FAOSTAT, 2015) Likewise, Figueiredo et al. (2016), on a genetic improvement study carried out for industrial tomatoes in Brazil found that no hybrid exceeds the hybrid test called Tinto, which showed fruit weight of 104.40 g.fruit^-1. However, plant materials used in this study were lower to those shown by cultivar SG 07-632 from Ohio University, with an average of 73.73 g, and lower to cultivar SG 07-627 with 98.18 g (Ohio University).

In relation to production parameter as fruits per plant, according to the study conducted by Santos, Melo, Santos, Carvalho Filho & Menezes, (2014), in general, high fruit production per plant should not be considered as the best production parameter, because most of these fruits become small on the same plant, reducing fruit quality and production. Under this consideration, in this study variation in fruit production per plant for the higher production group was 72.20 fruits per plant. However, studies published by Santos et al. (2014), reported a production of 96.15 fruits per plant under greenhouse conditions, and 57.89 fruits per plant under field conditions. These values are higher than those registered for tomatoes in Brazil used for industrial purposes (Giordano et al., 2005). However, cultivar SG 07-627 showed values of 50.20 fruits per plant and cultivar SG 07-603 displayed 61.78 fruits per plant, which showed to be below the data registered in Brazil (Giordano et al., 2005). This could be due to stress caused by high temperature in different tomato crop environments as this reduces the production of pollen grains in susceptible tomato genotypes (Santos et al., 2014).

We must state that industrial tomato cultivars assessed in this research have been exposed to tropical dry forest conditions, with high temperatures and increased levels of carbon dioxide (CO₂) in the atmosphere. Changes in CO₂ concentration related to environmental temperature can alter plant tissues in terms of growth and physiological behavior. Many of these effects have been studied in detail for tomatoes under cultivation (Moretti et al., 2010). Exposure of tomato fruits to temperatures above 30°C inhibits many normal fruit ripening parameters, including color development, sagging, respiration radios and ethylene production (Santos et al., 2014).

Production of the cultivars with the highest fruit qualities were those that showed lowest production. Thus, the cultivar UG 10109 produced 13.00 t.ha^-1, cultivar UG 16609 showed a yield of 10.20 t.ha^-1 and cultivar UG 1103913F was the lowest yield producing cultivar with only 5.1 t.ha^-1. These production values can be compared with hybrids studied by Figueiredo et al. (2016), who found the following production values: RVT-01 showed 35.17 t.ha^-1, RVT-02 showed 40.13 t.ha^-1 and RVT-09 showed 40.35 t.ha^-1, which are all placed in the same statistical significance range.

Likewise, pH is an extremely important factor for product acceptance level analysis by consumers, given that excessively acidic fruits are rejected by them (Casierra & Aguilar, 2008). According to Hernández Suárez, Rodríguez & Romero (2008), tomato is considered acid if it has a pH of less than 4.5. However, all assessed materials in this study with better quality characteristics obtained from the United Genetics home provider, were classified as acid. On the other hand, high temperature in tomato crops leads to low amount of citric, malic and oxalic acid, in contrast to high amounts of ascorbic acid, fructose, glucose and sucrose synthesis.

On the other hand, high increases in temperature allow fruit growth and coloration during their development (Moretti et al., 2010). These conditions were observed during this research.

Tomato flavor is generally determined by soluble solids content, acidity (measured as entitled acidity) and the presence of several volatile flavors. Moreover, some ethereal flavors are generated through enzymatic catalytic reactions (Wang, Sun, Feng, Cao & Jiang, 2008). However, the volatile flavors were not analyzed in this study. Average of cultivars with highest fruit quality was 3.62°brix (± 0.96). The best industrial tomato cultivars were as follows: UG 10109 with 5°brix, UG 16.609 with 5°brix, and UG 1103913F with 4°brix. These values were however, above the values found for the RVT-08 line studied by Figueiredo et al. (2016), who registered the highest value of 4.90°brix in their experiment with industrial tomatoes.

Furthermore, tomato juice with ca. 4 to 6°brix is a requirement for long-term tomato paste storage, which influences taste, pulp production and finally the price paid for this raw material (Barrett, Weakley, Diaz & Watnik, 2007). Hannan, Ahmed, Roy, Razvy, Haydar, Rahman, & Islam (2007), during their study about heterosis, combined ability and genetic over °brix, number of days of first fruiting and production, and observed additive and non-additive effects for total soluble solids and °brix. Likewise, they found a high statistical significance between tomato cultivars characterized by good tomato paste features, which were associated with a high level of °brix that contribute favorably to tomato flavor. Total soluble solids are one of the main components in tomato flavor and influences consumer preferences and industrial performance (Graça et al., 2015). On the other hand, soluble solids are important not only for their contribution to flavor, but also in terms of processing requirements. In this sense, it is desirable to produce tomatoes with high soluble content because they require less energy to evaporate the water in industrial processes (Graça et al., 2015).

Collecting tomato fruits in their very early developmental stage would reduce the organoleptic quality of these materials. In this research, titratable acidity ranges between 10.70 and 3.50 were found. This acidity variation among 27 industrial tomato cultivars, suggests that the harvest point of 75% was not suitable for the tomato material assessed. Consequently, Santos et al. (2014), observed that, titratable acidity and soluble solids content were 20 % higher and 10% lower respectively, in those tomatoes directly exposed to sunlight when mature.