M. Heckenberger, M. Bohn, D. Klein, and A. E. Melchinger*

Transcription

1 Identification of Essentially Derived Varieties Obtained from Biparental Crosses of Homozygous Lines: II. Morphological Distances and Heterosis in Comparison with Simple Sequence Repeat and Amplified Fragment Length Polymorphism Data in Maize M. Heckenberger, M. Bohn, D. Klein, and A. E. Melchinger* ABSTRACT Tools like doubled haploids, marker-assisted backcrossing, In addition to genetic distances (GDs) based on molecular markers, and genetic engineering allow to add a small morphological traits and heterosis have been proposed as possible number of genes to a protected variety and apply for tools to assess the genetic conformity between putative essentially PVP for this new variety. In addition, it is possible to derived varieties (EDVs) and their initial varieties (IVs). However, select intentionally for lines that are similar to their for maize (Zea mays L.) and other crops, no consensus has been parents. The efforts invested in breeding the original reached regarding methods and thresholds for identification of EDVs, variety can thus be exploited by the breeder of the mainly because reliable benchmark data are lacking. The objectives plagiarized variety without indemnification. The conof this study were to (i) evaluate the power of morphological traits cept of EDVs was, therefore, implemented into the reand heterosis to discriminate between homozygous progenies derived from F vised UPOV convention (UPOV, 1991) and several na-,bc 1, and BC populations (BC backcross), (ii) compare the findings to published data based on simple sequence repeats tional PVP acts. (SSRs) and amplified fragment length polymorphisms (AFLPs), and Regarding the EDV concept, a variety is deemed (iii) draw conclusions about the usefulness of the various criteria for essentially derived from an IV if it is clearly distinguishidentification of EDVs. Morphological distances (MDs) based on 5 able but conforms genetically to the IV. If the extent traits and midparent heterosis (MPH) for 1 traits were observed for of conformity exceeds a certain threshold, the concept a total of 58 European maize inbred lines comprising 38 triplets. A of EDVs indicates that the breeder of the EDV has to triplet consisted of one homozygous line derived from a F,BC 1,or reach an agreement with the breeder of the IV. How- BC population and both parental inbreds. In addition, all inbreds ever, no consensus has currently been reached on the were genotyped with 100 uniformly distributed SSR markers and 0 methods for determining the genetic conformity to dis- AFLP primer combinations in companion studies for calculation of GDs. Correlations between the coancestry coefficient, GDs, MDs, tinguish between EDVs and independently derived variand MPH were significant and high for the majority of traits. However, eties (IDVs). In addition, accepted or nonaccepted thresholds for EDVs to discriminate between F and BC 1 derived, breeding procedures have not yet been defined. or BC 1 and BC derived progenies using only MDs or heterosis Molecular markers, especially SSRs and AFLPs, have yielded considerably higher values for Type I ( ) and Type II (1 ) been recommended as appropriate tools for determinerrors than observed with GDs based on SSRs and AFLPs. Conse- ing EDVs in various crops including maize (Dillmann quently, morphological traits and heterosis are less suited for identifi- et al., 1997; Roldan-Ruiz et al., 000a; Borgo et al., cation of EDVs in maize than molecular markers. 00). By contrast, the use of morphological traits or heterosis is still under debate (International Association P of Plant Breeders for the Protection of Plant Varieties, lant variety protection (PVP) systems and their 1999). Until now, accurate morphological and agrolaws and regulations should balance commercial in- nomic descriptions of cultivars and varieties have been terests against sustainable development of new cultivars. the basis of tests for distinctness, uniformity, and stabil- On the one hand, registered plant varieties need to be ity (DUS) within worldwide PVP systems and have asprotected against plagiarism and misuse. On the other, sured farmers and breeders that they are using clearly protected germplasm should be accessible to secure fu- identifiable varieties to high standards of purity and qualture breeding progress. The concept of breeder s exemp- ity (Smith and Smith, 1989a). In addition, numerous tion was, therefore, introduced into the International studies showed significant correlations between MPH Union for the Protection of New Varieties of Plants and the coefficient of parentage (f) (Melchinger, 1999; (UPOV) convention to solve the obvious conflict be- Smith et al., 1991). For these reasons, proponents of the tween the different stakeholders within the PVP system use of morphological traits or heterosis for identifica- (UPOV, 1978). Accordingly, plant breeders have access tions of EDVs state that phenotypic information proto protected germplasm for the development of new va- vides the basis for PVP and should also be used for rieties. identification of EDVs. Studies on the ability of morphological traits to estimate the genetic conformity be- M. Heckenberger, D. Klein, and A.E. Melchinger, Institute of Plant Breeding, Seed Science, and Population Genetics, Univ. of Hohenheim, tween related ryegrass (Lolium perenne L.) varieties Stuttgart, Germany; M. Bohn, Crop Science Dep., Univ. of Illinois, S-110 Turner Hall, 110 South Goodwin Avenue, Urbana, Abbreviations: AFLP, amplified fragment length polymorphism; BC, IL Received 3 Feb Crop Breeding, Genetics & Cytology. backcross; DUS, distinctness, uniformity, and stability; EDV, essen- *Corresponding author (melchinger@uni-hohenheim.de). tially derived variety; GD, genetic distance; GR, genetic ratio; IDV, independently derived variety; IV, initial variety; MD, morphological Published in Crop Sci. 45: (005). distance; MPH, midparent heterosis; PIC, polymorphic information doi:10.135/cropsci content; PVP, plant variety protection; RFLP, restriction fragment Crop Science Society of America length polymorphism; SSR, simple sequence repeat; UPOV, Interna- 677 S. Segoe Rd., Madison, WI USA tional Union for the Protection of New Varieties of Plants. 113

