Field Crops Research 127 (2012) 64–70 Contents lists available at SciVerse ScienceDirect Field Crops Research journal homepage: www.elsevier.com/locate/fcr Genetic diversity in tef (Eragrostis tef) germplasm using SSR markers M. Zeid a,1 , K. Assefa b , A. Haddis c , S. Chanyalew b , M.E. Sorrells a,∗ a b c Department of Plant Breeding and Genetics, 240 Emerson Hall, Cornell University, Ithaca, NY 14853, USA Ethiopian Institute of Agricultural Research, Debre Zeit Agricultural Research Center, P.O. Box 32, Debre Zeit, Ethiopia CSIRO Plant Industry, Bld 1-Dough Lab, Clunies Ross St., Acton, ACT 2601, Australia a r t i c l e i n f o Article history: Received 11 June 2011 Received in revised form 27 October 2011 Accepted 27 October 2011 Keywords: Chloridoideae Allelic diversity Seed admixture Lodging a b s t r a c t The Institute of Biodiversity Conservation (IBC) in Ethiopia curates a large number of tef accessions, but there is little information about the genetic variation among the accessions. This study assessed the genetic diversity and relationships among 326 cultivated tef accessions, 13 wild relatives, and four commercial tef varieties from the U.S. using 39 SSR markers, 26 of which were flanking QTL intervals for yield, lodging index and stem strength related traits. We estimated the allelic diversity and identified markers associated with agronomic traits in this tef germplasm collection. Forty-seven loci were sufficient to differentiate 80.8% of the tef accessions. In contrast to earlier studies, genetic similarity estimates ranged from 0.21 to 0.99, indicating a high level of genetic diversity. In the course of this investigation, it was discovered that seed admixture is a serious problem affecting the integrity of almost all released tef varieties. Association was observed between the marker CNLTs540 and seed weight/plant. The majority of the alleles detected were present in tef breeding lines and varieties suggesting that tef plant breeders have been using a broad range of germplasm in their programs. The markers documented in this study will be useful to identify and verify hybrids from crosses between promising lines that lack morphological differences, an approach that was never attempted before in the tef breeding programs. © 2011 Elsevier B.V. All rights reserved. 1. Introduction Tef, Eragrostis tef (Zucc.) Trotter, is an annual self-pollinated grass species of the family Poaceae, subfamily Chloridoideae. Among cultivated cereals, only finger millet belongs to the same subfamily, thus the crop is only distantly related to the well-studied cereal crops. Within the genus Eragrostis, tef is the only species that is grown to produce grain (100 kernel weight = 0.18–0.38 mg) for human consumption, while the straw is fed to livestock. The crop is grown on a large scale in Ethiopia where it occupies 2.5 million hectares thereby ranking first among all cereals cultivated in the country. The grains are ground into flour, fermented and baked to produce flat pancake-like bread known as “injera”. Its flour lacks gluten prompting its use in food products for individuals diagnosed with celiac disease (Spaenij-Dekking et al., 2005; Hopman et al., 2006). Despite the high demand for tef in Ethiopia, as observed from the doubling of the production area from 1.2 million ha in 1990 to 2.5 million ha in 2009, the yield of tef per unit area has not changed for decades from an average of 1 ton/ha. Current tef varieties can yield 3–4 times more than the country’s average ∗ Corresponding author. Tel.: +1 607 255 2180; fax: +1 607 255 6683. E-mail address: mes12@cornell.edu (M.E. Sorrells). 1 Present address: Crop Science Department, Faculty of Agriculture, Alexandria University, Alexandria, Egypt. 0378-4290/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.fcr.2011.10.013 under experimental conditions without lodging (Teklu and Tefera, 2005). However, both farmers’ production practices and lodging are the two major limitations for yield improvement in tef nationwide. The current farmers’ production practices have not changed much from those traditionally used generations ago as described by Ketema (1997). In addition, lodging continues to be a serious problem in tef production causing an estimated average yield loss of 17% (Ketema, 1993). So far, it has not been possible to develop lodging resistant varieties mainly because of the lack of variation for lodging resistance within the available germplasm (Assefa et al., 2010). The fact that Ethiopia is considered to be the center of origin and diversity of tef (Vavilov, 1951) implies that the indigenous germplasm represents a rich resource base for plant breeders to improve this crop. Collection and conservation of the available tef material and its wild relatives is an ongoing process in Ethiopia because some of the regions have not yet been adequately represented in the germplasm collection (Demissie, 2001). Improved tef varieties are not replacing local landraces as fast as in the case of wheat, however, this might change soon, since tef production is becoming a more market-driven process where white-seeded varieties are favored over darker-seeded types (Belay et al., 2008). The fear for genetic erosion of tef diversity as observed in tetraploid wheat landraces (Tsegaye and Berg, 2007) due to the expansion of improved varieties is the main reason for collection and conservation of tef germplasm. The Institute of Biodiversity M. Zeid et al. / Field Crops Research 127 (2012) 64–70 Conservation (IBC) in Ethiopia has 4395 tef accessions (Demissie, 2001) that are shared with research institutes and tef breeders in Ethiopia, but most of the tef collection is lacking adequate passport data necessary for proper germplasm utilization. Several studies have examined the diversity available in some of the available tef germplasm using morphological and agronomic traits (Tefera et al., 1990; Assefa et al., 1999, 2000, 2001a,b,c, 2002a,b, 2003b; Ayele et al., 1999; Adnew et al., 2005). The results of