Introduction
Horseradish (Armoracia rusticana G. Gaertn., B. Mey. & Scherb.) is a herbaceous perennial plant that belongs to the Brassicaceae family. It is native to southern Russia and eastern Ukraine and has been naturalized throughout most of Europe (Ball, Reference Ball, Tutin, Heywood, Burges, Moore, Valentine, Walters and Webb1964). Although wild populations of the species are no longer reported, it is still cultivated today as a crop and is appreciated for the edible features of its underground parts (Sampliner and Miller, Reference Sampliner and Miller2009). The main cultivation areas of horseradish are in northern and eastern European countries and the USA (Shehata et al., Reference Shehata, Mulwa, Babadoost, Uchanski, Norton, Skirvin, Walters and Janick2009; Filipović et al., Reference Filipović, Popović and Aćimović2015).
Horseradish roots and rhizomes have long been used as a spice in traditional dishes (Courter and Rhodes, Reference Courter and Rhodes1969; Fahey et al., Reference Fahey, Zalcmann and Talalay2001; Agneta et al., Reference Agneta, Möllers and Rivelli2013) and as a condiment due to the presence of several compounds in the plant tissues. Glucosinolates (GSLs), which are secondary metabolites of Brassicaceae and precursors of isothiocyanates (ITCs), are responsible for the characteristic pungent and spicy taste obtained from the underground parts of the horseradish plant (Agneta et al., Reference Agneta, Möllers and Rivelli2013).
Bell et al. (Reference Bell, Kitsopanou, Oloyede and Lignou2021) characterized the pungent aroma of some Brassicaceae and identified 75 volatile compounds in the headspace of Armoracia roots, including 38 previously unreported compounds and GSL hydrolysis products such as allyl isothiocyanate (allyl-ITC) and phenethyl ITC, which had the highest relative abundances. Brassicaceae vegetables are a rich source of phytochemicals with high nutritive and nutraceutical value. Plant extracts obtained from Brassicaceae have numerous bioactive properties, including antimicrobial, antioxidant, anti-inflammatory, cardioprotective, immunomodulating, and antimutagenic functions (Favela-González et al., Reference Favela-González, Hernández-Almanza and De la Fuente-Salcido2020; Peña et al., Reference Peña, Guzmán, Martínez, Mesas, Prados, Porres and Melguizo2022). Over 130 GSLs have been characterized in plants to date (Blažević et al., Reference Blažević, Montaut, Burčul, Olsen, Burow, Rollin and Agerbirk2020), and after crushing or chewing, these secondary metabolites are hydrolysed and release ITCs, which are the main reaction products formed at neutral pH in the presence of the myrosinase enzyme. The presence of GSLs, which have different profiles and contents in all Brassicaceae species (Agerbirk and Olsen, Reference Agerbirk and Olsen2012; Blažević et al., Reference Blažević, Montaut, Burčul, Olsen, Burow, Rollin and Agerbirk2020), has been shown to have a chemopreventive effect by facilitating the inhibition of mutagenic factors and cancer growth (Velasco et al., Reference Velasco, Francisco, Cartea, Watson and Preedy2010; Zhang et al., Reference Zhang, Garzotto, Davis, Mori, Stoller, Farris, Wong, Beaver, Thomas, Williams, Dashwood, Hendrix, Ho and Shannon2020; Peña et al., Reference Peña, Guzmán, Martínez, Mesas, Prados, Porres and Melguizo2022).
To date, 17 GSLs have been identified in the roots and sprouts of A. rusticana (Agneta et al., Reference Agneta, Rivelli, Ventrella, Lelario, Sarli and Bufo2012, Reference Agneta, Lelario, De Maria, Möllers, Sabino Aurelio Bufo and Rivelli2014b). More recently, Dekić et al. (Reference Dekić, Radulović, Stojanović, Randjelović, Stojanović-Radić, Najman and Stojanović2017) identified, for the first time, the presence of 5-phenylpentyl ITC in the aerial parts of the species. This compound is a probable natural hydrolysis product of a 5-phenylpentyl GSL or glucoarmoracin, as proposed by the authors. It may have interesting antimicrobial and spasmolytic effects. The therapeutic effects of Armoracia are mainly attributed to its GSL content, and the plant's medicinal uses, which have been known since antiquity (Wedelsbäck Bladh and Olsson, Reference Wedelsbäck Bladh and Olsson2011; Walters, Reference Walters2021), are now limited to traditional uses still surviving in some countries (Sampliner and Miller, Reference Sampliner and Miller2009; Papp et al., Reference Papp, Gonda, Kiss-Szikszai, Plaszkó, Lőrincz and Vasas2018).
Several studies have investigated the health effects of horseradish and reported its ability to prevent cancer proliferation, decrease plasma cholesterol and exhibit antimicrobial and antioxidant activities (Agneta et al., Reference Agneta, Möllers and Rivelli2013; Popović et al., Reference Popović, Maravić, Čikeš Čulić, Đulović, Burčul and Blažević2020; Walters, Reference Walters2021). Most reports published on the characterization of horseradish from different origins have mainly focused on its GSL content. Li and Kushad (Reference Li and Kushad2004) evaluated 27 accessions of horseradish varieties from North America and eastern Europe and reported a wide range of variability in the GSL amount in different plant tissues. Wedelsbäck Bladh et al. (Reference Wedelsbäck Bladh, Olsson and Yndgaard2013) reported the GSL content of 168 accessions from Nordic European countries, while Ciska et al. (Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017) evaluated the GSL quantity of four European landraces.