2 HECKENBERGER ET AL.: MORPHOLOGICAL DISTANCES AND HETEROSIS 1133 revealed that they had only a limited power to distinracy. parental lines to guarantee heterosis estimates with high accuguish between IDVs and EDVs (Gilliland et al., 000; All trials received standard cultural practices of fertiliza- Roldan-Ruiz et al., 000b). tion as well as control of insects and weeds. In maize, a triangular instead of a linear relationship The experimental unit was a three-row plot with a row spacing of 0.75 m and a plot length of 4.0 m. Trials were was observed between MDs and GDs or the coancestry overplanted and later thinned manually to 6 plants per row, coefficient (f) (Dillmann and Guérin, 1998), which indiwith a final plant density of 8.7 plants m. Each row was cates that low GDs correspond necessarily with low harvested separately. To reduce neighbor effects between ad- MDs, whereas the reverse does not necessarily hold true jacent plots with different vigor (inbreds vs. hybrids), only because high GDs can correspond with both high and data of the middle row of each plot were used for further low MDs. In addition, genetic relationships among analyses. The experiment was performed using a randomized maize inbred lines based on morphology were essen- complete block design. Parameter values were observed for tially random compared with any relation derived from 3 morphological traits according to the UPOV guidelines heterosis or pedigree data (Smith and Smith, 1989b). (UPOV, 1978), and for six additional agronomic traits (Ta- However, data on the usefulness of heterosis or morphoparticular plot or by visual observation of the whole plot. ble 1) by measuring a minimum of five individual plants of a logical traits that reflect the degree of relatedness between maize inbred lines in terms of essential derivation is scanty. Molecular Analyses The main goal of this study was to investigate the All lines were genotyped with a set of 100 SSR markers relationship of homozygous progeny lines in maize de- uniformly covering the entire maize genome, as described in rived from F,BC 1,orBC populations to their parental detail by Heckenberger et al. (00). The 100 SSRs were inbreds based on heterosis and MDs in comparison with selected on the basis of reliable single-locus amplification, SSR- and AFLP-based GDs. In detail, our objectives absence of null alleles, high degree of polymorphism, and were to (i) evaluate the power of heterosis and MDs to high reproducibility of the bands. The SSR analyses were discriminate between progenies derived from F performed on a commercial basis. In addition, all lines were,bc 1, genotyped using commercial AFLPs. A total of 0 AFLP and BC populations, (ii) compare the findings to pubprimer combinations (PCs) was used as described in detail by lished data based on SSRs and AFLPs, and (iii) draw Heckenberger et al. (003). The AFLP markers were mapped conclusions about the usefulness of the various distance according to a proprietary integrated map of maize, as demeasures for the identification of EDVs. scribed by Peleman et al. (000). MATERIALS AND METHODS Plant Materials A total of 58 elite maize inbred lines was analyzed comprising 4 European flint and 34 European dent lines. These lines originated from the maize breeding programs at the University of Hohenheim (Stuttgart, Germany) and two commercial breed- ing companies in Germany. The 58 lines comprised 38 triplets. A triplet consisted of one progeny line O and both parental lines P1 and P. The materials consisted of 6 intrapool triplets of European dent and 1 intrapool triplets of European flint lines. European flint and dent lines, as well as the pooled U.S. lines, were regarded as separate germplasm pools. Progenies were either derived from F,BC 1,orBC populations. This is a subset of the material analyzed in our companion study (Heckenberger et al., 005). For each combination of lines within a triplet (P1 P, P1 O, and P O), seeds from the corresponding F 1 hybrid were generated. In addition, if more than one progeny line (O 1,O,., O j ) was derived from a cross of the same two parental lines, each possible F 1 hybrid (O 1 O,,O j 1 O j ) was generated. In total, 114 intrapool F 1 hybrids were tested in this study. Detailed information on all 38 triplets, the 58 maize inbreds, and the hybrids included in this study is available as supplemental data on the internet at Experimental Design Field experiments were conducted in 000 and 001 at three locations in South Germany, with two replications per loca- tion. All sites (Bad Krozingen, Eckartsweier, and Scherzheim) are located in the Upper Rhine Valley, a major area of grainmaize production in Germany. All inbred lines and hybrids of a triplet were grown together in one block. Within each triplet block, F 1 hybrids were grown side-by-side with their Statistical Analyses Grain yield for each single row was adjusted to 84.5% dry matter content. Heterosis was determined as MPH (F 1 MP)/MP, where F 1 is the F 1 hybrid performance, and MP (P 1 P )/, the midparent value in which P 1 and P are the performances of the inbred parents. Analyses of variance were performed for morphological traits of the inbreds using a mixed linear model considering genotypes as fixed effects, and environments as well as genotype environment interactions as random effects. Genotypic variance among lines ( g ) and the variance of genotype environment interaction ( ge ) were defined and estimated according to Scheffé (1959). Similarly, genetic ratios (GR) analogous to broad-sense heritability were calculated as GR g g er/ms g, where e is the number of envi- ronments, r is the number of replications, and MS g is the mean square due to genotypes. Dhillon et al. (1990) may be consulted for further details. Likewise, ANOVAs were calculated