Introduction

Cucumber (C. sativus L.; 2n = 2x = 14) improvement is a complex process, often involving refinement of populations derived from intercrossing elite and/or exotic (unadapted) germplasm, extraction of inbred lines from such populations, and subsequent identification of commercially acceptable F1 hybrids (Staub et al. 2008). Genetic diversity in C. sativus is relatively low (3–8%) when compared with other Cucumis species (10–25%) (Kupper and Staub 1988; Horejsi and Staub 1999; Luan et al. 2008). In fact, the genetic diversity of elite cucumber lines is extremely low (1–3%) (Dijkhuizen et al. 1996; Staub et al. 2002b; Behera et al. 2010), highlighting the need for continued introgression of exotic germplasm to ensure broad-based improvement of cucumber.

The concept of gene pools (primary, secondary, and tertiary) and their utilization for broadening species diversity was proposed by Harlan and de Wet (1971). The primary gene pool of C. sativus comprises primarily two interfertile botanical varieties, var. sativus L. and var. hardwickii R. Alef. (Lebeda et al. 2007). The secondary Cucumis gene pool houses C. hystrix Chakr. (H), which is sparely cross-compatible with C. sativus (C) (Chen et al. 2003) and has been suggested as a source for broadening the genetic base of cucumber (Bates and Robinson 1995). A synthetic amphidiploid species, C. hytivus Chen and Kirkbride 2000 (2n = 4x = 38; HHCC), was developed by interspecific mating between C. sativus (2n = 2x = 14; CC) and C. hystrix (2n = 2x = 24; HH) (Chen and Kirkbride 2000; Chen et al. 2003). Subsequently, allotriploids (2n = 3x = 36) and backcross-derived, fully fertile diploid (2n = 14; BC1S3) derivatives were produced from amphidiploid × diploid (C. sativus var. sativus; recurrent parent) mating. These genetic stocks may be useful for broadening the genetic base of commercial cucumber.

Novel genes, such as those that condition disease resistance [i.e., gummy stem blight (causal agent Didymella bryoniae)], are not found in cultivated cucumber but are present in C. hystrix (Chen et al. 2003). Backcross introgression has been suggested for incorporation of such genes (Wehrhahn and Allard 1965) and has shown potential for improving the genetic diversity and yield of cucumber (Owens et al. 1985).

Moderately saturated linkage maps have been developed for cucumber, and genomic regions associated with economically important quantitative and qualitative trait loci have been identified (Bradeen et al. 2001; Fazio et al. 2003a; Ren et al. 2009). The efficiency and effectiveness of selection during line and population development in cucumber can be increased through MAS (Fan et al. 2006; Robbins and Staub 2009). In fact, backcrossing with concurrent initial molecular-based genotyping and selection for genetic diversity has proven useful in cucumber (Fan et al. 2006; Delannay 2009; Delannay and Staub 2010a; 2010b). Therefore, a study was designed that employed marker-assisted backcrossing to increase genetic diversity in C. sativus by introgression of the C. hystrix genome into elite breeding lines. This process consisted of using mapped and unmapped markers (Fazio et al. 2003a; Kong et al. 2006; Nam et al. 2005; Ritschel et al. 2004) to survey C. hystrix × C. sativus-derived BC1 progeny for the construction of marker-genotyped BC2 progeny. These progenies were then backcrossed and/or successively self-pollinated to produce BC3 or BC2S1 lines, respectively, and compared with nonselected progeny. Such genetically diverse lines will have potential for direct use in cucumber improvement if they possess adequate yield and fruit quality characteristics.

Materials and methods

Parents

Three parents were used herein to produce genetically diverse BC2, BC2S1, and BC3 lines for evaluation. Line WI 7012A (P1), which is a late-flowering, indeterminate, monoecious BC1S3 line derived from a C. hytivus × C. sativus (long-fruited Chinese C. sativus cv. Beijingjietou; recurrent backcross parent) mating, was received from Nanjing Agricultural University, Nanjing, China (Chen et al. 2003). Although this line possesses a sequential fruiting habit and a comparatively high number of primary lateral branches (>4) and yield, it bears warty, light-green fruit of commercially unacceptable shape and quality. Gynoecious determinate line WI 7023A (P2) and indeterminate line WI 9-6A (P3) were created through selection and backcrossing {Gy-7 (recurrent parent; University of Wisconsin) and H-19 [donor parent; University of Arkansas (Fayetteville)]} to identify a small-statured genotype that develops high-quality fruit for once-over mechanical harvest operations (Robbins and Staub 2009). These lines originated from the same populations that were used to develop recombinant inbred lines for the mapping of quantitative trait loci (QTL) in US processing cucumber at the Agricultural Research Service, Madison, WI (Staub et al. 2002a; Robbins and Staub 2009).