all these studies generally agreed on the existence of high levels of variation for many of the phenotypic traits investigated. The most extensive study, however, was conducted by Ebba (1975), where more than 1000 field and market collections from tef were morphologically characterized for inter and intra-sample variation. Based on results of that study, 35 distinct landrace collections of tef were selected to represent the entire variation in the indigenous germplasm. Similar attempts were also undertaken at the DNA level using random amplified polymorphic DNA (RAPD) (Bai et al., 2000), amplified fragment length polymorphism (AFLP) (Bai et al., 1999b; Ayele and Nguyen, 2000). Those studies focused mainly on the genetic diversity among the 35 landrace collections of Ebba (1975), in addition to a few pure line accessions from the IBC, and accessions of Eragrostis pilosa and Eragrostis curvula were included for comparison. In those studies, it was possible to differentiate cultivated tef from its wild relatives and accessions within tef, however, the genetic similarity estimates among the tef accessions were generally high (0.73–0.98). It was concluded that the diversity in cultivated tef was very low and the authors stressed the need for inter-specific crosses with E. pilosa to overcome this problem. Since the low level of variation at the DNA level could not explain the large morphological variation observed among tef accessions, it was assumed that only a few genes controlled this broad morphological variation (Bai et al., 2000). The utilization of inter-simple sequence repeats (ISSR) markers (Assefa et al., 2003a) provided, for the first time, much lower estimates for genetic similarity (0.26–0.86) among 92 tef accessions including most of the accessions screened in the previous studies. These results suggested that the low level of diversity observed earlier was marker related and that microsatellites would provide better markers for future diversity studies in tef. This assumption was validated by Zeid et al. (2010), where 53% of the tef-specific SSR markers developed proved to be polymorphic between the parental lines of an inter-specific cross between E. tef (Kaye Murri) and E. pilosa (30–5). In addition, a number of SSR markers were screened on a limited number of tef accessions and a promising level of polymorphism among those accessions was observed. These markers have been used successfully for genetic map development and QTL analysis in tef for yield, lodging and related traits (Zeid et al., 2010) and were proposed as a useful tool for tef germplasm organization and fingerprinting. SSR markers have been very useful for studying the genetic diversity of self-pollinated cereals including wheat (Khlestkina et al., 2007; Mohammadi et al., 2009) and rice (Garris et al., 2005; Chao et al., 2007; de Oliveira Borba et al., 2009) and for studying the genetic integrity of collections in gene banks (Hirano et al., 2009). Furthermore, they were also employed for verifying pure line and hybrid seed identity in hybrid rice programs (Nandakumar et al., 2004; Sundaram et al., 2008). The objectives of this study were to (i) estimate the genetic diversity and the phylogenetic relationship among tef breeding lines in the core germplasm used in the tef breeding programs in Ethiopia and their wild relatives and some commercial tef varieties from the U.S.; (ii) fingerprint all released Ethiopian tef varieties; (iii) study the effect of seed admixture on the integrity of the released varieties; and (iv) to examine the allelic diversity in a set of QTL-flanking SSR markers in the tef germplasm and their potential association with stem lodging and related traits in tef. 65 2. Materials and methods 2.1. Plant material In this study, a total of 343 Eragrostis germplasm accessions were assayed. The Debre Zeit Agricultural Research Center (DZARC) in Ethiopia provided all of the accessions except for four that were provided by The Teff Company, Caldwell, Idaho, U.S. (Supplementary Table 1). The germplasm accessions provided by DZARC included 13 wild Eragrostis species and 271 tef breeding accessions that are actively being evaluated and used in the tef breeding program. Twenty-one of those breeding accessions had the same selection number with different suffix (Supplementary Table 1). In addition, 33 landraces that had been described earlier by Ebba (1975) to represent the entire tef variation grown in Ethiopia and 22 improved varieties that were developed between 1995 and 2007 were also included. 2.2. DNA extraction and genotyping DNA was extracted from bulked leaves of five seedlings from each accession based on the method of Tai and Tanksley (1990). Because of the tiny seed size of tef, seed admixture is a common problem. To study the effect of seed admixture on the identity of the most recent varieties, 12 lines were genotyped from seed lots provided by DZARC in 2003 and were compared to those of seed lots from 2009. Similarly, fingerprints of seedlings from the cultivar Kaye Murri and the wild accession E. pilosa (30–5) from three different seed lots were compared. Internal checks consisting of aliquots of DNA extracted from six accessions assigned to two different wells across the four 96-well plates assayed were employed as controls for any errors during sample handling. Sixty eight of the tef SSR primer-pairs developed by Zeid et al. (2010) were screened on a subset of 37 tef genotypes (19 of the most recent improved varieties and 18 landraces) in addition to E. pilosa (30–5) using the same protocol described by the authors (Supplementary Table 2). The PCR amplification products were separated on 4% polyacrylamide gels and visualized after staining with silver nitrate. Based on results of this preliminary screening, 13 primer-pairs of high polymorphic information content values (Botstein et al., 1980) that were also capable of differentiating the screened genotypes were selected for assaying the germplasm studied here (Supplementary Table 2). This