Evaluation of local onion lines from northwest Spain

Traditional onion (Allium cepa L.) varieties are still produced in certain regions of Spain due to their high quality and acceptance in local markets. The evaluation of morphological traits for 18 local northwestern Spanish onion lines showed the existence of three different groups attending exclusively to bulb shape traits and six classes if we consider bulb shape traits as well as skin and flesh colours. There was a positive correlation between storage quality and percentage of dry matter content. The importance of considering well defined descriptors in order to characterize correctly onion germplasm is also discussed in this paper. Additional key words: Allium cepa, characterization, dry matter, germplasm, storage quality.

Palabras clave adicionales: Allium cepa, calidad de almacenamiento, caracterización, germoplasma, materia seca. quality such as bulb size (Mattana and Lobo, 1980;Patil and Kale, 1985b;Rajapakse et al., 1992), dry matter (Mattana and Lobo, 1980), total soluble solids (Albert and Cuquerella, 1979;Patil and Kale, 1985a) or sugar contents (Patil and Kale, 1985a). In 1998, the horticultural research group at the CIAM (Centro de Investigaciones Agrarias de Mabegondo) collected seeds of local onion varieties in the main productive regions in northwest Spain. As a result of this sampling, 20 onion lines are kept at the CIAM germplasm bank. This work presents the results obtained from the evaluation of morphological, agronomical and storage quality characters for 18 of these lines in comparison with four commercial varieties, frequently used by growers in this part of the country.

Material and Methods
Twenty-two lines of onion were evaluated in northwest Spain over four years from 1999 to 2002. The site is located at Mabegondo (43º 15'N, 8º 18'W) near the coast. Eighteen of the total number of lines corresponded to local onion landraces collected as seeds obtained by farmers in the main growing regions of northwest Spain (Table 1). The characterizations  comprised morphological and agronomical measurements, carried out in 1999 and 2000, as well as a  storage quality evaluation which took place in 2001  and 2002. All of the lines as well as four commercial varieties were sown under greenhouse conditions, the seedlings were transplanted to the field in May and the harvest took place in September. The experimental design was a complete randomized block design with three replications for a total of 84 plants per plot (0.25 × 0.15 m).
The following measurements were recorded on 25 bulbs per landrace, which were randomly collected at the end of the experiment: bulb height (cm), diameter (cm), weight (g), and neck width (cm). The remaining parameters corresponding to bulb shape, skin colour, flesh colour and number of axes -were determined following the TG/46/6 UPOV (1999) guidelines.
In order to carry out the storage quality evaluations, 100 healthy, non sprouted, bulbs were collected randomly per line after harvest and stored in wooden boxes at 10-15ºC for the following 24 weeks. Every 4 weeks each line was evaluated counting and weighting the sprouted, rotted and healthy bulbs. At the beginning of the storage experiments the percentage of dry matter was estimated for each of the studied lines: a representative 500 g sample was obtained mixing 20 slices cut from 20 bulbs randomly sampled from each line. The sample was then dried at 80ºC for a period of 48 h by means of a stove, calculating the percentage of dry matter as a relation between fresh and dry weights.
An analysis of variance of these morphological traits was carried out independently for each year of evaluation. The statistical model used for these data was the following (Cubero and Flores, 1995): where: M u is the overall average , g j is the line effect, and E ijk is the residual effect.
Yield data from plots were analysed using the following model (Cubero and Flores, 1995): within plot) effect. The block effect was considered a random effect.
Multivariate relationships among lines were revealed with a principal component analysis (PCA) using a correlation matrix derived from the significant traits after the analysis of variance. Components with eigenvalues greater than one were used through hierarchical clustering analysis, based on Euclidean distance computed between each population. The dendogram formed by this method was cut at the third level cluster, each cluster being represented on the Principal Component Plan 1-2.
Regression analyses were performed to determine the influence of dry matter contents on storage quality (percentage of sprouted bulbs 24 weeks after harvest). Curves with the highest R 2 were chosen as best fit.

Morphological characterization of the lines
There were certain characters with great variation between lines, i.e.: traits related to bulb size (height, diameter and shape) as well as skin colour ( Table 2). The analysis of variance (ANOVA) for morphological quantitative characters showed highly signif icant differences between lines for all of the traits under study: bulb height, weight, diameter, height-diameter ratio and neck width (Table 3). According to these traits, the lines could be classified into three different groups by means of a PCA which absorbed 58% of the  accumulated variance (Fig. 1). These corresponded to: lines with rhombic bulb -such as 'Ribadeo', 'Oimbra', 'Vilagarcía', 'Outes' and 'Cea'-, with transverse narrow elliptic shape -'Betanzos', 'S. Xulián', 'Mondoñedo', 'Chata-Miño' and 'Pontearnelas' -and lines with bulb shapes ranging from transverse elliptic to transverse narrow elliptic -'Ameixenda', 'Baldaio', 'Caldas', 'Arcade', 'Ponteareas' and 'A Garda'-. A Duncan's multiple range test was carried out on these traits between the three defined groups obtaining significant differences between the groups for all of the traits except for the neck width in 2000 (Table 4). This classif ication differs to that carried out according to the bulb skin colour trait. In this case the lines can be grouped into four different classes: lines with brown, pale yellow, brownish yellow and brownish red skins. Most of the lines had symmetrical bulbs with one vegetative axe and white flesh ( Table 2).