Development of backcross families

The inbred backcross line (IBL) breeding method (Wehrhahn and Allard 1965) was employed to introgress C. hystrix genomic components into C. sativus via the C. hytivus-derived line WI 7012A. Line WI 7012A (considered as an inbred but “exotic” line) was crossed with WI 7023A to develop F1 plants, which were subsequently mated with WI 7023A (recurrent parent) to create BC1P2 lines without selection [designated BC1P2 (NSel)] (Fig. 1). Three hundred ninety-two BC1P2 plants were evaluated for fruit yield (number per plant), number of primary lateral branches, fruit length:diameter ratio (L:D), and seed cavity diameter (endocarp:total diameter) ratio (E:T) in a field nursery [Plainfield loamy sand (Typic Udipasamment) soil] at the University of Wisconsin Experimental Station in Hancock, WI (UWESH) during the summer of 2005 (Delannay 2009). Based on this field evaluation, 21 BC1P2 lines (plants) were selected [designated BC1P2 (Sel)] given their comparative fruit yield and quality over six harvests (i.e., base population for subsequent MAS or NSel). These selections were multiplied by stem cutting, grown in the greenhouse to produce healthy mature plants, and then genotyped with 24 DNA markers [8 mapped and 16 unmapped (Delannay 2009); see below]. Based on their comparative genetic diversity, 11 BC1 genotypes were selected for crossing to elite lines WI 7023A (P2) and WI 9-6A (P3), the progeny of which were designated as BC2P2 (Sel) and BC2P3 (Sel), respectively. These MAS BC2 lines were then backcrossed and/or self-pollinated without selection to produce BC2P2S1 (Sel) and BC2P3S1 (Sel) and also backcrossed to create BC3P2 (Sel) and BC3P3 (Sel) progeny (Fig. 1). In addition, non-MAS BC1P2 progeny were backcrossed once to P2 and then self-pollinated to produce BC2P2 (NSel) and BC2P2S1 (NSel) cross-progeny.

Fig. 1
figure 1

Schematic representation of the development of F1 and backcross populations in cucumber using C. hytivus × C. sativus. Evaluated at Hancock, WI in 2007. IBL inbred backcross line, RIL recombinant inbred line (Fazio et al. 2003a), Sel individuals selected for genetic diversity by 24 molecular makers, NSel individuals not receiving marker selection

Full size image

Molecular marker analysis

To assess genetic diversity, parent lines (WI 7012A, WI 7023A, and WI 9-6A) and derived cross-progeny populations were genotyped using nine single-nucleotide polymorphic (SNP; AI4SNP, AT1SNP, D11SNP, L1LSNP, M4LH2SNP, M7LG3SNP, M7LH3SNP, M8SNP, and W7SNP), nine sequence amplified characterized region (SCAR; AK5SCAR, BC231SCAR, BC526SCAR, P14SCAR, S_AV14SCAR, S_AV16-3SCAR, S_AU18-1SCAR, S_E9SCAR, S_M5SCAR), and six simple sequence repeat (SSR; CSWTAAA01SSR, CSWACC02SSR, CSWATT02SSR, CSWCT03SSR, CSWGAAA02SSR, and CSWTAA11BSSR) markers (Fazio et al. 2003a; Kong et al. 2006; Nam et al. 2005; Ritschel et al. 2004). The number of plants examined in each backcross population, along with their percentage polymorphism, average heterozygosity, and number of polymorphic loci, are presented in Table 1.

Table 1 Statistical measures of genetic variation in various cucumber populations as measured by 24 molecular markers
Full size table

Marker analysis was preformed by bulk sampling (at least 10 samples) of young tissue in the three-leaf stage from parental lines and cross-progeny populations, where samples were held at −4°C (~2 h) until transfer to −80°C storage for DNA extraction according to Fazio et al. (2003a). Subsequently, polymerase chain reactions (PCR) and electrophoresis using SNP, SCAR, and SSR makers (as defined above) were performed according to Fazio et al. (2002) and Robbins et al. (2008). Banding morphotypes identified by markers that detected differences between 5 and 30 bp were visualized using 3% agarose gels run at 250 V for 4 to 6 h, and amplicon differences greater than 30 bp were detected using 1.6% agarose gels run at 250 V for 2 h. For codominant markers that detected <5 bp differences, Alexa-labeled 2′-deoxyuridine 5′-triphosphates (dUTPs) was added to the PCR master mix, and then polymorphisms were identified using fragment analysis performed at the University of Wisconsin Biotechnology Center, Madison, WI. Banding and size differences were then analyzed using GeneMarker (version 1.2; Softgenetics, State College, PA).