While the bio-agronomic characterization of accessions is essential for the horseradish germplasm description and classification (Agneta et al., Reference Agneta, Möllers and Rivelli2013), only a few authors have reported the agronomic and morphological traits of the investigated accessions. Katalin (Reference Katalin2012) described the morphological and agronomic features of 78 cultivated varieties, mainly of Hungarian origin. Wedelsbäck Bladh et al. (Reference Wedelsbäck Bladh, Liljeroth, Poulsen, Yndgaard and Brantestam2014) evaluated the morphological traits, level of genetic diversity and GSL content of 176 accessions.
Horseradish is a rhizomatous geophyte (Pignatti, Reference Pignatti2017) with an underground part consisting of a rhizome and roots that are often difficult to distinguish. Most authors report roots as the main exploited part of the plant (Nguyen et al., Reference Nguyen, Gonda and Vasas2013), with a primary or marketable root being distinguished from secondary roots (Walters and Wahle, Reference Walters and Wahle2010). The presence of rhizome in horseradish is responsible for the species' characteristic of being propagated by its hypogeous parts, as it rarely produces viable seeds and is commercially propagated by rhizome cuttings (Sampliner and Miller, Reference Sampliner and Miller2009). However, this characteristic also presents an obstacle to improving the crop by developing new cultivars (Walters et al., Reference Walters, Bernhardt, Joseph and Miller2016). Horseradish cultivars with some degree of variability are grown worldwide (Walters and Wahle, Reference Walters and Wahle2010). In Italy, where the species is cultivated or naturalized near vegetable gardens (Pignatti, Reference Pignatti2017), Sarli et al. (Reference Sarli, Lisi, Agneta, Grieco, Ierardi, Montemurro, Negro and Montesano2012) reported the morphological traits of 32 cultivated accessions from Basilicata in southern Italy. The traditional uses of horseradish in the Basilicata region were reported by Agneta et al. (Reference Agneta, Möllers and Rivelli2013), who later evaluated six accessions from the same region for their morphological and productive traits, GSL content and genetic diversity (Agneta et al., Reference Agneta, Möller, De Maria and Rivelli2014a).
Despite the extensive cultivation or naturalization of horseradish being reported in the northern regions of Italy (Fiori, Reference Fiori1923; Pignatti, Reference Pignatti2017) and particularly in Trentino-Alto Adige (Dalla Torre and Sarnthein, Reference Dalla Torre and Sarnthein1909; Dalla Fior, Reference Dalla Fior1969; Aeschimann et al., Reference Aeschimann, Lauber, Moser and Theurillat2004), to our knowledge, no reports on the characterization of cultivated accessions from this area have been published to date. The interest in this region stems from the presence of typical small-scale mountain farming, known locally as ‘Maso’ (Mori and Hintner, Reference Mori and Hintner2013), where farmers grow horseradish in vegetable gardens for self-consumption. The underground parts of the plant are traditionally used to prepare a sauce mainly used as a condiment for meat dishes. Farmers usually propagate horseradish themselves, maintaining and conserving their cultivated accessions. This tradition is mainly preserved by the German-speaking ethnic group living in the region and probably originated from the use of the plant, which has been known since ancient times and is widespread in the areas situated further north of the Alps (Mattioli, Reference Mattioli1568). Moreover, recently, the Trentino-Alto Adige region has been the subject of studies on locally cultivated plants (Lucchin et al., Reference Lucchin, Barcaccia and Parrini2003; Heistinger and Pistrick, Reference Heistinger and Pistrick2007), highlighting the importance of this area for the presence of underutilized plant genetic resources.
This study aims to evaluate horseradish accessions cultivated in vegetable gardens in the Trentino-Alto Adige region by describing their morphological and productive traits and quantifying the main GSLs contributing to defining the quality of the obtained product when compared to a reference cultivar. Characterizing the investigated accessions could be important in expanding the existing knowledge on the germplasm consistence of the species and identifying possible new genetic resources suitable for cultivation.
Materials and methods
Plant materials
The investigated accessions were collected from local vegetable gardens in the Trentino-Alto Adige region (see online Supplementary Fig. S1) between 2012 and 2016. Exact details regarding their sites of origin are reported in Table 1 and online Supplementary Fig. S2.
Identification code and geographical origin (municipality, province, altitude and geographic coordinates) of A. rusticana accessions investigated in this study
The cultivated variety Bagameri 93/1 was obtained from the Leibniz Institute of Plant Genetics and Crop Plant Research (IPK) in Gatersleben, Germany. During the collection of accessions from horseradish growers, ethnobotanical information was acquired through interviews using a set of open-ended questions regarding the characteristics of the interviewee, plant origin, cultivation techniques used and product uses. The collected accessions were conserved in vivo, ex situ, at the experimental farm ‘Nicolini’ of the Research Centre for Forestry and Wood of the Council for Agricultural Research and Economics in Trento, Italy (46°02′53″N, 11°08′51″E; 370 m a.s.l.), an area characterized by a temperate sub-continental climate and a hyperskeleti-rendzic leptosol soil type (IUSS, 2015). On 2 March 2016, all collected accessions were simultaneously vegetatively propagated and transplanted into a field settled at the same experimental farm, with climatic conditions and soil characteristics representative of the growing sites from which the cultivated accessions originated.
Propagation was carried out using root cuttings that were 20 cm in length and 1 cm in diameter, transplanted at distances of 120 cm between rows and 70 cm within the row. The field was prepared beforehand by creating ground beds of approximately 30 cm in height, where the cuttings were laid at a 45° angle, following the polarity of the roots, with the proximal end (the point of attachment to the mother plant's rhizome) placed on top of the ground. Six plants per accession were transplanted in single rows and cultivated in 2016, following the local cultivation practices for the species, including watering at the time of transplantation and subsequently as needed and manual weed control (online Supplementary Fig. S3). Fertilization was carried out 1 month before transplantation by applying 50 kg/ha of N as ammonium nitrate, 50 kg/ha of K, 30 kg/ha of S as potassium sulphate and 50 kg/ha of P as mineral superphosphate on the entire area. No further manuring was done during cultivation.