set of 13 selected primer-pairs in addition to 26 other primers that were flanking QTL for various traits in tef including 100 seed weight, plant height, and lodging index in the study by Zeid et al. (2010) were labeled using fluorescent dye and used to fingerprint the entire germplasm collection under study (Table 1). SSR fragments were analyzed on the ABI PRISM 3730 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) and visualized and scored using Genographer 2.1.4 software (Benham et al., 1999). Scoring of products either from polyacrylamide gels or from the digital visualization from Genographer was performed by numbering the products of each primer-pair based on the number of alleles detected for that locus. The software NTSYSpc v2.21h (Rohlf, 2009) was used to calculate the shared band similarity coefficient (Lynch, 1990) and the UPGMA clustering method (Sneath and Sokal, 1973) was applied to the calculated similarity matrix. Relationships among accessions were analyzed using the principal coordinate analysis (PCoA) method (Gower, 1966) in NTSYSpc. Power Marker v3.25 (Liu and Muse, 2005) was used to calculate the number of alleles per locus, Wright’s fixation index (Fst) and polymorphism information content (PIC) for each locus. 66 M. Zeid et al. / Field Crops Research 127 (2012) 64–70 Table 1 SSR primer pairs used for genetic diversity and association analysis. Primer Linkage group Gene diversity PIC CNL 53 CNLT 12 CNLTs11 CNLTs17a CNLTs17b CNLTs85 CNLTs95 CNLTs96 CNLTs102a CNLTs102b CNLTs117 CNLTs136a CNLTs136b CNLTs205a CNLTs205b CNLTs220 CNLTs244 CNLTs255 CNLTs275a CNLTs295 CNLTs334 CNLTs338 CNLTs380 CNLTs392 CNLTs416a CNLTs416b CNLTs438 CNLTs455a CNLTs455b CNLTs485 CNLTs498 CNLTs540 CNLTs33 CNLTs42 CNLTs60 CNLTs91 CNLTs116 CNLTs133 CNLTs150a CNLTs150b CNLTs157 CNLTs216 CNLTs315 CNLTs328 CNLTs538 CNLTs484a CNLTs484b 13 9 6 – 13 1 9 24 6 – 6 3 – 3 – 5 3 3 15 19 2 1 23 9 1 – 2 15 – 7 9 9 – – 0.07 0.58 0.89 0.03 0.12 0.61 0.05 0.48 0.22 0.13 0.52 0.81 0.89 0.61 0.06 0.64 0.52 0.87 0.12 0.81 0.54 0.17 0.84 0.40 0.78 0.84 0.81 0.84 0.84 0.04 0.24 0.50 0.75 0.91 0.87 0.64 0.75 0.89 0.78 0.64 0.79 0.80 0.72 0.58 0.88 0.73 0.88 0.07 0.49 0.88 0.03 0.12 0.53 0.05 0.36 0.19 0.13 0.40 0.79 0.88 0.53 0.06 0.59 0.41 0.86 0.12 0.79 0.44 0.17 0.82 0.32 0.75 0.82 0.78 0.82 0.82 0.04 0.23 0.40 0.73 0.90 0.86 0.58 0.71 0.88 0.76 0.58 0.77 0.78 0.69 0.53 0.87 0.68 0.87 – – – – – – – – – 2.3. Phenotyping and association analysis Phenotypic data on traits mainly related to stem characteristics and lodging were measured on 271 accessions (Table 2). The experiment was conducted at DZARC in the central highlands of Ethiopia during the 2001 main season. Seeds (0.75 g plot−1 ) were sown in single 1 m long rows with 20 cm spacing between rows in a randomized complete block design with three replicates. Traits measured prior to harvesting on plot basis included days to heading, days to maturity, and lodging (scored on a 0–5 scale, where “0” represented upright standing plants and “5” for plants lying flat on the ground). Plant yield (seed weight/plant in g), and post harvest measures for stem characteristics including; penetrometer reading on the first, second and third internodes (measured by a digital force gauge, model MG10- 0.002 kgf, as the force needed to penetrate the stem at the internodes position), length (cm) and diameter (mm) of the first and second internodes, length (cm) and diameter (mm) of the peduncle and culm length (cm), were all measured as the mean of five plants per plot. The association between best linear unbiased predictors of the three replicates for the studied traits and alleles from all 39 SSR markers employed were analyzed using the mixed linear model (MLM) in TASSEL v2.1 (http://ww2.maizegenetics.net/) taking the average relationship between accessions (kinship matrix) into account. A false discovery rate cutoff was determined from the observed p-value distribution, FDR = 0.05. 3. Results All 68 SSR primer-pairs successfully amplified products for all of the 37 tef accessions in addition to E. pilosa (30–5) and the markers differentiated all 37 tef lines. The PIC values for the markers ranged from 0.05 to 0.86 (Supplementary Table 2). Eighteen markers amplified only two alleles, one specific to tef lines and the second specific to E. pilosa including the two markers (CNLTs37 and 68) with the lowest PIC values, 0.05 and 0.09, respectively. Thirteen primer-pairs out of the 69 screened were then selected to study the genetic diversity among the entire germplasm collection. The selection was based on both their PIC values and ability to differentiate the closely related lines as observed on silver nitrate stained polyacrylamide gels. The process of scoring alleles of a single marker on different gels is very difficult and error prone (Zeid et al., 2003), consequently the 13 selected primer-pairs for their high PIC values and the 26 other primer-pairs that were flanking QTL position in the study of Zeid et al. (2010) were fluorescent labeled and scored using the software Genographer v2.1.4. Fingerprints from each of the two aliquots of the six internal checks employed as controls for error during sample handling were identical for all 47 loci that were amplified from the 39 primer-pairs. Similarly, fingerprints from the three different seed lots for the landrace Kaye Murri were identical. On the other Table 2 Minimum, maximum, mean phenotypic values and standard deviation of 272 tef lines for the different traits. a b Trait Abbreviation Minimum Maximum Mean (SD) Lodginga (0–5 score) Days to headinga (days) Days to maturitya (days) Seed weight/plantb (g) Culm lengthb (cm) Peduncle lengthb (cm) Peduncle diameterb (cm) Length of first internodeb (cm) Length of second internodeb (cm) Culm diameter at first internodeb (cm) Culm diameter at second internodeb (cm) Penetrometer reading at first internodeb (kgf) Penetrometer reading at second internodeb (kgf) Penetrometer reading at third internodeb (kgf) Lodg DH DM SW CL PdL