Open field evaluation

Parental lines (WI 7012A, WI 7023A, and WI 9-6A) and cross-progeny populations as well as ‘Vlasset’ (Seminis Seed Company, Woodland, CA; control) were sown on May 18 in a greenhouse in Madison, WI, and then transplanted to a UWESH field nursery on June 20, 2007. The design was randomized complete block with three replications. Experimental plots consisted of 10 plants of each parental line (three) or cross-progeny family [BC1P2 (NSel), BC1P2 (Sel), BC2P2 (Sel), BC2P3 (Sel), BC3P2 (Sel), BC3P3 (Sel), BC2P2 (NSel), and BC2P2S1 (NSel)] (Table 1) spaced 13 cm apart within rows (5.2 m long) on 1.5-m row centers (~51,000 plants/ha) with end and side borders of ‘Vlasset.’ Data were collected on relative plant vegetative growth, number of primary lateral branches (LB), days to first flower (DF), yield (number of fruit), and fruit L:D and E:T on per-plant basis. Plant vigor in melon is often associated with plant architecture, including sex expression (gynoecious/monoecious), flowering date, and plant stature (determinate/indeterminate) (Lebeda et al. 2007). Given the differing architecture of the parents, in this work vegetative growth (relative size of vines) was scored by visual observation, where disease-free plants were given comparative values from 1 to 4, such that 1 = limited growth (38–43 cm; usually manifest as early flowering in determinate, gynoecious lines), 2 = moderate growth (44–58 cm; usually manifest as early flowering in determinate, predominantly gynoecious lines), 3 = vigorous growth (59–73 cm; usually manifest as late flowering in indeterminate, monoecious lines), and 4 = extremely vigorous grow (>73 cm; usually manifest as extremely late flowering in indeterminate, monoecious lines) (Fig. 2). DF was recorded as the number of days between sowing and the appearance of the first fully expanded corolla. LB was recorded when individual plants reached anthesis, where only LB longer than 5 cm on the first 10 nodes was observed. Fruits per plant, fruit length and width, fruit endocarp width, and total fruit diameter were recorded at each of six harvests at 1-week intervals. Harvest began when a majority of the fruit on a plant were greater than 2 cm in diameter (equivalent to USDA 2A grade). Mean fruit L:D and E:T were obtained by measuring the length and diameter of fruit ranging between 2.5 and 3.0 cm in diameter (equivalent to USDA 2–3A grade). Cumulative numbers of fruit per plant were calculated by dividing the total number of fruit harvested by the number of plants.

Fig. 2
figure 2

Comparative cucumber plant growth recorded as relative score (values 1–4) based on visual observation: 1 limited growth (38–43 cm), 2 moderate growth (44–58 cm), 3 vigorous growth (59–73 cm), and 4 extremely vigorous growth (>73 cm)

Full size image

Statistical analyses

Trait data were subjected to analyses of variance (ANOVA) using a mixed models procedure (PROC Mixed) to define blocks, and germplasm (parents and cross-progeny) effects, and genotype-by-environment interactions using SAS software (SAS 2003; Littell et al. 1996). To determine trait relationships, pairwise phenotypic Pearson correlations were calculated using SAS (2003). Data from each planting were initially combined for ANOVA to define block, and line (parents and cross-progenies) effects using the “proc glm” procedure in SAS. All variables were treated as random effects, and least-square means were calculated for each line using the “lsmeans” option in the “proc glm” procedure in SAS.

Based on allelic differences at marker loci, gene frequency, percent polymorphic loci, and mean heterozygosity were estimated using Tools for Population Genetic Analyses (TFPGA) version 1.3 software (Miller 1997). Estimates of genetic distance (GD; Nei 1973) and population differentiation (G ST; McDermott and McDonald 1993) were calculated using POPGENE version 1.32 software (Yeh and Boyle 1997).