for MPH values of each triplet to estimate the genotypic vari- ance for midparent values among triplets ( t ) and corresponding genetic ratio (GR t ), using the same procedures. For calculation of MDs, observations for each trait were standardized by dividing by the phenotypic standard deviation of the particular trait. Euclidean (MD EUC ) and Mahalanobis (1936) (MD MAH ) distances were calculated based on standard- ized observations for each pairwise comparison of inbred lines. Malécot s (1948) coancestry coefficient (f) was calculated between all pairwise line combinations. Polymorphic information content (PIC) values were estimated as suggested by Anderson et al. (1993). Genetic distances between lines based on SSR (GD SSR ) or AFLP (GD AFLP ) data were estimated using Rogers distance (Rogers, 197). The linear relationships between 1 f, GDs, MDs, and heterosis estimates were evalu- ated with a lack-of-fit test (Snedecor and Cochran, 1980). Empirical and approximated frequency distributions of MD

3 1134 CROP SCIENCE, VOL. 45, MAY JUNE 005 Table 1. Morphological and agronomic traits, their genotypic variance ( g) and genetic ratios (GR g ) observed for 58 flint and dent maize lines in four or six environments in South Germany. UPOV No. of Trait Code code test sites g GR g Ear diameter, mm EDI ** 96. number of kernels NKE ** 88.6 type of grain (1 9 scale) TGR ** 97.4 anthocyanin coloration of glumes of cob AGC ** 98.4 anthocyanin coloration of silks (1 9 scale) ACS ** 87.7 color of dorsal side of grain (1 9 scale) CDG ** 86.6 color of tip of grain (1 9 scale) CTG ** 96.5 length, mm ELG ** 9.6 number of rows of grain NGR ** 9.8 days to silk emergence TSE ** 96.4 Kernels thousand kernel weight, g TKW ** 95.4 grain yield, Mg ha 1 GYD ** 79.4 grain yield of four manually harvested ears, g GYE 6 3.7** 90.4 Leaf angle between blade and stem (1 9 scale) LAN ** 88.4 attitude of blade (1 9 scale) LAT ** 90.1 width of blade, mm WBL ** 81.3 Plant ear height, cm EHT ** 95.4 length, cm PLG ** 94.7 Tassel anthocyanin coloration of base of glume (1 9) ABG ** 94.7 anthocyanin coloration of glume excluding base (1 9) AEB ** 88.9 length of side branches, cm LSB ** 86.6 angle between main axis and lateral branches (1 9) TAN ** 88.7 anthocyanin coloration of anthers (1 9) AAH ** 85.0 attitude of lateral branches (1 9) ALB ** 80.6 length of main axis above lowest side branch, cm TLL ** 84.7 length of main axis above upper side branch, cm TLU ** 83. number of primary lateral branches NLB ** 91. days to anthesis TAH ** 96.7 ** Significant at the 0.01 probability level. UPOV International Union for the Protection of New Varieties of Plants. GR g genetic ratio on a line-mean basis pooled across flint and dent lines. values were compared with a Kolmogorov-Smirnov test (Leh- RESULTS mann, 1986) to check for significant deviations. Furthermore, simple correlations (r) were calculated between 1 f, GDs, Morphological Traits and Heterosis Data MDs, and heterosis estimates. Homogeneity of variance com- Estimates of genotypic variances ( g ) pooled across ponents of data from flint and dent germplasm was evaluated flint and dent inbred lines were significant (P 0.01) with Levene s test (Levene, 1960). Variance components and for all traits (Table 1). In addition, significant (P 0.01) correlations were not significantly different between flint and genotype environment interactions ( ge ) were obdent lines. Consequently, only results from pooled data were served for most traits due to cool and wet weather condireported. To evaluate potential EDV thresholds, the cumulative fre- tions in 000 and hot and dry weather conditions in 001 quency distributions for GDs were approximated by distrismaller than g. (data not shown). In most cases, ge was considerably butions (Johnson et al., 1995), as described in detail in our companion study (Heckenberger et al., 005). Frequency disfor MPH were observed for most traits (Table ). How- Significant (P 0.01) estimates of t among triplets tributions for MDs and MPH for F, BC 1, or BC derived progeny lines were approximated by normal distributions with ever, considerable differences were found between traits parameters chosen such that the mean and variance of the depending on the relative amount of MPH with highest original distribution were conserved. On the basis of these values for grain yield (GYD), grain yield of hand hardistributions, we calculated Type I ( ) and Type II ( ) errors vested ears (GYE), number of kernels per ear (NKE), for various EDV thresholds and various types of populations, and plant length (PLG). The GR for MPH of heterotic as suggested by Heckenberger et al. (005) for molecular traits ranged from 0.66 to 0.97 and were slightly smaller marker data. Here, corresponds to the probability that a true than for line per se performance. IDV will be wrongly judged as EDV, whereas corresponds to the probability that a true EDV will not be recognized as such and judged as IDV. We first investigated the situation that Simple Sequence Repeat and Amplified af derived progeny would be considered as IDV, and a Fragment Length Polymorphism Marker Data BC 1 derived progeny as EDV. Alternatively, we regarded a A total of 580 SSR alleles and 1053 polymorphic BC 1 derived progeny as IDV, and a BC derived progeny AFLP bands was identified for the set of 58 maize lines. as EDV. The number of alleles per SSR ranged from 3 to 1, Statistical analyses of marker data and f values were perwith a mean of 5.9. The PIC values for SSRs varied formed as described by Heckenberger et al. (005) using the PLABSIM software package (Frisch et al., 000). The ANOVAs between 0.08 and 0.86, and averaged The number for field experiments were calculated with the PLABSTAT of polymorphic bands per AFLP primer combination software (Utz, 001). All other statistical calculations were varied from 40 to 70, with an average of 54. The PIC performed with the R software package (Ihaka and Gentleman, 1996). 0.50, with a mean of values for individual AFLP bands ranged from 0.03 to 0.33.

