Introduction

The wild relatives of the cultivated tomato (Solanum lycopersicum L.) are distributed in Ecuador, Peru, northern Chile and the Galapagos Islands (^{Peralta & Spooner, 2001}), but Mexico is considered as its center of domestication (^{Jenkins, 1948}; ^{Peralta & Spooner, 2007}; ^{Rick & Fobes, 1975}; ^{Rodríguez et al., 2011}). Wild tomatoes grow in diverse habitats, from those found at sea level to almost 3,300 masl (^{Rick, 1973}; ^{Taylor, 1986}).

The species evaluated in this work have a wide distribution. Solanum pennellii L. grows in western Peru; its habitats are not evenly dispersed, but they are grouped along streams in the west of the country, usually between 500 and 1,500 masl (^{Rick, 1973}). Solanum pimpinellifollium L. is distributed along the coasts of Peru and Ecuador and has been used frequently in tomato breeding. The natural distribution area of Solanum peruvianum L. is mainly Peru, where it ranges from the west coast in the Andes to northern Chile (^{Chetelat, Pertuze, Faundez, Graham, & Jone, 2009}). Solanum chilense R. is found mainly in southern Peru to the north of Chile, from 0 to 3,000 masl. The last species studied, Solanum habrochaites S., is located from southwestern Ecuador to the southern part of Peru, between 500 and 3,300 masl (^{Spooner, Peralta, & Knapp, 2005}).

The evolution of the wild relatives of the tomato to the cultivated one resulted in an increase in productivity, but at the same time to a reduced genetic base of the present varieties (^{Ladizinsky, 1998}); therefore, cultivated varieties have been negatively affected by biotic and abiotic factors. To counteract this situation, the use of native germplasm or wild relatives is required for the introgression of new allelic combinations of tomatoes to increase their productivity, quality, resistance or tolerance to biotic and abiotic factors (^{Cervantes-Moreno, Rodríguez-Pérez, Carrillo-Fonseca, Sahagún-Castellanos & Rodríguez-Guzmán, 2014}; ^{Fernie, Tadmor, & Zamir, 2006}; ^{Foolad, 2007}; ^{Gur & Zamir, 2004}; ^{Hernández-Bautista, Lobato-Ortiz, Cruz-Izquierdo, García-Zavala, & Chávez-Servia, 2014}; ^{Hernández-Bautista et al., 2015}; ^{Marín-Montes, Rodríguez-Pérez, Sahagún-Castellanos, Hernández-Ibañez, & Velasco-García, 2016}).

Some authors indicate that the genetic diversity obtained from the tomato’s wild relatives is 95 %, while in the cultivated tomato only 5 % is obtained (^{Miller & Tanksley, 1990}). Currently, one of the strategies in tomato breeding is to use the diversity that was lost during the domestication processes of the current varieties (^{Zamir, 2001}); this diversity must be found in its wild relatives. Therefore, the aim of this work was to characterize agronomically, under greenhouse conditions, accessions of five wild relatives of the tomato for their incorporation into breeding programs of this vegetable.

Materials and methods

This research was carried out at the Colegio de Postgraduados, Montecillo Campus, Texcoco, State of Mexico (19° 27’ North latitude and 98º 54’ West longitude, 2,246 masl), in the greenhouses of the graduate school’s Program for the Conservation and Improvement of the Genetic Resources of the Tomato in Mexico. In total, 39 accessions of five wild species related to the cultivated tomato (Figures 1, 2, 3, 4 and 5, Table 1), provided by the Tomato Genetics Resource Center (TGRC) of the University of California, Davis, USA, were evaluated.

Figure 1 S. pimpinellifollium L. accession LA0373 

Figure 2 S. peruvianum L. accession LA1365 

Figure 3 S. habrochaites S. accession LA2409 

Figure 4 S. pennellii L. accession LA1272 

Figure 5 S. chilense R. accession LA2759 

A completely randomized experimental design was used with four replicates of ten plants each. Sowing was carried out on May 28, 2014, and the transplant (to 12-L polyethylene bags), 36 days after sowing. As substrate, volcanic sand (red tezontle) was used. The plants were irrigated with the nutrient solution proposed by ^{Steiner (1984)} at 25 % during the vegetative stage, at 50 % in flowering and at 100 % during fruit ripening. Additionally, Confidor^® (imidacloprid) and Ampligo^® (50 lambda cyhalothrin + 100 chlorantraniliprole) were used for the control of whitefly (Bemisia tabaci Gennadius), Captan^® 50 plus (carboxamide) and Ridomil Gold^® (metalaxyl-m + chlorothalonil) for late blight (Phytophthora infestans), and Amistar^® (azoxystrobin) for early blight (Altenaria solani).