PdD IL1 IL2 CD1 CD2 Pr1 Pr2 Pr3 1.94 27.06 107.07 1.31 23.93 12.17 1.08 4.95 6.84 1.20 1.11 0.942 0.69 0.47 2.07 44.31 113.18 2.79 47.2 20.20 1.85 5.98 8.31 2.35 2.21 2.42 1.82 1.09 2.02 34.83 110.05 1.73 36.26 16.00 1.45 5.34 7.50 1.73 1.70 1.61 1.20 0.76 Measurements were scored on a plot basis. Measurements were based on the mean of 5 randomly selected plants/plot. ± ± ± ± ± ± ± ± ± ± ± ± ± ± 0.02 3.50 0.98 0.23 4.05 1.37 0.14 0.14 0.25 0.21 0.20 0.29 0.21 0.12 M. Zeid et al. / Field Crops Research 127 (2012) 64–70 hand, only one line (DZ-Cr-82), of the recently developed varieties, had identical fingerprints for both the 2003 and 2009 seed lots, while the other 11 lines showed variations in their fingerprints and consequently their genetic similarity estimates (GS) ranged between 0.45 and 0.98 with a mean GS of 0.79. Genetic similarity between seeds from the three seed lots of the wild accession E. pilosa (30–5) varied between GS = 0.45 and GS = 0.67 with a mean GS of 0.55. The 47 loci scored were able to clearly differentiate the 277 accessions (80.8%) while 27 accessions showed similarity levels of GS = 1 to at least one other accession and up to five out of the 343 accessions under study (Supplementary Table 3). After excluding duplicate accessions, the genetic similarity estimates between tef accessions ranged between 0.20 and 0.99. A dendrogram was generated to summarize the relationship between all 343 accessions included in this study (Supplementary Fig. 1). All wild Eragrostis species, except for the E. cilianensis accession, clustered together separately from cultivated tef accessions. The dendrogram revealed a small number of closely clustered accessions. Graphical representation of the relationships between accessions in the PCoA showed a good differentiation of the tef breeding lines on the first coordinate that explained 7.5% of the variance. The third coordinate (4.3% of the variance), on the other hand, clearly differentiated the tef accessions from their wild relatives (Fig. 1). As shown in this figure, all the 33 landraces and the four U.S. varieties were differentiated from one another. 3.1. Allelic diversity and association analysis The number of alleles amplified per locus across all 343 accessions ranged from 8 to 23 in the case of those markers selected for their high PIC value and from 2 to 23 for the QTL-flanking markers chosen from previous studies (Fig. 2). Although the tef breeding lines and varieties harbored the majority of alleles (85%) observed in this study (Fig. 2), a number of unique alleles were amplified from wild relatives of tef (9.9%), tef landraces (3.8%) and the U.S. lines (1.3%). Those unique alleles could be valuable for tef breeders in cases where the number of alleles available in the tef accessions and varieties is very low as observed for QTL-flanking loci in the left part of the histogram of Fig. 2. Association between stem related traits in addition to lodging and yield/plant and the 47 loci were analyzed for the 271 unique tef accessions. The marker locus CNLTs540 was significantly associated with seed weight/plant but no other significant associations were observed. 4. Discussion Molecular characterization of germplasm accessions is a useful tool for studying and managing the genetic diversity available in the genebanks. This tool becomes even more valuable when passport data is very scarce or lacking. In this study, microsatellite markers were used to characterize the core collection of tef breeding germplasm available in Ethiopia including breeding lines, released varieties and wild species, some of which have been employed for producing inter-specific crosses. The 271 breeding lines studied here along with the 33 landrace collections are the primary germplasm resources used by breeders for tef improvement. Contemporary tef breeding programs have shifted from mass selection within landraces towards targeted inter- and recently intra-specific crosses since the proper hybridization methods were established by Berhe (1975) according to Assefa et al. (2010). The first set of 26 SSR markers employed here were distributed across 12 of the 30 linkage groups identified by Zeid et al. (2010) and were flanking QTL positions for various traits in tef. An additional set of 13 markers with high PIC values that were not previously 67 mapped were selected from 68 markers. These 39 markers differentiated 80.8% of the accessions assayed and identified the rest as redundant accessions (Supplementary Table 3 and Fig. 1). When the fingerprints of either of the two sets of markers were used to assess the variation in the 20 tef varieties from the 2009 seed lot, all varieties, except DZ-Cr-255 and DZ-Cr-82 were differentiated, and the correlation between the GS estimates for the two marker sets was r = 0.71. These markers thus, established the first tef fingerprints that could be used for variety identification and protection. Because of the scarcity of passport data on the tef accessions studied here, identical (redundant) accessions were expected. We identified 27 cases where accessions were identical to one or more other accessions (Supplementary Table 3). Only one accession representing those redundant accessions was used in further analysis, while the rest (N = 39) were excluded. On the other hand breeding accessions carrying the same selection number with a different suffix (N = 21) as seen in Supplementary Table 1, were shown to be dissimilar, and the basis for the addition of the suffix was not documented. Knowing that farmers traditionally grow tef in mixtures even with contrasting seed colors (Assefa et al., 2002a), the most probable explanation is that plant materials collected from the same farmer’s field during germplasm acquisition were assigned the same selection number, and a different suffix was attached if morphological differences were apparent. This identification method is an indication that the methodologies employed during tef germplasm acquisition were efficient and successful