Morphological characterization
Morphological characteristics were recorded following the UPOV guidelines for horseradish TG/191/2 ‘Guidelines for the Conduct of Tests for Distinctness, Uniformity and Stability of Horseradish (Armoracia rusticana Gaertn. Mey. et Scherb.)’ (UPOV, 2001), using the complete set of 29 characters. Characters related to aerial parts (UPOV descriptor nos. 1–17), except for the height of stems at full flowering, were recorded in the first year of cultivation on all six plants per accession since no missing plants were recorded, on the fourth fully developed leaf and the first incised leaf, following the UPOV guidelines. Characters related to the underground parts were recorded on three plants per accession during the harvest, conducted manually between 8 and 16 March 2017, at the beginning of the vegetative growth phase in the second year after transplant. The harvest was conducted during this period to follow the local practice linked to the traditional use of the product during the Easter festival.
The height of stems at full flowering (UPOV descriptor no. 29) was recorded during the second year after transplant for the plants left after harvest. Some of the quantitative morphological traits included in the UPOV descriptor list, related to the dimension of the rhizome (UPOV descriptors nos. 20–22), were measured instead of being described by classes. The total fresh weight of all underground parts of the plant, which is not included in the UPOV guidelines, was also measured as an additional character. The quantitative traits related to the dimension of the leaf, including length, width, their ratio and those of the petiole (respectively, UPOV descriptor nos. 2, 3, 4, 13 and 14), were measured and transformed into classes by following the method described by Sarli et al. (Reference Sarli, Lisi, Agneta, Grieco, Ierardi, Montemurro, Negro and Montesano2012), which established the limit values of each class so that the arithmetic mean was the central value, and each class included at least 10% of cases concerning the total.
Sample preparation and GSL analysis
To determine the GSL content, samples of rhizomes from three harvested plants per accession were prepared as follows: immediately after recording the morphological traits of the underground parts, the rhizomes were cleaned with distilled water, dried with a paper towel and a central portion of their tissue was cut and separated from the rest. The tissue was quickly put in liquid nitrogen, then freeze-dried and stored in sealed paper bags at room temperature until analysis. The residual parts of the rhizome were then dried in a thermostatic ventilated oven at 105°C until steady weight to determine dry matter content. Two hundred milligrams of finely powdered freeze-dried samples were extracted with 70% boiling ethanol. The GSLs were desulphated and analysed by HPLC-UV in compliance with the EU official ISO 9167-1:2019 method, with minor modifications (Pagnotta et al., Reference Pagnotta, Agerbirk, Olsen, Ugolini, Cinti and Lazzeri2017).
Desulphated-GSLs were detected by monitoring their absorbance at 229 nm and identified based on the retention times and UV spectra of desulpho-GSL standards available in our library (Wathelet et al., Reference Wathelet, Iori, Leoni, Rollin, Quinsac and Palmieri2004; Galletti et al., Reference Galletti, Bagatta, Branca, Argento, De Nicola, Cianchetta, Iori and Ninfali2015; Müller et al., Reference Müller, Schulz, Pagnotta, Ugolini, Yang, Matthe, Lazzeri and Agerbirk2018). Their content was quantified using isolated epi-progoitrin [(2S)-2-hydroxy-but-3-enyl-GSL] as an internal standard (250 μl of a solution 0.031 μmol/μl). The epi-progoitrin standard was isolated from Crambe abyssinica Hochs t. ex R. E. Fries (Daubos et al., Reference Daubos, Grumel, Iori, Leoni, Palmieri and Rollin1998) at a purity of 96% on a weight basis and stored at −20°C until required. Sulphatase from Helix pomatia (Sigma Aldrich, USA) was purified according to ISO 9167-1:2019 and stored at −20°C until required.
Statistical analysis
Quantitative measured traits were subjected to one-way analysis of variance, with the percentage data previously transformed into angular values (arcsin/√x), and statistical mean comparisons were calculated following Tukey's (honestly significant difference (HSD)) test. Agglomerative hierarchical clustering (AHC) was performed using the unweighted pair group medium average fusion method and the Euclidean distances as criteria. These statistical analyses, principal component analysis (PCA) and correlation analysis were conducted using the XLSTAT software package (Addinsoft, New York, USA).
Results
Ethnobotany and uses
Among the 11 interviewees who grew horseradish, six were male. Only one was under 30 years old, while eight were over 50 years old, and two were over 70 years old. Five of them were professional farmers who owned the typical local mountain farm, called ‘Maso’, while two were professionals in the food business. They all reported cultivating their horseradish accessions since they received the plant from their forefathers. They used the obtained product for self-consumption and grew plants in a limited part of their home garden. Only one interviewee cultivated it on a larger surface (around 100 m2) to sell locally. The edible parts of the plants consumed in line with tradition were rhizomes and roots, which were used fresh and ground to prepare a sauce by mixing it with various ingredients, including apples (five interviewees), cream (four interviewees), vinegar, bread (four interviewees), salt (two interviewees), sugar, pepper and cranberry jam (one interviewee).
Two interviewees reported using edible flowers, and one grower mentioned that young leaves were eaten according to tradition. Both flowers and leaves were used for preparing salads. All interviewees stated that spring (March or April) is the optimal period for harvesting horseradish underground parts, consistent with the traditional use of the obtained sauce. Five recalled the custom of combining the sauce with boiled ham, a traditional dish prepared during the Easter period. In some villages of the province of Bolzano, the tradition of bringing the foods, including the horseradish root, consumed during the Easter lunch to the Easter Mass for blessing is still alive (online Supplementary Fig. S4). Three respondents highlighted that roots harvested in autumn are too stringy to prepare a good sauce. One interviewee recalled the medicinal use of horseradish underground parts, applied fresh externally to cure contusions. Another reported the plant protection features of horseradish. It was cultivated under fruit trees to prevent plant pathogens, possibly due to the well-known nematicidal effect of GSLs in the plant's underground parts (Van Dam et al., Reference Van Dam, Tytga and Kirkegaard2009).