Commercial production
Commercial production of the best local onion lines was slightly higher than that of the commercial varieties (with no significant differences) in 2001, and significantly higher in 2002 (Table 6). There were also great differences between the yields obtained in the different years, probably due to the different climatic conditions. The lines that had the highest yields were 'Ameixenda' -the most productive line in 2001 and 2002-and 'Baldaio' -outstanding in 2002-which clearly exceeded the productions obtained by the commercial cultivars. The yield range of the local lines was shorter than that of commercial varieties, either in 2001 or 2002, and the mean yields of these lines were higher than those of commercial ones.

Dry matter contents
The percentage of dry matter was, in general, slightly higher in the local lines, with the exception of that obtained in the variety 'Paja Virtudes', which was significantly higher than the rest of the varieties and local lines, either in 2001 or 2002 (Table 6). Among the local lines, the highest values were obtained for 'S. Xulián' and 'Betanzos', which were clearly evident in the two experiments. Either the means or the percentage ranges were similar in the two group of lines, obtaining higher levels in 2002 than in the previous experiment.

Storage quality
The storage quality, defined by Castell and Díez (2000) as the percentages of sprouted bulbs 12, 16, 20 and 24 weeks after harvest, was better in certain local lines than in the commercial varieties: 'Betanzos', 'San Xulián', 'Caldas de Reis' and 'Vilagarcía' (lines 1 and 2) had similar or even lower percentages of sprouted bulbs than 'Paja Virtudes', the commercial variety with the best storage qualities ( Table 7). The quality ranged, either in the local or in the commercial cultivars, from  lines with very low quality to those specified before which had high level quality following the descriptors of Castell and Díez (2000).

Discussion
Considering the grouping criteria used by Castell and Díez (2000) -employed to characterize local onion lines collected all over Spain-based mainly on the bulb type as well as the skin and flesh colours, the situation differs from that described in this work. Most of the Galician local lines are situated in the 4.3.2 subgroup of the red storage type (Table 5), including either transverse elliptic or transverse narrow elliptic bulb lines. Only five lines were situated in different groups: 'S. Xulián' can be considered as a Red Storage type line belonging to the 4.3.4 subgroup and 'Ribadeo', 'Oimbra', 'Outes' and 'Cea' may be included in the 4.4.1. subgroup of the conical type class. It is important to mention that four of the eighteen lines cannot be clearly included in any of the groups defined by Castell and Díez (2000), mainly because of the skin colour. Most of the northwestern local lines have a skin colour ranging from dark brown to pale yellow and should not be all included in the same subgroup. A similar situation takes place with bulb shape: the descriptors used for characterization of the Spanish local lines do not distinguish well between transverse elliptic and transverse narrow elliptic shapes (UPOV, 1999) being all considered oval bulbs. This is specially important if we consider the fact that there were no northwestern entries in the catalogue described by these authors (Castell and Díez, 2000). The redefinition of classes using more precisely defined descriptors -such as the UPOV (1999) descriptors for exampleand an increase in the number of entries, including several from the northwest and other parts of Spain, would improve the characterization.
Percentages of dry matter content are negatively correlated with percentages of sprouted bulbs (Fig. 2) -though the R 2 of the regression is relatively lowwhich is to say that storage quality, measured by the percentages of sprouted bulbs, is positively correlated with percentages of dry matter content. Mattana and Lobo (1980) reported increases in the storage quality of onion lines with higher percentages of dry matter contents as well as of lines with smaller bulbs. This usually takes place with onions that develop bulbs during long-day periods.  reported that storage losses were positively correlated with protein content and negatively correlated with ash, potassium, and dry matter content, as well as total soluble solids and nonreducing sugars. Storage quality is also reported to be negatively correlated with certain bulb morphological traits such as bulb and neck Table 6. Yields (t ha -1 ) and dry matter content (%) of the local and commercial onion (Allium cepa L.) lines studied  (Patil and Kale, 1985b;Rajapakse et al., 1992). These correlations could not be confirmed by the data presented in this paper. If further research confirms that storage quality is correlated with certain chemical or morphological traits, the measurements of these characters would be of great use in breeding programs.
From the evaluations carried out in this work it can be concluded that the 18 local lines of northwestern origin can be grouped into three different classes according to morphological traits and into six classes if we also consider the flesh and skin colour. The most productive lines had the worst storage qualities and vice versa. Some of these lines had excellent storage qualities as well as good yields and will be employed in breeding programs in the future. A. Rivera Martínez et al. / Span J Agric Res (2005) 3(1), 90-97 Table 7. Storage quality of local and commercial onion (Allium cepa L.) lines studied