in recognizing the available variation within a specific collection area. After excluding 39 redundant accessions, the genetic similarity estimates among the remaining tef accessions ranged between 0.20 and 0.99 revealing a broad base of genetic diversity available in tef. These results indicated that the high GS estimates reported in previous studies (Ayele et al., 1999; Bai et al., 1999b; Ayele and Nguyen, 2000; Bai et al., 2000) that use the same plant material (landraces) included in this study, was a marker dependent issue rather than an inherent low polymorphism in tef as previously suggested. These results are in agreement with Assefa et al. (2003a), where estimates of GS based on eight ISSR markers ranged from 0.26 to 0.86 among a smaller set (92) of tef accessions, most of which were included in the present study as well. Furthermore, the present findings also showed that the 33 landraces, that were originally selected by Ebba (1975) from farmer’s fields, based solely on morphological traits to represent the entire variation in the tef germplasm were equivalent in their range of variation for the GS values (0.20–0.99) to that of the 271 breeding accessions. However, significant differences were detected between the gene diversity for the breeding lines (Fst = 0.60) and landraces (Fst = 0.07). These results indicate the importance of the breeding line accessions as a major source of genetic diversity for the tef breeding programs. In this study, 70% of the markers were polymorphic between the two landrace accessions, Kaye Murri and Fesho. These two accessions were the parents of the first population used to construct a linkage map in tef (Bai et al., 1999a), using AFLP markers of which only 6.1% were polymorphic. The high level of polymorphism observed between the two landraces as shown in our study indicates the potential of those SSR markers for future mapping work utilizing intra-specific crosses and for combining the results obtained from the inter-specific cross E. tef × E. pilosa reported by Zeid et al. (2010). The four U.S. commercial varieties studied here had a mean GS value of 0.41 to the other tef accessions and a maximum GS value of GS = 0.69, suggesting that those are unique lines that were carefully selected and/or bred independent of the DZARC germplasm collection. Those four lines could be evaluated for their performance in Ethiopia for any beneficial traits they might possess. All 13 wild Eragrostis accessions studied here except E. cilianensis were differentiated from the cultivated tef accessions in 68 M. Zeid et al. / Field Crops Research 127 (2012) 64–70 PCo3 0.7 Breeding lines 0.6 Landraces 0.5 Wild accessions US lines Kaye Murri 0.4 Varieties 0.3 0.2 E. unioloides-46-1 E. cilianensis 0.1 PCo1 0 -0.6 -0.4 -0.2 -0.1 0 0.2 0.4 0.6 0.8 -0.2 -0.3 -0.4 Fig. 1. Principal co-ordinate analysis of 326 tef accessions (breeding lines, landraces and varieties) and 13 wild Eragrostis accessions representing the entire breeding program of tef in Ethiopia in addition to 4 U.S. commercial lines based on 47 loci. The first principal coordinate explained 7.5% of the variation, while the third coordinate explained 4.3% of the variation. the cluster analysis (Supplementary Fig. 1) and also on the third PCoA coordinate (Fig. 1). Previous studies employing molecular markers (Ayele et al., 1999; Bai et al., 1999b; Ayele and Nguyen, 2000; Bai et al., 2000) differentiated the three species; E. tef, E. pilosa and E. curvula, however, this is the first time that other wild species have been compared to a large number of tef accessions. Although E. cilianensis has been used as a cereal crop in parts of West Africa (Dalziel, 1937) and was suggested in many studies to be the progenitor of tef (Ingram and Doyle, 2003), our data suggested that an error due to mislabeling is the reason for a presumed close relationship to tef accessions. This, however, needs to be verified by screening another source of seed from that species. Mislabeling and lack of passport data are major problems for the tef germplasm collection and its wild relatives in Ethiopia and this study is a first step towards identifying and resolving this problem. Most of the wild accessions preserved at DZARC are the same accessions described by Jones et al. (1978), however, other accessions such as Dakota-50-1, Kiubins-20-1 and Oferifiana (Supplementary Table 1) are examples of cases where information is lacking except for being described as wild Eragrostis species, a description that is clearly supported by our results presented here (Fig. 1). Fig. 2. Frequency of alleles detected using (A) 32 loci flanking QTL positions and (B) 15 loci of high PIC values, for 326 tef accessions (breeding lines, landraces and varieties), 13 wild Eragrostis accessions and 4 U.S. commercial lines. M. Zeid et al. / Field Crops Research 127 (2012) 64–70 Tef is a predominantly self-pollinated crop, and for that reason any change in the genetic integrity of a specific line could be attributed to seed admixture as pollen flow is very limited in selfpollinated cereals (Matus-Cadiz et al., 2004; Wang et al., 2009). To address this issue and its effects on preserving line identity, fingerprints of two different seed lots from 2003 and 2009, of 12 tef lines were compared. Only one (DZ-Cr-82) of the 12 lines showed identical fingerprints between the seeds lots, and three lines shared a GS of less than 0.70, thereby signifying the seriousness of the seed admixture problem. Seed admixture is a common problem in small grain cereals and can easily be detected using SSR markers (Donini et al., 1998; Khlestkina et al., 2007). Tef breeders generally depend on morphological traits, especially those of the panicle form and lemma color, to differentiate lines in the field, and this task is becoming even more difficult, since all new lines are of the loose panicle type that is associated with higher yields. In addition the tiny seeds of tef make it very hard to maintain truly pure seed stocks because admixture