Morphological characterization
Great variability in qualitative traits was observed among accessions, except for the leaf's absent or very weak glaucosity, the mostly smooth surface texture of the rhizome, generally whitish internal colour, and the presence of more than three shoots on the crown of the rhizome in all accessions. Regarding the leaf, the predominant shape was elliptic (66.7%). Most accessions were characterized by medium-sized leaf blades (medium length 58.3%, medium width 75.0%) and petioles (medium length 66.7%, medium width 50%), medium glossiness of the upper side (66.7%), medium twisting of the tip (58.3%) and medium serration (66.7%) of the leaf blade. The undulation of the margin was strong in half of the investigated accessions, and the colour of the midrib was whitish in most of them (66.7%), while just two accessions presented anthocyanin colouration at the base of the petiole. The incised leaf's appearance was mostly early (58.3%), and incisions, when present, were weak (25%). As for the rhizome, it was obtriangular (75%), straight (50%) and had weak colouration of the flesh (50%) in the majority of accessions, with a medium density of side roots both at the upper third (66.7%) and at the base (66.7%) (online Supplementary Tables S1–S3). The photographs of the investigated accessions are included in online Supplementary Figs S5–16.
Productive traits
The measured biometric traits of the underground parts are given in Table 2.
Morphological quantitative traits of underground parts of A. rusticana accessions
R.TD, rhizome top diameter; R.BD, rhizome basal diameter; R.FW, rhizome fresh weight; T.FW, total rhizome and root f.w.
Mean values (N = 3) ± SD followed by different letters are significantly different at P < 0.01 according to Tukey's (HSD) test.
Abbreviated codes of accessions are indicated in Table 1.
On average, the top diameter of the rhizomes of the accessions was 9.9 cm, and the basal diameter was 6.2 cm. The top diameters of ANT, LAI and NOV and the basal diameters of BAD and BAG were larger than those of ALD and BED accessions. The average fresh weight of rhizomes was 521.6 g, with BAD having a higher weight (721.6 g) than BED accession (188.1 g). The percentage of the dry matter content of rhizomes (data not shown) was 25.9% on average, with higher values in BAD (28.2%) compared to ALD, ANT and BED. The mean fresh weight of the underground parts, including rhizomes and roots, was 1485.3 g, with higher weights in BAD, BAG and LAI than ALD and BED accessions. BAD, LAI and the reference cultivar BAG showed better productive characteristics, such as larger rhizome diameters and higher fresh weights of rhizomes and whole underground parts, compared to other accessions.
Qualitative traits
Five major GSLs were identified and quantified in the investigated accessions: glucoiberin (GIB), sinigrin (SIN), gluconapin (GNA), glucobrassicin (GBS) and gluconasturtiin (GST). Their contents are presented in Table 3, and a representative HPLC-UV profile of GSLs is shown in online Supplementary Fig. S17.
GSL content in rhizome of A. rusticana accessions
GIB, glucoiberin; SIN, sinigrin; GNA, gluconapin; GBS, glucobrassicin; GST, gluconasturtiin.
Reported values expressed as μmol∙g−1 DW.
Mean values (N = 3) ± SD followed by different letters are significantly different at P < 0.01 according to Tukey's (HSD) test.
The total amount of identified GSLs ranged from 133.5 to 79.5 μmol∙g−1 dry weight (DW) in NOV and BED accessions, respectively, with an average content of 119.9 μmol∙g−1 DW. SIN was the predominant GSL in all accessions, ranging from 69.5 in BED to 123.9 μmol∙g−1 in SCE accession, with an average content of 110.0 μmol∙g−1 DW, followed by GST, GNA, GBS and GIB. Highly significant differences among investigated accessions resulted in total and single GSL content (P < 0.01). Eleven out of 12 accessions produced more than 100 μmol∙g−1 DW of total GSL, and values higher than 120 μmol∙g−1 DW were found in eight accessions (ANT, BAD, BUR, LAI, NOV, PUS, SCE and ZIV) which were significantly different (P < 0.01) from BED, except for PUS. The cultivar BAG showed no difference from the others for the content of total GSL, which was 113.97 μmol∙g−1 DW. The SIN content of BED was significantly lower (P < 0.01) compared to half of the investigated accessions (BAD, BUR, LAI, NOV, SCE and ZIV). Nine accessions produced more than 100 μmol∙g−1 DW of SIN. In particular, in BAD, BUR, NOV, SCE and ZIV, more than 120 μmol∙g−1 DW were extracted. GST content ranged from 3.9 in BAD to 8.2 μmol∙g−1 DW produced in ALD, accounting for an average of 5.1% of the total GSL. The remaining GSLs were present in amounts lower than 1.5 μmol∙g−1 DW and with percentages lower than 1.2% of the identified GSLs.
Multivariate analysis
Measured biometric traits of underground parts and rhizome GSL content are the most important traits in defining the attractive quantitative and qualitative characteristics of the obtainable commercial and edible product of the investigated accessions (Agneta et al., Reference Agneta, Möller, De Maria and Rivelli2014a). A PCA was carried out using the mean values of these parameters per accession to explore and summarize the existing variability among them for these traits. A correlation matrix was obtained first among the considered traits (online Supplementary Table S4). As expected, a highly significant positive correlation was observed among the different measured biometric characters of the underground parts of the plant. Additionally, SIN and total GSL content were positively correlated with the evaluated biometric traits, except for the lack of correlation between total GSLs and the rhizome basal diameter. Correlation indexes between biometric traits and SIN ranged from 0.624 to 0.762, while those between the same traits and total GSL ranged from 0.709 to 0.745.
