Introduction
The Rubus genus is widely distributed around the world and comprises 12 subgenera; despite the high diversity of this genus, only a few species including several blackberries, have been domesticated (Jennings et al., Reference Jennings, Daubeny and Moore1991; Wang et al., Reference Wang, Wang, Chen, Zhang, Tang, Luo and Liu2015). In Ecuador, more than 5000 hectares of blackberries are grown, with the species Rubus glaucus being the predominant species; although it is unknown if other Rubus species are widely included in large-scale cultivations (Alwang et al., Reference Alwang, Barrera, Andrango, Dominguez, Martinez, Escudero and Montufar2019). In 1990, Ballington et al. (Reference Ballington, Luteyn, Thompson, Romoleroux and Castillo1993) completed an expedition to the Ecuadorian Andes to collect Rubus germplasm. In this prospection species that could contribute valuable genes such as R. macrocarpus, R. roseus, R. adenothallus and R. robustus were collected.
Studies on morphoagronomic characteristics conducted in the highlands of Ecuador in R. glaucus revealed diversity in terms of plant growth, shape and type of the fruit; presence or absence of thorns and vigour of the plants among other traits (Carrillo-Perdomo et al., Reference Carrillo-Perdomo, Aller, Cruz-Quintana, Giampieri and Alvarez-Suarez2015). Genetic diversity of Rubus species has been determined using several DNA molecular markers, including random-amplified polymorphic DNA (RAPDs), simple sequence repeats (SSRs), amplified fragment length polymorphisms (AFLPs) and, more recently, next generation sequencing technologies (Lindqvist-Kreuze et al., Reference Lindqvist-Kreuze, Koponen and Valkonen2003; Weber, Reference Weber2003; Ward et al., Reference Ward, Bhangoo, Fernández-Fernández, Moore, Swanson, Viola, Velasco, Bassil, Weber and Sargent2013; Lee et al., Reference Lee, Lee, Kang, Lee, Raveendar, Shin, Cho and Ma2016; VanBuren et al., Reference VanBuren, Bryant, Bushakra, Vining, Edger, Rowley, Priest, Michael, Lyons and Filichkin2016). Amsellem et al. (Reference Amsellem, Noyer and Hossaert-McKey2001) used AFLPs to compare R. alceifolius growing natively with plants from the same species that were introduced in a different area. More genetic variation was identified within plants growing in their native range, with diversity in non-native ranges being dependent on distance from the origin. Marulanda et al. (Reference Marulanda, Lopez and Uribe2012) also used AFLP and SSR markers to characterize genetic diversity of wild and cultivated genotypes of Rubus spp. collected from the central Andes in Colombia. Molecular data revealed high genetic variability among and within Rubus species, suggesting that sexual reproduction could be important in conserving the genetic diversity in R. glaucus. This study reported high similarity indexes among species of different species, such as R. urticifolius and R. glaucus, indicating a hybrid origin or a common predecessor of such species, as reported by Kollmann et al. (Reference Kollmann, Steinger and Roy2000) in pollination studies with R. armeniacus. Marulanda et al. (Reference Marulanda, Lopez and Uribe2012) used SSR markers to characterize 44 Colombian R. glaucus, including thorny and thornless genotypes, revealing a genetic differentiation between both phenotypes. Although Andean blackberry is an important fruit crop in Ecuador, little is known about its genetic diversity or the use of other Rubus species in the actual production system. We aimed to screen the genetic diversity in Andean blackberry cultivars and related berries species from the central Andes of Ecuador. For this purpose, a set of cultivated and wild blackberry accessions was screened with different arbitrary molecular markers.
Materials and methods
Biological material
Plant samples from INIAP collections were used for this study. In total, 106 accessions were sampled. Cultivated samples were identified as R. glaucus and one as Olallieberry cultivar, which is the marketing name for the Olallie blackberry released by USDA and ARS. Wild samples were identified as R. urticifolius, R. niveus, R. loxensis, R. lanciniatus and R. boliviensis. Accessions were taxonomically identified at the National Herbarium (INABIO) from Quito, Ecuador.