According to the tomato descriptors manual of the International Plant Genetic Resources Institute (^{IPGRI, 1996}), 12 traits were evaluated: days to flowering (DF), days to maturity (DM), leaf length (LL, cm), leaf width (LW, cm), stem diameter (SD, cm), number of flowers per cluster (FC), cluster length (CL, cm), fruit weight (FW, g), fruit length (FL, cm), fruit width (FWi, cm), total soluble solids (TSS, °Brix) and number of seeds per fruit (SF). For the measurement of SD, FL and FWi, a Truper^® digital standard and millimeter Vernier caliper was used. The LL, LW, CL traits were measured with a Truper^® model FH-3M flexometer. The FW was obtained with an Ohaus^® model SP2001 digital scale. An ATAGO^® model PAL-1 digital refractometer with a range of 0.0 to 53.0 °Brix was used to evaluate TSS.

Average, range, coefficients of variation and standard deviation were calculated for each variable. Likewise, analysis of variance and Tukey’s range test (P ≤ 0.05) were performed with the Statistical Analysis System package (^{SAS Institute Inc., 2002}). These tests were carried out in order to determine if there are significant statistical differences within the species evaluated and identify those accessions that had the highest and lowest parameters.

Results and discussion

The traits with the greatest variation among species were DF, FW and SF, with 45.57, 62.30 and 62.57 %, respectively. By contrast, those with the smallest variation were DM, FL, FWi and TSS with 14.70, 16.78, 18.07 and 17.82 %, respectively (Table 2).

In Solanum pennellii L. the traits with the greatest variation were FW, TSS, SF and FC with 27.87, 18.42, 16.22 and 15.48 %, respectively (Table 2). Likewise, significant statistical differences were observed (P ≤ 0.05) among accessions in DM, LL, SD, FW, SF and TSS (Table 3). These differences are the product of the ecological niche and adaptability of the accessions to each of the environments. In spite of the above, the accessions of S. pennellii were the ones with the least variation with respect to the other species, which agrees with the findings reported by ^{Rick and Tanksley (1981)}, who found that S. pennellii L. has stable and less variable characteristics between individuals and accessions.

Accessions LA2580 and LA0716 showed self-compatibility by presenting less variability with respect to the rest of the evaluated accessions, which presented self-incompatibility, in DF (2.7 %), SD (1.7 %), CL (1.6 %) and TSS (1 %) (Table 3). This agrees with what was reported by ^{Mercer and Perales (2010)}, who indicate that the genetic variation of individuals is influenced by the type of reproduction, since individuals who have autogamy (self-compatibility) systems have less variation within the population and more between populations.

As for the quality of the fruit, ^{Fernie et al. (2006)} indicate that the increase in TSS in S. pennellii L. is the result of an increase in sucrose and glucose. The comparison of means showed that LA1272 and LA1367 have the highest amount of TSS (9.3 and 9.7, respectively) (Table 3). Therefore, these accessions can be used to improve the fruit quality of the elite tomato varieties.

On the other hand, among accessions of Solanum pimpinellifollium L., the traits with the greatest variation were DM (21.24 %), FC (23.43 %), CL (37.57 %) and SF (33.71 %) (Table 2). Likewise, there were significant differences (P ≤ 0.05) between the means of collections, with the exception of DF and FW (Table 4). ^{Rick and Chetelat (1995)} indicate that in S. pimpinellifollium L. the type of inflorescence, stem diameter, and days to flowering and maturity are very similar to those of the cultivated tomato, which makes this wild species the most used in tomato hybridization. In addition, both species are self-compatible and have red fruit, with the shape and size of the fruits being a relevant factor in genetic improvement (^{Rick & Forbes, 1975}).

^{Galiana-Balaguer, Roselló, and Nuez (2006)} concluded that the TSS content in S. pimpinellifollium L. is high. In general, all the accessions of S. pimpinellifollium L. evaluated in the present work had higher TSS than those commonly presented by 'Saladette'-type hybrids, which oscillate between 3.9 and 5.2 °Brix (^{Bonilla-Barrientos et al., 2014}; ^{Hernández-Leal et al., 2013}). ^{Rodríguez, Pratta, Zorzoli, and Picardi (2006)}, when studying a population of recombinant lines derived from the cross between S. lycopersicum cv. Caimanta and the accession LA722 of S. pimpinellifollium L., found an increase of 1.6 °Brix and 19 days of shelf life with respect to the female parent. Therefore, accession LA1576, which presented 9.3 °Brix (Table 4), can be an alternative to improve the internal quality of tomato fruits.