Furthermore, GST content was negatively correlated with three rhizome biometric traits: the basal diameter (−0.798), the fresh weight of the rhizome (−0.709) and total underground parts (−0.614). SIN was positively correlated with total GSLs, with a correlation index of 0.987, which was highly expected since SIN was the main GSL in the rhizomes. A positive correlation was also found between SIN and GNA (0.587) and, consequently, between GNA and Total GSLs (0.609). SIN and GNA belong to the Met-derived GSL group, for which studies on the Arabidopsis and Brassica genus (Sønderby et al., Reference Sønderby, Geu-Flores and Halkier2010; Liu et al., Reference Liu, Liu, Yang, Tong, Edwards, Parkin, Zhao, Ma, Yu, Huang, Wang, Wang, Lu, Fang, Bancroft, Yang, Hu, Wang, Yue, Li, Yang, Wu, Zhou, Wang, King, Pires, Lu, Wu, Sampath, Wang, Guo, Pan, Yang, Min, Zhang, Jin, Li, Belcram, Tu, Guan, Qi, Du, Li, Jiang, Batley, Sharpe, Park, Ruperao, Cheng, Waminal, Huang, Dong, Wang, Li, Hu, Zhuang, Huang, Huang, Shi, Mei, Liu, Lee, Wang, Jin, Li, Li, Zhang, Xiao, Zhou, Liu, Liu, Qin, Tang, Liu, Wang, Zhang, Lee, Kim, Denoeud, Xu, Liang, Hua, Wang, Wang, Chalhoub and Paterson2014) revealed that the two GSLs share many steps of the biosynthetic pathway starting from the same precursor amino acid, methionine. Considering that the final step of synthesizing these GSLs is catalyzed by the same gene (AOP2), even if the metabolic way splits into two different pathways, we could hypothesize that the content of SIN and GNA in A. rusticana accessions may be correlated. The results of the PCA are reported in Table 4.
The first three principal components (PCs), with an eigenvalue higher than 1, explained 87.13% of the total variability. PC1 was highly correlated with SIN, total GSL content (T.GSL), and all the evaluated biometric traits. PC2 was highly correlated with GBS and, to a lesser extent, with GNA content, while PC3 was positively correlated with GST and negatively with GIB content (see Table 4). In the score plot obtained for the first two PCA components (Fig. 1), most of the investigated accessions positively correlated with PC1 formed a main cluster, almost entirely included in the right quadrants, together with the reference cultivar BAG. Accessions BED and ALD, negatively correlated with PC1, formed a separate cluster.
Score plot for the 12 evaluated A. rusticana accessions (abbreviated codes are indicated in Table 1).
To better understand the existing similarities among the accessions for the considered traits and to more accurately identify any clusters of accessions with similar traits, AHC was carried out using the unweighted pair group method with arithmetic mean (UPGMA) fusion method and the Euclidean distances. The accessions are grouped into three main clusters in the resulting dendrogram (Fig. 2).
UPGMA dendrogram based on the relative scores for the three extracted PCs (abbreviated codes are indicated in Table 1).
The first cluster, comprising accessions ALD and BED, was characterized by lower values of productive traits such as the dimensions and fresh weight of rhizome and whole underground parts, lower amounts of GIB, SIN, GNA and total GSL, and the highest amount of GST. The second cluster, composed of seven cultivated accessions and the reference cultivar BAG, was characterized by the highest values of productive traits, particularly regarding the fresh weight of rhizome and whole underground parts and the highest amounts of SIN and total GSL. The third cluster grouped accessions LAI and PUS and was characterized by the highest amounts of GIB, GNA and GBS, in addition to sharing with cluster 2 high values of rhizome and underground part fresh weight, as well as high amounts of SIN and total GSL content.
Discussion
A wide range of variability was observed for the considered qualitative morphological traits. The expression level in BAG accession was identical to those reported in the UPOV guidelines for the same reference cultivar, except for the number of rhizome shoots. It was higher than three in this study and lower than three in the UPOV guidelines. In a study focused on the morphological characterization of A. rusticana accessions from the Basilicata region (Sarli et al., Reference Sarli, Lisi, Agneta, Grieco, Ierardi, Montemurro, Negro and Montesano2012), the authors reported the absence of glossiness in the upper side of the leaf, the green colour of the midrib and the absence of anthocyanin colouration at the base of the petiole in all the accessions. However, a certain variability was observed for these characters in our study.
The rhizome biometric data recorded in this study agree with, or in some cases, are higher than, those reported in the literature. However, a comparison was not always easy due to the different criteria used to differentiate the underground parts of the plant. For this purpose, we referred to the UPOV guidelines (UPOV, 2001), distinguishing a central rhizome from side roots. This differs from other authors who distinguished a main root (Rivelli et al., Reference Rivelli, Caruso, De Maria and Galgano2017) or a marketable root (Agneta et al., Reference Agneta, Möller, De Maria and Rivelli2014a) from the others.