DNA extraction and molecular analyses
Leaf tissues were collected from actively growing shoots and were conserved on silica gel until DNA extraction. Genomic DNA was extracted using the PureLink Genomic Plant DNA Purification kit (Life Technologies cat. no. K1830-01, CA, USA). DNA concentration and quality were determined by gel electrophoresis using the Low DNA Mass Ladder as reference (Life Technologies, cat. no. 10068-013, CA, USA).
DNA polymorphism was screened using 100 RAPD primers (Williams et al., Reference Williams, Kubelik, Livak, Rafalski and Tingey1990) and 72 inter-simple sequence repeat (ISSR) markers (Zietkiewicz et al., Reference Zietkiewicz, Rafalski and Labuda1994); we use a subset of samples (n = 20) representing different localities and phenotypic fruit variation. Polymerase chain reaction (PCR) amplification was performed in 7.5 μl reactions using 5 ng of genomic DNA, 1.0 μM of RAPD or ISSR marker, 1 × PCR buffer (50 mM Tris, pH 8.5; 10 mM KCl; 2 mM MgCl2); 500 μg/ml bovine serum albumin; 0.01% xylene cyanol and 1.5% of Ficoll 400 (Sigma-Aldrich, F4375, St. Louis, MO, USA), 0.1 μM dNTPs, and 0.6 U of Taq polymerase (5 U/μl) (Invitrogen cat. no. 10342-020, Carlsbad, CA, USA). PCR temperature profiles were as follows: 94°C (5 min), followed by 40 cycles of 94°C (30 s), annealing temperatures ranged from 42 to 60°C, depending on the specific primer conditions, extension at 72°C (2 min) and final extension at 72°C (7 min). PCR products were visualized in 2% agarose gels and amplification profiles processed using the Lab Works 3.0 image software (UPV, Upland, CA, USA). PCR bands were scored based on differences in size using the 1 kb DNA Ladder as reference (Invitrogen, cat. no. 10381-010, Carlsbad, CA, USA) as reference.
AFLP genotyping
Based on the results of the screening performed with RAPD and ISSR markers, we selected a set of 29 representative accessions to screen a higher number of DNA polymorphisms using the standard AFLP protocol published by Vos et al. (Reference Vos, Hogers, Bleeker, Reijans, Lee, Hornes, Friters, Pot, Paleman and Kuiper1995). We applied M13 technology in a LI-COR 4300S (IR2; Li-Cor Biosciences, Nebraska USA) using the IRDye Fluorescent AFLP kit recommended for large genomes (Li-Cor Biosciences cat. no. 830-06195, Nebraska, USA). Passport data of selected accessions genotyped with AFLP markers are presented in Table 1.
Rubus samples used in the AFLP analysis
NA, not available.
a Herbarium identification.
For the digestion with the restriction enzymes EcoRI and MseI we used 500 ng of genomic DNA for each accession. The following AFLP combinations were screened using DNA from four different R. glaucus accessions differentiated by ISSR markers: IRDye-labelled EcoRI markers ACG, ACC, AGC, ACT, AAC with unlabelled MseI markers CAC, CTC, CTT, CTA, CTG, CAA and CAT. Of these, five AFLP combinations were selected based on the number of polymorphisms: E-AAC/M-CAT, E-AAC/M-CAT, E-ACC/M-CTC; E-AGC/M-CTC; E-ACG/M-CTG and E-AAC/M-CAT. AFLP gels were genotyped using the SAGA-MX™ AFLP® analysis software program (IR2; Li-Cor Biosciences, Nebraska, USA).