4 HECKENBERGER ET AL.: MORPHOLOGICAL DISTANCES AND HETEROSIS 1135 Table. Estimates of mean, range, genotypic variance ( t), and genetic ratio of midparent heterosis (GR t ) among triplets observed for different morphological and agronomic traits of 114 flint and dent hybrids and their parental lines tested in four or six environments in South Germany as well as correlations (r) of MPH with coancestry (1 f ), genetic distance based on 100 simple sequence repeats (GD SSR ) or 0 amplified fragment length polymorphism primer combinations (GD AFLP ), and Euclidean (MD EUC ) or Mahalanobis (MD MAH ) morphological distances. MPH r Trait Test sites Mean Min. Max. t GR t 1 f GD SSR GD AFLP MD EUC MD MAH EDI ** 0.78** 0.76** 0.65** 0.55** NKE ** ** 0.86** 0.87** 0.69** 0.60** ELG ** ** 0.85** 0.89** 0.68** 0.63** NGR ** ** 0.56** 0.50** 0.43** 0.39** TSE ** ** 0.66** 0.70** 0.58** 0.51** TKW ** ** 0.71** 0.76** 0.55** 0.53** GYD ** ** 0.84** 0.86** 0.7** 0.61** GYE ** ** 0.90** 0.9** 0.66** 0.63** EHT ** ** 0.78** 0.80** 0.58** 0.57** PLG ** ** 0.85** 0.87** 0.63** 0.55** TAH ** 0.76** 0.74** 0.49** 0.4** ** Significant at the 0.01 probability level. For abbreviations, see Table 1. GR t Genetic ratio on a triplet-mean basis for midparent heterosis pooled across flint and dent lines. Relationships among Distance Measures, Heterosis, and Coancestry Threshold Scenarios for Identification of EDVs Estimates of MD EUC and MD MAH for F, BC 1, and BC derived progenies were approximately normally distributed in the joint analysis of flint and dent lines. Considerable overlaps between the frequency distributions of MD EUC and MD MAH were observed for F vs. Table 3. Simple correlations between coancestry coefficient (1 f ), genetic distances based on 100 simple sequence repeats (GD SSR ) and 0 amplified fragment length polymorphism primer combinations (GD AFLP ), as well as Euclidean (MD EUC ) and Mahalanobis (MD MAH ) morphological distances based on 5 traits (see Table 1) for 4 flint (below diagonal) and 34 dent inbreds (above diagonal). Correlations (r) between 1 f and GDs based on SSRs (GD SSR ) and AFLPs (GD AFLP ) were highly signifi- cant (P 0.01) and exceeded 0.85 in both flint and dent lines, with a single exception (Table 3). By comparison, r values between GDs and MDs were moderate (0.40 r 0.68). Likewise, r values between MD EUC and MD MAH were only of moderate sizes. Correlations were consistently higher for flint lines than for dent lines. Coancestry was moderately correlated with MD EUC, but poorly correlated with MD MAH ; both relationships showed a triangular form (Fig. 1). In contrast, the relationships between GD SSR,GD AFLP, and 1 f on the one hand, and MPH on the other, were linear for most heterotic traits (Fig. ). Corresponding r values were highly significant (P 0.01) and moderate to high, depending on the trait (Table ). In general, these correlations were considerably higher than those of MD EUC or MD MAH with GD SSR,GD AFLP,or1 f. BC 1, as well as for BC 1 vs. BC derived progenies (Fig. 3). For thresholds based on MDs, the power 1 to classify a BC 1 derived progeny line as EDV amounted to 18% for MD EUC and 3% for MD MAH, when choosing 0.05 for F derived lines (Table 4). Assuming 0.05 for BC 1 derived lines, corresponding values of 1 for BC derived lines were considerably higher for 1 f GD SSR GD AFLP MD EUC MD MAH 1 f 0.75** 0.85** 0.58** 0.31** GD SSR 0.88** 0.9** 0.57** 0.40** GD AFLP 0.88** 0.97** 0.68** 0.40** MD EUC 0.55** 0.65** 0.65** 0.6** MD MAH 0.44** 0.49** 0.59** 0.76** ** Significant at the 0.01 probability level. Fig. 1. Relationship of coancestry coefficient (f) with (A) Euclidean and (B) Mahalanobis distances based on 5 morphological traits observed for 1767 pairwise comparisons of maize inbred lines.