The rhizome diameters of the accessions investigated in this study comply with the Standards for Grades of Horseradish Roots of the US Department of Agriculture, for which the minimum diameter of marketable horseradish roots varies from 1.27 to 3.8 cm, depending on the commercial grade (USDA, 2016). The fresh weight of the rhizomes of the accessions fell within and sometimes exceeded the range of 150 and 600 g as indicated by Achleitner and Kaufmann (Reference Achleitner, Kaufmann, Gartenbau and Hoppe2013) for a commercial standard. In a study investigating the chemical profile of a horseradish accession (Agneta et al., Reference Agneta, Rivelli, Ventrella, Lelario, Sarli and Bufo2012), smaller diameters of the primary root, both at the top (6 cm) and at the base (2 cm), and lower total fresh root weight (1056 g) were reported compared with our results. However, the same author observed a marketable root weight (733 g) comparable to the highest rhizome weight measured in this study. In Agneta et al. (Reference Agneta, Möller, De Maria and Rivelli2014a), a study on the evaluation of different accessions from the Basilicata region, the marketable root top diameter values varied from 2.6 to 15.3 cm, a wider range compared to our results. However, the basal diameter, with values from 1.8 to 3.4 cm, had a smaller range than our data. The same authors reported fresh weights of marketable roots ranging from 78 to 732 g and total roots from 121 to 1063 g. These values were lower than most obtained in the present study in both cases. Rivelli et al. (Reference Rivelli, Caruso, De Maria and Galgano2017), in a similar study focused on two accessions from the same region in Italy, reported the fresh weight of the main root ranging from 189 to 293 g and that of the total roots from 323 to 424 g. In both cases, these values were lower than those reported in this study.
Regarding the dry matter content, Agneta et al. (Reference Agneta, Rivelli, Ventrella, Lelario, Sarli and Bufo2012) reported a percentage of 37, ranging from 24.9 to 36.4% in a subsequent study (Agneta et al., Reference Agneta, Möller, De Maria and Rivelli2014a). In both cases, the dry matter content was higher than our results. The higher weights and larger dimensions of the rhizomes found in this study, compared to those reported in previous papers, could be due to many factors, such as different environmental conditions and cultivation practices.
Regarding cultivation practices, the supply of fertilizers adopted, nitrogen in particular, may have influenced the development of the plants. However, in a study focused on the effect of fertilization on horseradish cultivation, Rivelli et al. (Reference Rivelli, Lelario, Agneta, Möllers and De Maria2016) reported that plants cultivated with an N supply of 100 kg∙ha−1, double the dose applied in this study, had a rhizome fresh weight of 824 g and a total root weight per plant of 1042 g. These values were slightly higher and lower compared to our results. The dose of fertilizers used in our study was very low compared to those reported in the literature for A. rusticana (Walters and Wahle, Reference Walters and Wahle2010; Filipović et al., Reference Filipović, Popović and Aćimović2015; Ciska et al., Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017).
The importance of environmental and climatic conditions for the cultivation of horseradish has been reported by various authors (Walters and Wahle, Reference Walters and Wahle2010; Filipović et al., Reference Filipović, Popović and Aćimović2015). Agneta et al. (Reference Agneta, Rivelli, Ventrella, Lelario, Sarli and Bufo2012, Reference Agneta, Möller, De Maria and Rivelli2014a) and Rivelli et al. (Reference Rivelli, Lelario, Agneta, Möllers and De Maria2016, Reference Rivelli, Caruso, De Maria and Galgano2017) conducted their experimental trials at cultivation sites situated four degrees of latitude south of the site reported in this study and characterized by a different climate. Rivelli et al. (Reference Rivelli, Lelario, Agneta, Möllers and De Maria2016) reported a long-term annual average rainfall of 546 mm and mean monthly temperatures during the experimental period ranging from 8 to 26°C, whereas the site of cultivation in this study recorded a mean annual average rainfall of 1127.8 mm over the long term period (1979–2017), and minimum and maximum monthly temperatures of 0.9°C (January 2017) and 23.3°C (July 2016), respectively, throughout the experimental season. The different climatic conditions of the growing sites may have influenced the plants' development, including the underground parts and the dry matter content. The latter parameter could possibly cause the higher fresh weight of the underground parts obtained in this study.
The results reported in this study regarding the productive traits of the accessions refer to soil and climatic conditions similar to the vegetable gardens where the plants are locally cultivated. The genetic background of the cultivated accessions was an important factor in this study and probably played a primary role, as it generally does, in affecting the phenotypic variation of plants and the development of commercially important traits described in this study as the dimensions of the rhizome.
Regarding the qualitative traits of underground parts, the detected GSLs are consistent with A. rusticana GSLs found in previous studies (Li and Kushad, Reference Li and Kushad2004; Agneta et al., Reference Agneta, Rivelli, Ventrella, Lelario, Sarli and Bufo2012, Reference Agneta, Lelario, De Maria, Möllers, Sabino Aurelio Bufo and Rivelli2014b; Wedelsbäck Bladh et al., Reference Wedelsbäck Bladh, Olsson and Yndgaard2013; Ciska et al., Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017). Among the investigated horseradish accessions, SIN was the predominant GSL. GST, which is typically the major GSL in Brassicaceae roots and is interesting for its toxic effect on soil-borne organisms due to its hydrolysis products (Van Dam et al., Reference Van Dam, Tytga and Kirkegaard2009), was confirmed to be the second most abundant GSL in the roots of A. rusticana. Several studies have shown that the total GSL content can vary significantly in A. rusticana. For example, among 27 horseradish accessions, mainly from eastern Europe and North America, the total GSLs ranged from 2 to 296 μmol∙g−1 DW, with an average content of 81.0 μmol∙g−1 DW, and eight accessions had a total content greater than 150 μmol∙g−1 DW (Li and Kushad, Reference Li and Kushad2004). In Ciska et al. (Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017), total GSL contents from 70 to 120 μmol∙g−1 DW were found in different landraces from central and northern Europe. However, Wedelsbäck Bladh et al. (Reference Wedelsbäck Bladh, Olsson and Yndgaard2013) reported a total GSL content lower than 50–60 μmol∙g−1 DW in a screening of 168 Nordic accessions. In Agneta et al. (Reference Agneta, Möller, De Maria and Rivelli2014a), six local accessions collected in the Basilicata region produced significantly different total GSL contents, ranging from 1.73 to 37.7 μmol∙g−1 DW, a narrower range with lower values compared to many of those found in previous studies and our investigation.