can occur during any step of the seed production and handling process unless extreme care is practiced. For example, the close relationship between the line DZ-Cr-37 from the 2003 seed lot with DZ-Cr-385 (GS = 0.89 (a line derived from the cross of DZ-01-2785 × E. pilosa (30–5)) and HO-Cr-136 (GS = 0.87) (of unknown origin) was unexpected. However, two different seed lots of the same line DZ-Cr-37 were also different with GS = 0.86. Another clear case of seed admixture was observed with the accession E. pilosa (30–5), where the average GS between the three fingerprinted seed lots was GS = 0.55. These discrepancies in relationships between lines and seed lots of the same line are evidence of loss of integrity that could be attributed to seed admixture. In contrast, identical fingerprints for all three seed lots Kaye Murri were observed. This could be attributed to the unique morphology of that line, as it is characterized by its thick culms, compact and long panicles, red lemmas and white grains that make it easy to distinguish. This line has been involved in marker development, linkage mapping and QTL analysis (Zhang et al., 2001; Yu et al., 2006a,b, 2007) and is used in the current tef breeding program as a donor parent for strong culms. Interestingly, our results (Fig. 1) showed that this line was positioned closer to the wild Eragrostis species than the majority of tef accessions on the third principal coordinate. The fingerprint of that line substantiates the unique morphology because the mean GS between Kaye Murri and all tef accessions is equal to the mean GS between that line and all the wild species (GS = 0.42). 4.1. Allelic diversity and association analysis The phenotypic variation in this study did not reach the levels of variation reviewed earlier by Assefa et al. (2001a) despite the fact that there was overlap of accessions in both studies. Culm length, for example, a trait of major interest for better lodging resistance in tef, ranged from 23.93 to 47.2 cm in the present study (Table 2), while Assefa et al. (2001a) reported a range of 11–82 cm. Nevertheless, environmental effects influence these traits limiting the value of the comparison. The penetrometer reading adopted in this study as a measurement of stem strength at the lowest 3 internodes was difficult to measure because of the extremely thin stems of tef as compared to maize. Furthermore, no correlation was observed between lodging and the penetrometer readings at any of the three internodes (data not shown). There were no significant associations observed between the SSR markers and any of the traits except for the marker CNLTs540 previously reported by Zeid et al. (2010) to be flanking a QTL for lodging index. In the current study, however, this marker was associated with seed weight/plant rather than the lodging score. In general, lodging index has been shown to be positively correlated with seed weight/plant (Yu et al., 2007; Zeid et al., 2010), but in 69 the study by Zeid et al. (2010), no QTL for seed weight co-localized with the QTL for lodging index. In this work, lodging score using 0–5 scale rather than the lodging index (Caldicott and Nuttall, 1979) was employed, however, neither of the scales was correlated with any of the stem related traits. These observations suggest the need for an improved measurement method if lodging improvement in tef is to be achieved. In rice for example, lodging is known to be a complex trait that is the product of many stem and root characteristics. Pushing resistance, which is an index of resistance to stem bending and root lodging, was adopted as a more reliable measure for lodging resistance (Kashiwagi et al., 2008). This method employs a prostrate tester (Daiki Rika Kogyou Co., Tokyo, Japan) and could be explored in tef. Nevertheless, our study revealed much allelic diversity for some of the loci that were associated with agronomic traits in previous studies. The accessions in this study were selected from farmers’ landraces through pure-line selection (Assefa et al., 2010) and therefore, they represent the alleles harbored by those lines (Fig. 2). Thus the germplasm collection and conservation efforts were successful in capturing a large amount of the available diversity while enabling plant breeders to access this diversity. Other alleles that were only present in landraces or U.S. lines could be easily integrated into the breeding pipeline if they prove to be of value to the breeding program. A more difficult option would be to introgress useful alleles from wild relatives of tef. In both cases, however, molecular markers would be useful for monitoring this process and ensure a faster and more efficient transfer of the desired alleles. The tef breeding program has been successful in producing seven improved varieties through hybridization methods (Assefa et al., 2010), while relying only on morphological traits to ensure that the cross was successful. Those morphological traits have been the bases for selecting the parental lines of the cross, verifying the success of the hybridization process and for selecting among lines of the segregating populations. The most recent and successful selections, Quncho and Gamachis (Supplementary Table 1), were derived from the same cross (DZ-01-974 × DZ-01-196). While the former was selected for its desired white seed color, the latter was selected for its early maturity which is beneficial for moisturestress prone areas (Belay et al., 2008). Our marker results are in agreement with the pedigree relationship between the parental lines and the two selected lines indicating the SSR markers used in this study provide an unprecedented tool for improving the efficiency of the tef breeding program once implemented. Molecular markers have been used for assessing seed purity in hybrid seed lots in rice (Nandakumar et al., 2004; Sundaram et al., 2008). In tef however, these markers would also give breeders the chance to identify and verify hybrids from the cross between two promising