5 1136 CROP SCIENCE, VOL. 45, MAY JUNE 005 Fig.. Relationship of genetic distances based on 100 simple sequence repeats (GD SSR ) or 0 AFLP primer combinations (GD AFLP ) with midparent heterosis of 84 intrapool hybrids with given pedigree relationships of their parental maize lines.

6 HECKENBERGER ET AL.: MORPHOLOGICAL DISTANCES AND HETEROSIS 1137 Fig. 3. Cumulative histograms (columns) and approximated normal distributions (curves) for (A) Euclidean or (B) Mahalanobis morphological distances based on 5 morphological traits for F, BC 1, and BC derived progeny lines. Variables n,, and SD refer to the number of values, the mean, and the standard deviation of Mahalanobis distance values for the particular distribution, respec- tively. MD EUC and MD MAH. The power 1 for thresholds determined by to classify BC 1 orbc derived progenies as EDVs increased considerably compared with the values of This increase in 1 was associated with higher values of. When potential thresholds were based on MPH, the power 1 to classify a BC 1 derived progeny line as EDV ranged from to 30%, assuming 0.05 between F derived lines (Table 4). Choosing 0.05 for BC 1 derived lines, the values of increased for BC derived lines. For, the power to classify BC 1 orbc derived progenies as EDVs increased substantially; however, this was again associated with higher values of. In general, values of and were of similar magnitude for MDs and MPH. DISCUSSION The maize inbred lines examined in our study represent a cross-section of present-day elite flint and dent inbred lines from commercial and public maize breeding programs in Germany. Morphological traits were chosen according to the UPOV guidelines of DUS. Heterosis and morphological traits were determined in extensive field trials across yr and three locations, thus exceeding by far the number of environments employed for DUS testing within PVP systems. Furthermore, SSRs were selected as a suitable marker system due to their known map positions and high degree of polymorphism. The AFLPs were chosen due to the greater number of markers per assay unit and their high reproducibility (Heckenberger et al., 003). Thus, the present study is the first larger investigation after a series of pioneering studiess based on isozymes and restriction fragment length polymorphisms (RFLPs) (Smith and Smith, 1989a, 1989b; Smith et al., 1991) to provide critical data on the ability of MDs and heterosis for the identification of EDVs in maize in direct comparison with SSR and AFLP data. For this reason, our results provide a well-founded comparison of different distance measures for the identification of EDVs in maize and may serve as an example for other crops. Data Quality and Relatedness between Different Measures for Genetic Conformity Despite the contrasting climatic conditions during the vegetation seasons in 000 and 001, high GR values were observed for morphological traits and MPH, the former being considerably higher than those reported by Rebourg et al. (001). In addition, UPGMA cluster analysis based on MD EUC showed a clear grouping of flint and dent lines, which further corroborates the high quality of morphological data (available as supplemental data to the online version of this paper; see crop.scijournals.org/). However, the dendrogram based on MD MAH deviated considerably from the expectations based on pedigree data or GDs, and only moderate correlations between MD EUC and MD MAH were observed. This can be explained by the different statistical properties of MD EUC and MD MAH, because MD MAH adjusts for the correlations of traits. The graphs between MDs and GDs (Fig. 1) confirmed the triangular relationship between morphological and GDs reported in previous studies (Dillmann et al., 1997; Rebourg et al., 001). This indicates that low GDs corre- spond necessarily with low MDs, whereas the reverse does not necessarily hold true because high GDs can correspond with both high and low MDs (van Eeuwijk and Baril, 001). In addition, the triangular shape has several biological explanations (Nuel et al., 001) and is also expected if only molecular markers tightly linked with the genes controlling the phenotypic trait(s) were used (Burstin and Charcosset, 1997). In the present investigation, flint and dent lines showed similar estimates of ˆ g and correlations among the vari- ous criteria, and the same applied to flint and dent triplets. This is in harmony with a previous study of Bar- Hen et al. (1995), who examined 974 maize inbred lines with morphological traits and RFLPs. Correlations of 1 f, GD SSR,orGD AFLP, with MPH were higher than

7 1138 CROP SCIENCE, VOL. 45, MAY JUNE 005 Table 4. Determination of EDV thresholds (T, measured in the scale of the particular trait) and power (1 ) for various conceivable EDV scenarios (F vs. BC 1 ;BC 1 vs. BC ; 0.05; ) based on morphological distances, midparent heterosis, and genetic distances (GDs) calculated from simple sequence repeats (SSRs) and amplified fragment length polymorphisms (AFLPs). F vs. BC 1 BC 1 vs. BC Parameter T 1 T T 1 T Morphological distances Euclidean (MD EUC ) Mahalanobis (MD MAH ) Heterosis Grain yield (GYE), Mg ha Plant length (PLG), m Number of kernels per ear (NKE) Cumulative Genetic distances 100 SSRs (GD SSR ) AFLP PCs (GD AFLP ) Average relative heterosis of five traits (GYE, ELG, NKE, PLG, EHT) showing highest correlation with 1 f; for abbreviations, see Table 1. reported by Ajmone Marsan et al. (1998) for AFLPs based on molecular markers have clear advantages for but similar to correlations of