According to the reports above, SIN accounts for over 80% of the root total GSL content, and its variability depends on SIN content (Li and Kushad, Reference Li and Kushad2004; Wedelsbäck Bladh et al., Reference Wedelsbäck Bladh, Olsson and Yndgaard2013; Agneta et al., Reference Agneta, Möller, De Maria and Rivelli2014a). In our samples, SIN accounted for an average of 91.5% of the identified GSLs, with a less variable range compared to that (53–87%) reported in Agneta et al. (Reference Agneta, Möller, De Maria and Rivelli2014a) and similar to the percentage values shown in Wedelsbäck Bladh et al. (Reference Wedelsbäck Bladh, Olsson and Yndgaard2013). The SIN content of our accessions was higher than half of the SIN values reported by Li and Kushad (Reference Li and Kushad2004) and higher than most values reported by Ciska et al. (Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017), Wedelsbäck Bladh et al. (Reference Wedelsbäck Bladh, Olsson and Yndgaard2013) and Agneta et al. (Reference Agneta, Möller, De Maria and Rivelli2014a, Reference Agneta, Lelario, De Maria, Möllers, Sabino Aurelio Bufo and Rivelli2014b).
The range of GST concentrations reported here was similar to the values found in Ciska et al. (Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017) and Wedelsbäck Bladh et al. (Reference Wedelsbäck Bladh, Olsson and Yndgaard2013) and higher than the values indicated in Agneta et al. (Reference Agneta, Möller, De Maria and Rivelli2014a, Reference Agneta, Lelario, De Maria, Möllers, Sabino Aurelio Bufo and Rivelli2014b). The range of GNA (1.3–1.5 μmol∙g−1 DW), GBS and GIB content (0.8–1.7 and 0.4–1.1 μmol∙g−1 DW, respectively) found in the investigated accessions was similar to the values reported in previous studies. In many species of the Brassicales order, to which Brassicaceae belongs, GSLs are usually present with different contents and profiles depending on the analysed tissue. Typically, four to five GSLs are predominant, while many others with different structures are present in traces in the same species (Verkerk et al., Reference Verkerk, Schreiner, Krumbein, Ciska, Holst, Rowland, De Schrijver, Hansen, Gerhäuser, Mithen and Dekker2009). Among the seventeen GSLs identified in A. rusticana, the five major GSLs found in our samples were included. Our data confirmed that A. rusticana has a very rich GSL profile compared to other genera of Brassicaceae (Fahey et al., Reference Fahey, Zalcmann and Talalay2001; Verkerk et al., Reference Verkerk, Schreiner, Krumbein, Ciska, Holst, Rowland, De Schrijver, Hansen, Gerhäuser, Mithen and Dekker2009).
Unlike other species, the roots of horseradish do not have a more diversified GSL profile compared to other plant tissues, where SIN, GIB and GNA, belonging to the aliphatic GSL group, predominate (about 95%), with SIN accounting for around 90% of the total GSL content. GBS and GST, which belong to the indolic and phenethyl GSL structural groups, are present in much lower concentrations, about 2.6 and 2.4% of the total GSLs, according to Agneta et al. (Reference Agneta, Lelario, De Maria, Möllers, Sabino Aurelio Bufo and Rivelli2014b) and the results obtained in this study. Additionally, horseradish has a wide range of total GSL content, reaching some of the highest concentrations measured in root vegetables (Li and Kushad, Reference Li and Kushad2004; Verkerk et al., Reference Verkerk, Schreiner, Krumbein, Ciska, Holst, Rowland, De Schrijver, Hansen, Gerhäuser, Mithen and Dekker2009).
It is well-known that photoperiod, temperature, soil type and biotic and abiotic factors can affect GSL accumulation (Björkman et al., Reference Björkman, Klingen, Birch, Bones, Bruce, Johansen, Meadow, Mølmann, Seljåsen, Smart and Stewart2011; Burow, Reference Burow and Kopriva2016). In many species of the Brassica genus, it has been shown that GSL biosynthesis is subject to fluctuations in temperature, with an increase in GSL content induced at temperatures less than or around 10°C and at stress temperatures above 30°C (Engelen-Eigles et al., Reference Engelen-Eigles, Holden, Cohen and Gardner2006; Schonhof et al., Reference Schonhof, Kläring, Krumbein, Clauen and Schreiner2007; Verkerk et al., Reference Verkerk, Schreiner, Krumbein, Ciska, Holst, Rowland, De Schrijver, Hansen, Gerhäuser, Mithen and Dekker2009). Rao et al. (Reference Rao, Chen, Wang, Zhu, Yang and Zhu2021) reported that in Brassica rapa seedlings grown under controlled conditions, the content of aliphatic GSLs could increase during short-term high-temperature stress, while indole GSLs increase during long-term stress or stress recovery. In the roots of four horseradish landraces from central-north Europe (Ciska et al., Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017), SIN content fluctuated according to the harvest time of the year, reaching the highest concentration in October, with SIN values ranging between 70 and 116 μmol∙g−1 DW, higher than the SIN contents in horseradish from Basilicata (Agneta et al., Reference Agneta, Möller, De Maria and Rivelli2014a, Reference Agneta, Lelario, De Maria, Möllers, Sabino Aurelio Bufo and Rivelli2014b) and some accessions analysed by Li and Kushad (Reference Li and Kushad2004).