With respect to Solanum peruvianum L., the traits with the greatest variability were DF, FC, CL, FW, SF and FL, with 36.34, 35.31, 30.58, 57.82, 21.11 and 26.27 %, respectively (Table 2). This variability is due to its reproduction system (allogamy). Given that cross-pollination is required in individuals, due to their self-incompatibility, they have greater variation compared to those with autogamy (^{Rick, 1988}). The above can be observed in Table 2, where, with the exception of DM, the traits have coefficients of variation greater than 13 %. Accession LA1982 had later flowering (40 days) and ripening (89 days), greater leaf length (30.2 cm) and width (16 cm), and greater stem diameter (0.99 mm) and cluster length (41 cm) (Table 5); this suggests that LA1982 can be exploited in breeding programs of cultivated tomatoes.

^{Chetelat et al. (2009)} reported that the number of seeds of the evaluated accessions of S. peruvianum L. varies between 22.5 and 50 seeds per fruit. These values are much lower compared to those obtained in this research, which varied between 77 and 166.

Most of the traits evaluated in the Solanum habrochaites S. accessions had coefficients of variation greater than 14 % (Table 2). The traits with the greatest variation were FW (59.9 %), SF (35.37 %) and LW (20.48 %). Among accessions there were significant statistical differences (P ≤ 0.05) in the traits evaluated, except for FC (Table 6). Five accessions of this species were characterized as self-compatible, while the other three were self-incompatible, so their propagation is through cross-pollination; this generates greater diversity (^{Peralta & Spooner, 2001}). Only LA1223 produced fruit without the need to manually pollinate.

^{Carter, Gianiagna, and Sacalis (1989)}, in a study of tolerance to the Colorado beetle (Leptinotarsa decemlineata Say), concluded that the leaves of S. habrochaites S. contain zingiberene, a compound that promotes partial tolerance to this insect. In the accessions evaluated in the present work, LA2650 had the largest leaves, so this represents an alternative for tolerance to this pest, by associating leaf size with greater zingiberene production.

On the other hand, accession LA1777 has been widely used in tomato breeding, since it has alleles that increase fruit yield and TSS, detected on chromosomes 1 and 4, respectively (^{Bernacchi et al., 1998}; ^{Monforte & Tansksley, 2000}). However, it was found that LA2158 had greater weight and size, statistically different from those of LA1777, so it could be a better alternative for the breeding program’s objectives.

Among the seven accessions of Solanum chilense R., it was found that the traits with the greatest variation were DF, LL, SD, FC, FWi and SF (Table 2) with significant differences (P ≤ 0.05) in most of them, except in FW (Table 7). LA1960 presented the highest value in number of seeds per fruit (82), while LA2748 and LA2759 were superior in the number of flowers per cluster (28). These traits are important for determining gene flow, and therefore the genetic and evolutionary structure of the species (^{Barrett, 2008}).

The species S. chilense R. is self-incompatible (^{Breto, Asins & Carbonell, 1993}), which promotes cross-pollination, giving rise to a wide genetic diversity among and within the accessions. This explains the fact that the evaluated accessions of this species have eight traits with a coefficient of variation greater than 20 %; on the contrary, a self-compatible species such as S. pimpinellifollium L. presented only four traits with coefficients of variation greater than 20 %. ^{Rick (1988)} indicates that S. chilense R. has a wide diversity, since cross-pollination is required; therefore, these accessions represent an ample source of genes, not only for the traits evaluated, but also in the resistance to viral diseases (^{Griffiths & Scott, 2001}, ^{Stamova & Chetelat, 2000}).

On the other hand, ^{Chetelat et al. (2009)} found 20 to 50 seeds in collections of S. chilense, S. peruvianum and S. pennellii made in the Atacama Desert in northern Chile, a range in which the values of the species evaluated here are found.

Conclusions

The accessions within species showed wide variation, which makes them a promising germplasm to be used in the development of breeding programs. In this sense, the accessions with better characteristics were LA1272, LA1367, LA1576 and LA177, which can be a source of new allelic versions to improve the fruit, while LA1982 and LA2650 can help improve the archetype of the cultivated tomato.

The traits with the greatest variation among species were DF, FW and SF, while those with the smallest variation were DM, FL, FWi and TSS. The species with the greatest differentiation were S. peruvianum L., S. chilense R. and S. habrochaites S., because they had higher coefficients of variation compared to the rest of the species evaluated.

Solanum pennellii L. presented the lowest coefficients of variation among accessions in most of the variables evaluated.