MPH with 1 f and GDs identification of EDVs. based on RFLPs in intrapool crosses (Boppenmaier et al., 1993; Smith et al., 1990). In addition, our study con- Power of Morphological Distances and firms the findings of Smith and Smith (1989b) that corre- Heterosis for Identification of EDVs lations of molecular markers or 1 f are considerably For MD EUC as well as for MD MAH, extensive overlaps higher with MPH than with MD EUC or MD MAH. of the frequency distributions of F, BC 1, and BC Distinctness versus Conformity derived progenies were found in spite of the significant correlations with 1 f. Type I ( ) and Type II ( ) To confirm an essential derivation in the sense of the errors observed for MDs were thus considerably higher UPOV convention (UPOV, 1991), three separate criteria than observed for GDs based on molecular markers must be fulfilled. An EDV must (i) be distinct from the (Table 4). Consequently, MDs provide only a rough IV, (ii) be predominantly derived from the IV, and (iii) estimate of the true relatedness between two lines and conform to it in the expression of its essential characterprogenies. These results confirm data from ryegrass (Gilli- can only poorly discriminate F, BC 1, and BC derived istics. Distinctness can be determined based on morpholand et al., 000) and maize (Smith and Smith, 1989b), logical traits by established procedures for DUS testing. showing that morphological conformity could give an Establishing a predominant derivation will either reinitial indication of the relatedness between two cultiquire a directly documented evidence, for example, by vars, particularly for highly conforming pairs of inbreds. breeding books, or could be determined with molecular Nevertheless, a small MD between two varieties cannot evidence similar to forensic approaches in the human be taken as a definitive proof that they are in fact closely sector (Gill et al., 1995). However, the question whether related because of the triangular relationship between conformity in the expression of essential characteristics 1 f and MDs. should be assessed by phenotypic rather than molecular In contrast to the triangular relationship between GDs data is still unsolved. While differences in the expression and MDs, a linear relationship of MPH with GDs or of one single trait are sufficient to prove distinctness 1 f was observed, as expected by quantitative genetic between two varieties, assessment of conformity should theory (Melchinger, 1999). However, in spite of the be based on a large number of morphological traits and higher correlation between MPH and 1 f or GDs, still could not give a definite answer due to the triangular MPH was not markedly superior to MDs regarding the relationship mentioned above. power to discriminate between F, BC 1 or BC Proponents of phenotypic data state that the term derived progeny. This is attributable to the larger exper- conform in the expression of its essential characteristics imental error and genotype environment interactions that result from the genotype (UPOV, 1991) implies of MPH in comparison with line per se performance the use of phenotypic and agronomic data rather than (data not shown), as reflected by the comparison of GR t molecular data (Troyer and Rocheford, 00). In con- vs. GR g. trast, opponents state that even highly heritable pheno- For nearly all of the examined scenarios, GDs based typic traits can only offer an indirect measure of the on SSRs or AFLPs were superior to MDs or MPH for relatedness between two cultivars (Roldan-Ruiz et al., any trait or combination of traits in their power 1 000b). Molecular data provide a direct estimate of the to discriminate among F, BC 1, and BC derived progtrue relatedness between two genotypes because they enies for given values of. However, different from are unbiased by environmental effects and reflect the MDs and MPH, the use of SSR or AFLP markers would percentage of the genome in common between the IV require thresholds specific for a given germplasm pool. and a putative EDV. On the basis of our results, GDs This is necessary because flint and dent lines differ sig-

8 HECKENBERGER ET AL.: MORPHOLOGICAL DISTANCES AND HETEROSIS 1139 nificantly in their mean GD between unrelated lines inbreds. III. Performance of crosses within versus between heterotic groups for grain traits. Plant Breed. 111:17 6. due to the different levels of polymorphism within each Borgo, L., B. Serani, and S. Conti. 00. Molecular profiling of genetigermplasm pool (Heckenberger et al., 005). The lower cally modified maize lines as related to the issue of essential deriva- T values for AFLPs compared with SSRs are due to tion. Poster presented at the Annual Congr. of the Italian Society their lower degree of polymorphism, but this has no of Agricultural Genetics, Giardini Naxos, Italy Sept. 00. influence on the ability to discriminate different types Italian Society of Agricultural Genetics, Portici. Burstin, J., and A. Charcosset Relationship between phenotypic of progeny. and marker distances: Theoretical and experimental investigations. Heredity 79: CONCLUSIONS Dhillon, B.S., C. Paul, E. Zimmer, P.A. Gurrath, D. Klein, and W.G. Pollmer Variation and covariation in stover digestibility traits in diallel crosses of maize. Crop Sci. 30: On the basis of our results, MDs and MPH can pro- Dillmann, C., A. Bar-Hen, D. Guerin, A. Charcosset, and A. Murigvide only a vague indication for putative EDVs, but neux Comparison of RFLP and morphological distances reliable identification of EDVs by MPH or MDs is not between maize (Zea mays L.) inbred lines Consequences for ger- possible due to the large overlaps in the frequency distri- mplasm protection purposes. Theor. Appl. Genet. 