Data analysis
AFLP polymorphic bands were scored as 1 if present, 0 if absent and 9 if ambiguous. Only bands of similar intensity across samples were recorded to minimize missing data in the matrix. Bands observed in a single genotype were not used for the statistical analysis. Genetic variability parameters were calculated and genetic structure was inferred using the Darwin program (Perrier and Jacquemoud-Collet Reference Perrier and Jacquemoud-Collet2006). Cluster and factorial analyses were performed based on Jaccard similarity coefficient and neighbour joining (NJ) method. Bootstrap analysis was performed using 1000 replicates. Analysis of molecular variance (AMOVA) (Excoffier et al., Reference Excoffier, Smouse and Quattro1992) was calculated using GenAlEx 6.5 (Peakall and Smouse, Reference Peakall and Smouse2006). STRUCTURE version 2.3.4 was used to assign individual genotypes to populations using a Bayesian approach (Pritchard et al., Reference Pritchard, Stephens and Donnelly2000). The length of the Monte Carlo Markov chain was set at 100,000 with a 10% burn-in. The program Structure Harvester (Earl, Reference Earl2012) was used to determine the optimal value of K, which represents the optimal number of populations or groups.
Results
Preliminary screening
The 100 RAPD markers tested did not show any polymorphism for DNA R. glaucus samples; these markers revealed molecular differentiation between R. glaucus and the screened wild Rubus samples in almost all primers. Similar results were obtained for the screening with 72 ISSR markers. One ISSR primer (ISSR-872; 5′GATAGATAGATAGATA3′) was found to be useful for revealing polymorphisms in R. glaucus accessions. This primer was used to characterize the 106 Rubus samples collected, producing a total of 16 DNA polymorphisms, of which nine were polymorphic for R. glaucus samples. Thus, the amplification patterns obtained in R. glaucus with this primer were useful for selecting samples to conduct the AFLP analysis.
AFLP genotyping
The five AFLP combinations revealed 203 AFLP bands in the 29 selected accessions. Bands ranged from 29 to 50 markers in the five selected primer combinations (Table 1). Among the 203 bands, a total of 139 markers were scored only in accessions identified as R. glaucus (Table 2).
AFLP combinations and number of molecular markers observed
Genetic diversity
The cluster analysis revealed the presence of three main clusters identified in Fig. 1 as clusters P1 and P2, and W. R. glaucus samples were distributed in clusters P1 and P2 whereas cluster W comprises the wild accessions corresponding to the species R. urticifolius, R. niveus, R. loxensis, R. lanciniatus and R. boliviensis. Interestingly two accessions identified previously as R. glaucus (AP003 and AP025) were also included in this cluster. The sample placement shown in the cluster analysis did not show any geographical location correlation; however some bootstrap values support a robust grouping of many genotypes. In addition, structure analysis indicated by colour scale differentiation shown in Fig. 1 revealed some possible admixed genotypes of R. glaucus with wild Rubus species.
NJ tree using Jaccard similarity coefficient based on 203 AFLPs for Andean blackberry sampled in Ecuador. Bootstrap values are shown in blue. Groups assigned by structure analysis are displayed to the right.
Factorial analysis shown in Fig. 2 supported the results obtained by NJ cluster analysis. STRUCTURE showed three main lineages (optimal K = 3) for R. glaucus and closely-related species. AMOVA identified a highly significant differentiation (P = 0.001) among R. glaucus, P1 group, and the wild group, and P2 (PhiPT = 0.40 and 0.28, respectively) (Table 2). An estimated low gene flow between these two cultivated groups is suggested by the number of migrants (N m) value of 0.503 (Table 3), even if some admixed genotypes are evidenced by the assignation test mainly in the P1 group (Fig. 1). Among the admixed genotypes identified by assignation values, the accessions AP003 and AP025 are more differentiated from the other R. glaucus samples.
Factorial analysis plotting the three groups differentiated by cluster and structure analysis.
Genetic values obtained for the three genetic groups revealed by molecular markers
Note: N m = number of effective migrants, GD = Nei genetic distance, P1 = group 1 R. glaucus, P2 = group 2 R. glaucus, W = Rubus spp.