lines that lack clear morphological differences. This will allow many new cross combinations that may combine short culm length and high yield, a combination that was not achieved with the cross between Kaye Murri and E. pilosa due to the weak stem and low yield potential of the wild parent (Tefera et al., 2003; Yu et al., 2007). The markers could also be used to organize the entire tef germplasm collection, reducing the number of duplicate accessions and grouping accessions based on their genetic relationship. The SSR markers presented here are reliable PCR-based markers that are easy to reproduce and once implemented in the tef breeding program, marker-assisted breeding and gene introgression will become feasible. Acknowledgements We gratefully acknowledge financial support from the McKnight Foundation’s Collaborative Crop Research Program for the African Chloridoid Cereals project (Grant No. 06-448). Thanks are due to all 70 M. Zeid et al. / Field Crops Research 127 (2012) 64–70 technicians at the Debre Zeit Agricultural Research Center, Ethiopia for their help collecting the field data. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.fcr.2011.10.013. References Adnew, T., Ketema, S., Tefera, H., Sridhara, H., 2005. Genetic diversity in tef [Eragrostis tef (Zucc.) Trotter] germplasm. Genet. Resour. Crop Evol. 52, 891–902. Assefa, K., Ketema, S., Tefera, H., Hundera, F., Kefyalew, T., 2001a. Genetic diversity for agronomic triats in tef. In: Tefera, H.G.B., Sorrells, M. (Eds.), Narrowing the Rift: Tef Research and Development. Ethiopian Agricultural Research Organization, Addis Ababa, pp. 33–48. Assefa, K., Ketema, S., Tefera, H., Kefyalew, T., Chundera, F., 2000. Trait diversity, heritability and genetic advance in selected germplasm lines of tef [Eragrostis tef (Zucc.) Trotter]. Hereditas 133, 29–37. Assefa, K., Ketema, S., Tefera, H., Nguyen, H.T., Blum, A., Ayele, M., Bai, G., Simane, B., Kefyalew, T., 1999. Diversity among germplasm lines of the Ethiopian cereal tef [Eragrostis tef (Zucc.) Trotter]. Euphytica 106, 87–97. Assefa, K., Merker, A., Tefera, H., 2002a. Qualitative trait variation in tef [Eragrostis tef (Zucc.) Trotter] germplasm from western and southern Ethiopia. Euphytica 127, 399–410. Assefa, K., Merker, A., Tefera, H., 2003a. Inter simple sequence repeat (ISSR) analysis of genetic diversity in tef (Eragrostis tef (Zucc.) Trotter). Hereditas 139, 174–183. Assefa, K., Merker, A., Tefera, H., 2003b. Multivariate analysis of diversity of tef (Eragrostis tef (Zucc.) Trotter) germplasm from western and southern Ethiopia. Hereditas 138, 228–236. Assefa, K., Tefera, H., Merker, A., 2002b. Variation and inter-relationships of quantitative traits in tef (Eragrostis tef (Zucc.) Trotter) germplasm from western and southern Ethiopia. Hereditas 136, 116–125. Assefa, K., Tefera, H., Merker, A., Kefyalew, T., Hundera, F., 2001b. Quantitative trait diversity in tef [Eragrostis tef (Zucc.) Trotter] germplasm from central and northern Ethiopia. Genet. Resour. Crop Evol. 48, 53–61. Assefa, K., Tefera, H., Merker, A., Kefyalew, T., Hundera, F., 2001c. Variability, heritability and genetic advance in pheno-morphic and agronomic traits of tef (Eragrostis tef (Zucc.) Trotter) germplasm from eight regions of Ethiopia. Hereditas 134, 103–113. Assefa, K., Yu, J.K., Zeid, M., Belay, G., Tefera, H., Sorrells, M.E., 2010. Breeding tef [Eragrostis tef (Zucc.) Trotter]: conventional and molecular approaches. Plant Breed., doi:10.1111/j.1439-0523.2010.01782.x. Ayele, M., Nguyen, H.T., 2000. Evaluation of amplified fragment length polymorphism markers in tef, Eragrostis tef (Zucc.) Trotter, and related species. Plant Breed. 119, 403–409. Ayele, M., Tefera, H., Assefa, K., Nguyen, H.T., 1999. Genetic characterization of two Eragrostis species using AFLP and morphological traits. Hereditas 130, 33–40. Bai, G., Ayele, M., Tefera, H., Nguyen, H.T., 2000. Genetic diversity in tef [Eragrostis tef (Zucc.) Trotter] and its relatives as revealed by random amplified polymorphic DNAs. Euphytica 112, 15–22. Bai, G., Tefera, H., Ayele, M., Nguyen, H.T., 1999a. A genetic linkage map of tef [Eragrostis tef (Zucc.) Trotter] based on amplified fragment length polymorphism. Theor. Appl. Genet. 99, 599–604. Bai, G.H., Ayele, M., Tefera, H., Nguyen, H.T., 1999b. Amplified fragment length polymorphism analysis of tef [Eragrostis tef (Zucc.) Trotter]. Crop Sci. 39, 819–824. Belay, G., Tefera, H., Getachew, A., Assefa, K., Metaferia, G., 2008. Highly client-oriented breeding with farmer participation in the Ethiopian cereal tef [Eragrostis tef (Zucc.) Trotter]. Afr. J. Agric. Res. 3, 022–028. Benham, J., Jeung, J.U., Jasieniuk, M., Kanazin, V., Blake, T., 1999. Genographer: a graphical tool for automated fluorescent AFLP and microsatellite analysis. J. Agric. Genomics 4, 399. Berhe, T., 1975. A break-through in tef breeding techniques. Int. Bull. Cereal Improv. Prod. 12, 11–13. Botstein, D., White, R.L., Skolnick, M., Davis, R.W., 1980. Construction of a genetic linkage map in man using restriction fragment length polymorphisms. Am. J. Hum. Genet. 32, 314. Caldicott, J.J.B., Nuttall, A.M., 1979. A method for the assessment of lodging in cereal crops. J. Nat. Inst. Agric. Bot. 15, 88–91. Chao, S., Zhang, W., Dubcovsky, J., Sorrells, M., 2007. Evaluation of genetic diversity and genome-wide linkage disequilibrium among US wheat (Triticum aestivum L.) germplasm representing different market classes. Crop Sci. 47, 1018. Dalziel, J.M., 1937. The Useful Plants of West Tropical Africa. (Appendix to the Flora of West Tropical Africa). Crown Agents for the Colonies, London. de Oliveira Borba, T.C., Brondani, R.P.V., Rangel, P.H.N., Brondani, C., 2009. Microsatellite marker-mediated analysis of the EMBRAPA rice core collection genetic diversity. Genetica 137, 293–304. Demissie, A., 2001. Tef genetic resources in Ethiopia. In: Tefera, H.G.B., Sorrells, M. (Eds.), Narrowing the Rift: Tef Research and Development. Ethiopian Agricultural Research Organization, Addis Ababa, pp. 