95:9 10. butions of MDs and MPH of F Dillmann, D., and D. Guérin Comparison between maize inbred, BC 1, and BC derived lines: Genetic distances in the expert s eye. Agronomie 18: progenies. In addition, MDs and MPH have several Frisch, M., M. Bohn, and A.E. Melchinger Plabsim: Software disadvantages compared with molecular markers. First, for simulation of marker-assisted backcrossing. J. Hered. 91: assessment of morphological traits and MPH requires Gill, P., C.P. Kimpton, A. Urquhart, N. Oldroyd, E.S. Millican, S.K. extensive field trials across several years and locations to Watson, and T.J. Downes Automated short tandem repeat (STR) analysis in forensic casework: A strategy for the future. minimize environmental effects. These measurements Electrophoresis 16: are therefore more expensive and time consuming than Gilliland, T.J., R. Coll, E. Calsyn, M. de Loose, M.J.T. Van Eijk, molecular marker analyses. Second, heterosis estimates and I. Roldan-Ruiz Estimating genetic conformity between requires production and testing of hybrids. In addition, related ryegrass (Lolium) varieties. 1. Morphology and biochemical reciprocal crosses should be evaluated to minimize the characterisation. Mol. Breed. 6: Heckenberger, M., M. Bohn, and A.E. Melchinger Identificarisk of maternal effects (Melchinger et al., 1986). Third, tion of essentially derived varieties obtained from biparental the scoring of morphological traits is subjective to a crosses of homozygous lines. I. Simple sequence repeat data from certain extent. Therefore, a number of check inbreds maize inbreds. Crop Sci. 45: , this issue. must be included in the study to warrant a high quality Heckenberger, M., M. Bohn, J.S. Ziegle, L.K. Joe, J.D. Hauser, M. of morphological traits across different years and scoramong accessions within maize inbred lines and implications for Hutton, and A.E. Melchinger. 00. Variation of DNA fingerprints ing persons. identification of essentially derived varieties. I. Genetic and techni- In conclusion, we strongly recommend the use of mo- cal sources of variation in SSR data. Mol. Breed. 10: lecular markers for the identification of suspected EDVs Heckenberger, M., J. Rouppe van der Voort, A.E. Melchinger, J. by fingerprinting with at least 100 SSR markers or 0 Peleman, and M. Bohn Variation of DNA fingerprints among accessions within maize inbred lines and implications for identifica- AFLP PCs. If doubts still remain whether a variety is tion of essentially derived varieties. II. Genetic and technical IDV or EDV, the corresponding genotypes should be sources of variation in AFLP data and comparison to SSR data. fingerprinted with an additional set of SSRs or AFLPs. Mol. Breed. 1: Use of MPH for the identification of EDVs is problemand Ihaka, R., and R. Gentleman R: A language for data analysis graphics. J. Comput. Graphical Stat. 5: atic because the rationale for using MPH is merely its International Association of Plant Breeders for the Protection of Plant linear relationship with 1 f, and the biological mecha- Varieties Essential derivation and dependence Practical infornisms underlying heterosis are not yet fully understood. mation [Online]. Available at papers/derive.htm [verified 30 Nov. 004]. ASSINSEL, Nyon, Switzerland. ACKNOWLEDGMENTS Johnson, N.L., S. Kotz, and N. Balakrishnan. (ed.) Continuous We are indebted to the Gesellschaft zur Förderung der univariate distributions. Vol.. J. Wiley & Sons, New York. privaten Pflanzenzüchtung (GFP), Germany, for a grant to Lehmann, E.L. (ed.) Testing statistical hypotheses. John Wiley & Sons, New York. support Dr. Bohn and the molecular analyses of the presented Levene, H Robust tests for equality of variances. p study. Financial support for M. Heckenberger and the field In Contributions to probability and statistics: Essays in honor of trials was provided by the European Union Grant No. QLK- Harold Hotelling. Stanford Univ. Press, Stanford, CA. CT (MMEDV). Mahalanobis, P.C On the generalized distance in statistics. Proc. Natl. Acad. Sci. India : Malécot, G Les mathematiques de l heredité. Masson & REFERENCES Cies, Paris. Ajmone Marsan, P., P. Castiglioni, F. Fusari, M. Kuiper, and M. Melchinger, A.E Genetic diversity and heterosis. Int. Symp. Motto Genetic diversity and its relationship to hybrid perfor- on Genetics and Exploitation of Heterosis in Crop Plants. p mance in maize as revealed by RFLP and AFLP markers. Theor. In J.G. Coors and S. Pandey (ed.) The genetics and exploitation Appl. Genet. 96:19 7. of heterosis in crops. ASA, CSSA, and SSSA, Madison, WI. Anderson, J.A., G.A. Churchill, J.E. Autrique, S.D. Tanksley, and Melchinger, A.E., H.H. Geiger, and F.W. Schnell Epistasis in M.E. Sorrells Optimizing parental selection for genetic link- maize (Zea mays L.).. Genetic effects in crosses among early flint age maps. Genome 36: and dent inbred lines determined by three methods. Theor. Appl. Bar-Hen, A., A. Charcosset, M. Bougoin, and J. Guiard Rela- Genet. 7: tionship between genetic markers and morphological traits in a Nuel, G., S. Robin, and C.P. Baril Predicting distance using a maize inbred lines collection. Euphytica 84: linear model: The case of varietal distinctness. J. Appl. Stat. 8: Boppenmaier, J., A.E. Melchinger, G. Seitz, H.H. Geiger, and R.G Herrmann Genetic diversity for RFLPs in European maize Peleman, J., R. van Wijk, J. van Oeveren, and R. van Schaik. 000.