*P < 0.001.
AFLP variation in R. glaucus
Since RAPD and ISSR primers were not useful for detecting a high genetic differentiation in R. glaucus, AFLP technology performed to screen a high number of DNA polymorphisms. The unweighted pair group method with arithmetic mean cluster based on 139 polymorphic AFLPs for R. glaucus cultivars showed two main groups: P1 and P2 (Fig. 2). As shown in Fig. 3 these genetic groups and the wild group were sampled in the centre region in Ecuador. Group P1 includes non-thorned R. glaucus accessions (including a commercial variety) and some thorned ones, whereas group P2 includes the remaining thorned accessions. AMOVA revealed significant to moderate genetic differentiation (PhiPT = 0.33) within R. glaucus groups.
Distribution map of Rubus samples screened with AFLPs in Ecuador. Colour dots display the groups based on AFLPs analysis. P1 = yellow; P2 = blue; W = green.
Discussion
The population dynamics structure and genetic diversity of Rubus berries from Ecuador were assessed using arbitrary molecular markers. AFLP technology efficiently revealed more DNA polymorphisms in contrast to the other molecular markers used in a preliminary screening, of which almost any DNA differences were observed except for one ISSR marker. Thus, this marker was useful for selecting a set of R. glaucus accessions screened with AFLP technology. An intrinsic population structure suggesting the prevalence of two main lineages of R. glaucus was revealed by AFLP. This structure is not associated with geographic distribution or phenotype. Compared to morphological characterization assessed by Mejía (Reference Mejía2011) for this same germplasm, no evident association is determined between both studies. However, the P2 genetic group (Fig. 3) comprises only thorned R. glaucus accessions, whereas thorned and no-thorn accessions are in the P1 group. A low genetic exchange among the two R. glaucus groups is suggested; in contrast, a possible genetic flow between wild Rubus species and R. glaucus is proposed based on the identification of admixed genotypes and a closer genetic affinity of P2 group. Grouping of R. glaucus in two main clusters is congruent with earlier studies (Morillo et al., Reference Morillo, Morillo, Zamorano, Vásquez and Muñoz2005; Marulanda et al., Reference Marulanda, Aguilar and Lopez2007, Reference Marulanda, Lopez and Uribe2012). This result suggests that R. glaucus populations studied could differ at the ploidy level, as described in previous studies (Alice, Reference Alice2002; Marulanda et al., Reference Marulanda, Aguilar and Lopez2007). A relative low gene flow among these groups is suggested. The results display possible genetic flow between cultivated and wild Rubus species according to Graham et al. (Reference Graham, Squire, Marshall and Harrison1997) who, using RAPD, evaluated the genetic diversity among populations of Rubus at 12 locations across the UK and reported little gene flow between wild and cultivated accessions. Amsellem et al. (Reference Amsellem, Noyer and Hossaert-McKey2001) studied the spatial diversity among native and introduced plants of R. alceifolius and suggested that reproduction differs according to the introgression site, meaning that sexual reproduction is common for native species and apomixes is common in non-native locations. This suggestion could be applied to the results obtained in this study. AMOVA results point to a high genetic variability among and within Rubus species in general. AFLP markers differentiated the species, particularly R. glaucus and R. urticifolius, which in previous studies had not been resolved (Marulanda et al., Reference Marulanda, Aguilar and Lopez2007).