27–31. Donini, P., Stephenson, P., Bryan, G.J., Koebner, R.M.D., 1998. The potential of microsatellites for high throughput genetic diversity assessment in wheat and barley. Genet. Resour. Crop Evol. 45, 415–421. Ebba, T., 1975. Tef (Eragrostis tef) cultivars. Morphology and classification. Part II. Agric. Expt. Sta. Bull. No. 66. Addis Ababa University, College of Agriculture, Dire Dawa, Ethiopia. Garris, A.J., Tai, T.H., Coburn, J., Kresovich, S., McCouch, S., 2005. Genetic structure and diversity in Oryza sativa L. Genetics 169, 1631. Gower, J.C., 1966. Some distance properties of latent root and vector methods used in multivariate analysis. Biometrika 53, 325. Hirano, R., Jatoi, S.A., Kawase, M., Kikuchi, A., Watanabe, K.N., 2009. Consequences of ex situ conservation on the genetic integrity of germplasm held at different gene banks: a case study of bread wheat collected in Pakistan. Crop Sci. 49, 2160. Hopman, E.G.D., le Cessie, S., von Blomberg, B.M.E., Mearin, M.L., 2006. Nutritional management of the gluten-free diet in young people with celiac disease in the Netherlands. J. Pediatr. Gastroenterol. Nutr. 43, 102. Ingram, A.L., Doyle, J.J., 2003. The origin and evolution of Eragrostis tef (Poaceae) and related polyploids: evidence from nuclear waxy and plastid rps16. Am. J. Bot. 90, 116–122. Jones, B.M.G., Ponti, J., Tavassoli, A., Dixon, P.A., 1978. Relationships of the Ethiopian cereal tef (Eragrostis tef (Zucc.) Trotter): evidence from morphology and chromosome number. Ann. Bot. (Lond.) 42, 1369–1373. Kashiwagi, T., Togawa, E., Hirotsu, N., Ishimaru, K., 2008. Improvement of lodging resistance with QTLs for stem diameter in rice (Oryza sativa L.). Theor. Appl. Genet. 117, 749–757. Ketema, S., 1993. Tef [Eragrostis tef (Zucc.) Trotter]: Breeding, Genetic Resources, Agronomy, Utilization and Role in Ethiopian Agriculture. Institute of Agricultural Research, Addis Ababa, Ethiopia. Ketema, S., 1997. Tef, Eragrostis tef (Zucc.) Trotter. Promoting the conservation and use of underutilized and neglected crops. 12. Institute of Plant Genetics and Crop Plant Research. Gatersleben/International Plant Genetics Resources Institute, Rome, Italy. Khlestkina, E.K., Varshney, R.K., Röder, M.S., Graner, A., Börner, A., 2007. A comparative assessment of genetic diversity in cultivated barley collected in different decades of the last century in Austria, Albania and India by using genomic and genic simple sequence repeat (SSR) markers. Plant Genet. Resour. 4, 125–133. Liu, K., Muse, S.V., 2005. Power marker: an integrated analysis environment for genetic marker analysis. Bioinformatics 21, 2128. Lynch, M., 1990. The similarity index and DNA fingerprinting. Mol. Biol. Evol. 7, 478. Matus-Cadiz, M.A., Hucl, P., Horak, M.J., Blomquist, L.K., 2004. Gene flow in wheat at the field scale. Crop Sci. 44, 718. Mohammadi, S.A., Khodarahmi, M., Jamalirad, S., Kamali, M.R.J., 2009. Genetic diversity in a collection of old and new bread wheat cultivars from Iran as revealed by simple sequence repeat-based analysis. Ann. Appl. Biol. 154, 67–76. Nandakumar, N., Singh, A.K., Sharma, R.K., Mohapatra, T., Prabhu, K.V., Zaman, F.U., 2004. Molecular fingerprinting of hybrids and assessment of genetic purity of hybrid seeds in rice using microsatellite markers. Euphytica 136, 257–264. Rohlf, F.J., 2009. NTSYSpc: Numerical Taxonomy System. Ver. 2.21h. Exeter Software. Setauket, New York. Sneath, P.H.A., Sokal, R.R., 1973. Numerical Taxonomy. W.H. Freeman, San Francisco, CA. Spaenij-Dekking, L., Kooy-Winkelaar, Y., Koning, F., 2005. The Ethiopian cereal tef in celiac disease. N. Engl. J. Med. 353, 1748–1749. Sundaram, R.M., Naveenkumar, B., Biradar, S.K., Balachandran, S.M., Mishra, B., IlyasAhmed, M., Viraktamath, B.C., Ramesha, M.S., Sarma, N.P., 2008. Identification of informative SSR markers capable of distinguishing hybrid rice parental lines and their utilization in seed purity assessment. Euphytica 163, 215–224. Tai, T.H., Tanksley, S.D., 1990. A rapid and inexpensive method for isolation of total DNA from dehydrated plant tissue. Plant Mol. Biol. Rep. 8, 297–303. Tefera, H., Assefa, K., Belay, G., 2003. Evaluation of interspecific recombinant inbred lines of Eragrostis tef × E. pilosa. J. Genet. Breed. 57, 21–30. Tefera, H., Ketema, S., Tesemma, T., 1990. Variability, heritability and genetic advance in tef (Eragrostis tef (Zucc.) Trotter) cultivars. Trop. Agric. 67, 317–320. Teklu, Y., Tefera, H., 2005. Genetic improvement in grain yield potential and associated agronomic traits of tef (Eragrostis tef). Euphytica 141, 247–254. Tsegaye, B., Berg, T., 2007. Genetic erosion of Ethiopian tetraploid wheat landraces in Eastern Shewa, Central Ethiopia. Genet. Resour. Crop Evol. 54, 715–726. Vavilov, N.I., 1951. The origin, variation, immunity and breeding of cultivated plants. Soil Sci. 72, 482. Wang, T., Shi, Y., Li, Y., Darmency, H., 2009. Testing coexistence and genetic containment for an autogamous crop. Transgenic Res. 18, 809–813. Yu, J., Sun, Q., La Rota, M., Edwards, H., Tefera, H., Sorrells, M.E., 2006a. Expressed sequence tag analysis in tef (Eragrostis tef (Zucc.) Trotter). Genome 49, 365. Yu, J.K., Graznak, E., Breseghello, F., Tefera, H., Sorrells, M.E., 2007. QTL mapping of agronomic traits in tef [Eragrostis tef (Zucc.) Trotter]. BMC Plant Biol. 7, 13. Yu, J.K., Kantety, R.V., Graznak, E., Benscher, D., Tefera, H., Sorrells, M.E., 2006b. A genetic linkage map for tef [Eragrostis tef (Zucc.) Trotter]. Theor. Appl. Genet. 113, 1093–1102. Zeid, M., Belay, G., Mulkey, S., Poland, J., Sorrells, M.E., 2010. QTL mapping for yield and lodging resistance in an enhanced SSR-based map for tef. Theor. Appl. Genet., doi:10.1007/s00122-00010-01424-00124. Zeid, M., Schön, C.C., Link, W., 2003. Genetic diversity in recent elite faba bean lines using AFLP markers. Theor. Appl. Genet. 107, 1304–1314. Zhang, D., Ayele, M., Tefera, H., Nguyen, H.T., 2001. RFLP linkage map of the Ethiopian cereal tef [Eragrostis tef (Zucc.) Trotter]. Theor. Appl. Genet. 102, 957–964.