9 1140 CROP SCIENCE, VOL. 45, MAY JUNE 005 Linkage map integration: An integrated genetic map of Zea mays Smith, J.S.C., O.S. Smith, S.L. Bowen, R.A. Tenborg, and S.J. Wall. L. Poster P47. Plant & Animal Genome Conf. VIII, San Diego, The description and assessment of distances between inbred CA. 9 1 Jan lines of maize. III. A revised scheme for the testing of distinctiveness Rebourg, C., B. Gouesnard, and A. Charcosset Large scale between inbred lines utilizing DNA RFLPs. Maydica 36: molecular analysis of traditional European maize populations. Re lationships with morphological variation. Heredity 86: Smith, O.S., J.S.C. Smith, S.L. Bowen, R.A. Tenborg, and S.J. Wall. Rogers, J.S Measures of genetic similarity and genetic distance Similarities among a group of elite maize inbreds as measured p In Studies in Genetics VII. Publ Univ. of by pedigree, F 1 grain yield, grain yield, heterosis, and RFLPs. Theor. Texas, Austin. Appl. Genet. 80: Roldan-Ruiz, I., E. Calsyn, T.J. Gilliland, R. Coll, M.J.T. van Eijk, Snedecor, G., and W. Cochran Statistical methods. Iowa State and M. de Loose. 000a. Estimating genetic conformity between Univ. Press, Ames. related ryegrass (Lolium) varieties.. AFLP characterization. Mol. Troyer, A.F., and T.R. Rocheford. 00. Germplasm ownership: Re- Breed. 6: lated corn inbreds. Crop Sci. 4:3 11. Roldan-Ruiz, I., F.A. van Eeuwijk, T.J. Gilliland, P. Dubreuil, C. UPOV International convention for the protection of new varie- Dillmann, J. Lallemand, M. de Loose, and C.P. Baril. 000b. A ties of plants comparative study of molecular and morphological methods of content.htm [verified 30 Nov. 004]. International Union for the describing relationships between perennial ryegrass (Lolium per- Protection of New Varieties of Plants, Geneva. enne L.) varieties. Theor. Appl. Genet. 103: UPOV International convention for the protection of new varieties Scheffé, H The analysis of variance. John Wiley & Sons, of plants [Online]. Available at New York. conventions/1991/content.htm [verified 30 Nov. 004]. International Smith, J.S.C., and O.S. Smith. 1989a. The description and assessment Union for the Protection of New Varieties of Plants, Geneva. of distance between inbred lines of maize: I. The use of morphological Utz, H.F Plabstat, ein Computerprogramm zur statistischen traits as descriptors. Maydica 34: Analyse von pflanzenzüchterischen Experimenten (in German). Smith, J.S.C., and O.S. Smith. 1989b. The description and assessment Version P vom 14. Univ. Hohenheim, Germany. of distances between inbred lines of maize: II. The utility of mor- van Eeuwijk, F.A., and C.P. Baril Conceptual and statistical phological, biochemical, and genetic descriptors and a scheme for issues related to the use of molecular markers for distinctness and testing of distinctiveness between inbred lines. Maydica 34: essential derivation. Acta Hortic. 546:35 53.

10 HECKENBERGER ET AL.: MORPHOLOGICAL DISTANCES AND HETEROSIS 1141 SUPPLEMENTAL DATA for Crop Science manuscript C (1), Heckenberger et al., 005, Crop Sci. 45( ). Supplemental Table A. Parent offspring triplets and their pedigree relationships. Supplemental Table B. Pedigree relationships among progeny lines developed from the same parental lines. P1 P Progeny f (P1,P) Type No. O1 O f (O1,O) Dent Dent D F 19 P009 P D16 D F 0 D63 D D06 D F 1 D64 D D F D66 D D408 D F 3 UH00 UH D06 D61 P F 4 M006 M D83 D46 UH F 5 D63 D D704 D76 M F 6 D64 D D06 D F 7 D63 D D64 D06 P F Flint D64 D06 P F 8 UH003 F D711 D704 M F 9 UH003 F D711 D704 M F 30 F009 F D711 D701 M F 31 F009 UH D711 D757 M F 3 F011 UH D3 D06 UH F 33 F014 F D3 D06 UH F 34 F004 UH D BC 1 35 F004 F D BC 1 36 UH001 F D BC D BC D16 D BC D17 D BC D BC D BC D17 D BC Flint D146 C1 F F D167 C1 F F CO55 D503 R F D149 D150 F F D149 D150 F F D149 D150 UH F D149 D171 UH F D149 D171 F F D149 D171 F F D147 D171 F F D147 D171 UH F D147 D171 F F

11 114 CROP SCIENCE, VOL. 45, MAY JUNE 005 Fig. A. Continued on next page.

12 HECKENBERGER ET AL.: MORPHOLOGICAL DISTANCES AND HETEROSIS 1143 Fig. A. Dendrograms based on pairwise (A) Euclidean, or (B) Mahalanobis morphological distances from 5 morphological and agronomic traits. Flint and dent lines are indicated by F or D, respectively.