Genetic variability in cultivated Rubus species is lower in Ecuador compared to genetic assessments in Colombia. Using three AFLP combinations, Marulanda et al. (Reference Marulanda, Aguilar and Lopez2007) found that 91.6% of produced bands were polymorphic; in this study, five primer combinations revealed 77% polymorphism. After testing several molecular markers, 68% of all collected R. glaucus samples were found to be duplicates. This result can be explained by the facultatively apomictic reproductive behaviour of blackberry. Apomictic reproduction occurs in plants where botanical seeds are produced asexually by pseudogamy, resulting in low genetic variability (Kollmann et al., Reference Kollmann, Steinger and Roy2000; Šarhanová et al., Reference Šarhanová, Sharbel, Sochor, Vašut, Dančák and Trávníček2017; Hojsgaard and Hörandl, Reference Hojsgaard and Hörandl2019). Vegetative reproduction, commonly employed by traditional farmers (Kollmann et al., Reference Kollmann, Steinger and Roy2000; Viteri et al., Reference Viteri, Martínez, Jácome, Ayala, Villares, Viera, Sotomayor, Posso, Hinojosa, Galarza, Garcés, Velásquez, Sánchez and Zambrano2016) contributes to the lack of genetic diversity.
Plants collected from the same location and grower can have different genotypes, e.g. AP026 and AP025. We also found significant differences among plants from the same province; this result agrees with Graham et al. (Reference Graham, Marshall and Squire2003), who found variability between sites and attributed this to small changes in temperature, altitude and adaptation of reproduction that could result in asynchrony of flowering. AMOVA indicated a greater variability between genotypes than between populations (Fig. 3, which agrees with the factorial analysis). This result corroborates previous reports of diversity the sense of a limited genetic diversity of R. glaucus in other germplasm collections. The use of highly resolutive markers, such as the AFLPs, in closely-related genotypes allowed us to discriminate between genotypes in R. glaucus and to identify a strong genetic structure. This structure is well supported by other molecular markers, such as universal primers of chloroplast DNA and microsatellites created in our laboratory (data not shown) and could have its origin in genetic introgression. Therefore, this structure is a product of sexual recombination, where even low levels of germination may be sufficient to maintain genetic variability in a predominantly apomictic population (Kollmann et al., Reference Kollmann, Steinger and Roy2000). It could also be related to the level of ploidy of the species, since Rubus has a wide plasticity of forms of reproduction, which are directly linked to its level of ploidy. Diploid plants reproduce sexually, whereas polyploid species are apomictic, pseudogamous and self-fertile (Antonius and Nybom, Reference Antonius and Nybom1995). Currently, the origin of this structuring of the species cannot be determined with certainty. This information not only contributes to a better understanding of the diversity of R. glaucus, but has direct implications in the breeding of this native fruit which will allow the selection of highly differentiated parents.
This study is the first to analyse the genetic diversity of Andean blackberry in Ecuador and closely-related species producing blackberries. Our molecular survey results suggest that the centre region in Ecuador could be a hot spot of the diversity of R. glaucus and related berries species. Knowledge about genetic diversity provides an insight into the current status of the cultivated Andean blackberry, revealing that two main lineages exist. The low genetic variability among R. glaucus species is associated with the common cultural practices for this cultivation, which is based on asexual reproduction using stolons. This practice is predominant in this agricultural system and should be reconsidered because it reduces the genetic diversity of the cultivated species. Studies aiming to differentiate ploidy level, differentiate between physical variables such as adaptation of reproduction to temperature differences, as well as restraint cultivars, are all needed to use and preserve the genetic diversity of R. glaucus.
Molecular screening using molecular markers, such as RAPDs and ISSRs, suggested that R. glaucus has a narrow genetic base; however molecular characterization applying AFLP-M13 technology in a set of selected cultivars revealed genetic variability and an unexpected strong genetic structure. A closer genetic affinity of one group of cultivars was displayed with the wild Rubus group. Admixed genotypes in R. glaucus suggest a possible introgression of wild Rubus species in cultivars that could increase genetic variability in this crop. These findings are relevant for Andean blackberry genetic breeding and germplasm use of this important fruit crop.
Acknowledgements
We thank National Herbarium of Ecuador in Quito for sample identification, and William Viera and Alvaro Monteros from INIAP for their participation in sampling. We also thank Gwen Bloomsburg from UDLA-Quito for the English revision of the manuscript. This research was supported by the project Mora-Naranjilla financed by FONTAGRO.