Although we found significant differences between the lowest and highest total GSL and SIN contents in our samples, which varied almost twofold, the high SIN concentrations detected in most horseradish accessions are consistent with the results reported by Ciska et al. (Reference Ciska, Horbowicz, Rogowska, Kosson, Drabińska and Honke2017). This could be explained by the climate characterizing the cultivation site, which is likely similar to that of continental European areas, and by the genetic background of the investigated accessions, considering that they all originated from the Trentino-Alto Adige region in northern Italy. This combined effect of environment and genotype might partially explain the difference in root GSL content between the horseradish accessions evaluated in this study and those analysed in Agneta et al. (Reference Agneta, Möller, De Maria and Rivelli2014a), which originated from Basilicata, a region in southern Italy characterized by different pedoclimatic traits. Regarding the effect of sulphur fertilization on sulphur-containing compounds, since GSLs can account for up to 30% of the total sulphur content of plant tissues, sulphur supply in cultivated plants of Brassicaceae species can lead to an increase in GSL content (Falk et al., Reference Falk, Tokuhisa and Gershenzon2007). Thus, besides the genotype effect, GSL root accumulation in the underground parts of A. rusticana can also be modulated by fertilization, particularly affecting SIN content and its degradation product, allyl-ITC, which mainly gives a bitter, pungent taste to the roots and contributes to the valued flavour of traditional dishes (Agneta et al., Reference Agneta, Möllers and Rivelli2013; Rivelli et al., Reference Rivelli, Lelario, Agneta, Möllers and De Maria2016).
In Rivelli et al. (Reference Rivelli, Lelario, Agneta, Möllers and De Maria2016), a major root yield and an increasing GSL concentration concerning the control were detected in A. rusticana plants fertilized with nitrogen and sulphur and harvested during vegetative regrowth in early spring. Consistent with Rivelli et al. (Reference Rivelli, Lelario, Agneta, Möllers and De Maria2016) and, above all, with the local cultivation practices of the sites where the investigated accessions originated, the underground parts of plants in this study were harvested during early spring vegetative regrowth. Furthermore, the soil had been fertilized before cultivation according to local practices. This fact may raise the question of whether the biosynthesis and accumulation of GSLs were favoured in the first stages of the plant's vegetative cycle. However, it seems unlikely that such fertilization could have influenced the GSL content of the accessions investigated. It must be taken into account that sulphur in the soil is easily washed away, a low sulphur dose (30 kg∙ha−1 of S as potassium sulphate) was used in this study, and a long time had passed since fertilization to harvest, which was over 1 year. These three factors probably reduced the sulphur effect on the GSL content of the investigated accessions.
Regardless of any unverifiable hypotheses, the fact remains that the accessions investigated in this study were all cultivated under the same experimental conditions, and the differences found in their GSL content are thus effective. Plant genetic background plays a primary role in affecting SIN content (Magrath et al., Reference Magrath, Herron, Giamoustaris and Mithen1993). Studies examining possible effects of the interaction of genotype and environment on GSL content have confirmed that the genotype has a greater influence than climate factors, mainly on aliphatic GSL synthesis (Farnham et al., Reference Farnham, Wilson, Stephenson and Fahey2004; Verkerk et al., Reference Verkerk, Schreiner, Krumbein, Ciska, Holst, Rowland, De Schrijver, Hansen, Gerhäuser, Mithen and Dekker2009). In the present study, the variability in quantified GSLs among investigated accessions cultivated under the same experimental and environmental conditions could be attributed to their different genetic profiles.
Finally, some of the results of multivariate analysis, particularly the significant positive correlation between SIN, GSL content, and some of the evaluated biometric traits of underground parts, are partly in contrast with the results of Agneta et al. (Reference Agneta, Möller, De Maria and Rivelli2014a), who reported that higher GSL concentrations were not always associated with the best root yield traits in six accessions from southern Italy. In our study, the positive correlation between SIN and total GSL content and the main biometric traits of underground parts suggests that the quality and yield of A. rusticana's edible product can be correlated.
Conclusions
The accessions examined in this study demonstrated high variability in morphological qualitative and quantitative traits, encompassing both aerial and underground parts. Most of them exhibited biometric traits indicative of productive features, such as rhizome diameters and weights, with values comparable to the reference cultivar and, in some instances, surpassing those found in the existing literature. High variability was also observed in the case of GSL content, where the mean relative percentages aligned with those reported in the literature, while the absolute amounts of GSL often exceeded those documented in prior studies.
GSL content correlated with the rhizome's biometric traits, highlighting their significance in evaluating intraspecific diversity in horseradish. Taking these characteristics into account, the agronomic features of most locally cultivated plants proved to be comparable to those of the reference cultivar. Two types cultivated in the study area, specifically those originating from Laion and Pusteria, distinguished themselves from the others due to their intriguing qualitative and quantitative traits.
By exhibiting notable productive and qualitative features, the accessions cultivated in the Trentino-Alto Adige region serve as a valuable source of germplasm for the cultivation of this species.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262123000254
Acknowledgements
The Ministry of Agriculture, Food and Forestry Policies supported this work under the ‘Implementation of the FAO International Treaty on Plant Genetic Resources for Food and Agriculture’ project. This project aims to facilitate research and experimentation in collecting, characterizing and evaluating plant genetic resources. The authors thank the horseradish growers in Trentino-Alto Adige for generously providing propagation materials from their cultivated accessions. We also thank Heinrich Abraham of the Laimburg Research Centre and the Südtiroler Bäuerinnenorganisation in Bolzano, Italy, for offering valuable information about the study area and granting permission to publish the image of the Easter basket (available in online Supplementary Fig. S4). Furthermore, we appreciate the Leibniz Institute of Plant Genetics and Crop Plant Research (IPK) in Gatersleben, Germany, for supplying the Bagameri 93/1 reference cultivar. Lastly, the authors express their gratitude to Katharine Ann Gethin for her meticulous revision of the text in English.