Results and discussion

The marker-based genetic and phenotypic analyses conducted herein formed the basis for comparative examination of backcross progeny. Such comparisons were critical in determining whether changes in genetic diversity occurred as a result of MAS and if the alterations in allelic frequency detected were the result of selection primarily for genomic segments of C. hytivus during MAS. The assessment of genetic (marker-based) and phenotypic (economically important traits) diversity conducted herein also provided for an estimation of the potential value of backcross progeny for plant improvement.

Genetic diversity

The genetic variation among the backcross populations examined was not significant (P < 0.05; data not shown). Nevertheless, backcross progeny derived from line WI 9-6A (P3) were more polymorphic than those derived from WI 7023A (P2). Likewise, although the genetic diversity as measured by the gene diversity coefficient (Hc) and Shannon’s information index (I) were relatively high in all backcross progeny derived from WI 9-6A, values for such estimates were comparatively low in progeny derived from WI 7023A (Table 1). In fact, the estimated percentage of polymorphism was relatively high in marker-selected backcross progeny, ranging from 70.3% in BC2P2 (Sel) to 86.5% in BC2P3S1 (Sel). Such polymorphism levels might be expected given the use of the C. hystrix-derived amphidiploid C. hytivus in the creation of donor parent WI 7012A {[C. hytivus × C. sativus) × C. sativus] S3}. Similar differences between distantly related C. sativus var. sativus and var. hardwickii (Kupper and Staub 1988) have been exploited to create high-yielding, multiple disease-resistant lines (Staub et al. 1992). The differences observed in the marker-selected backcross progeny described herein portend their potential value in broadening the narrow genetic base of cucumber (3–12% polymorphism; Dijkhuizen et al. 1996; Horejsi and Staub 1999).

The observed levels of heterozygosity between marker-selected and nonselected backcross progeny ranged from 0.25 in BC2P2 (NSel) to 0.31 in BC2P3 (Sel) (Table 1). In contrast, heterozygosity differences between marker-selected backcross genotypes ranged from 0.27 in BC2P2 (Sel) to 0.29 in BC1P2 (Sel). Moreover, the heterozygosity detected in BC1P2-derived progeny was increased when mated to WI 9-6A to create BC2P3 and BC2P3S1 progeny (heterozygosity ~0.31), suggesting that this parent (originating from LB sequential fruiting C. sativus line H-19) be used in future experiments directed towards increasing the genetic diversity of cucumber in combination with advanced lines originating from the C. hytivus-derived line WI 7012A. Emphasis should be placed on alignment of complementary quantitative trait loci (QTL) associated with fruit yield and quality through continued joint phenotypic and MAS recurrent selection (Robbins et al. 2008; Robbins and Staub 2009).

Gain from selection is to a great extent dependent upon the complementary characteristics of often genetically diverse parental lines. The mean genetic identity (I; synom. genetic similarity) among pairs of parents was lower than the I among any two pairs of backcross populations (Table 1). When backcross populations are considered as a group, I predictably increased dramatically during advanced backcrossing [i.e., among any backcross pair (BC2P2 versus BC3P2)] to as high as 0.91, suggesting that relatively few alleles contributed significantly to the among-population genetic variation; for example, the lowest I (0.320) was detected between backcross populations derived from WI 7012A (C. hytivus derived) and WI 7023A (C. sativus derived), where a gradual increase in genetic similarity was detected over successive backcrossing and selfing generations [e.g., 0.49 (BC1P2 NSel versus WI 7012A) to 0.59 (BC3P2 Sel versus WI 7012A)]. In contrast, the highest identity (0.91; most similar) was detected between populations BC2P2 (Sel) and BC3P2 (Sel) and populations BC3P3 (Sel) and BC2P3S1 (Sel). Shannon’s information index (Id) showed very close values (range 0.38–0.46) among all backcross populations. These data suggest close genetic affinities among all backcross populations examined, and confirm earlier reports of MAS that dramatic gains from selection for yield and quality components in cucumber often occur during early backcross [(C. sativus × C. sativus) × (C. sativus × C. sativus)] generations (Fazio et al. 2003b; Fan et al. 2006).

The largest genetic distance among the parents used was detected between the parents WI 7012A and WI 7023A (Nei’s GD = 0.68; Table 2). Predictably, all marker-selected backcross progeny derived from WI 7012A were genetically more diverse than those derived from WI 7023A. Curiously, WI 7023A-derived backcross progenies were more similar to WI 9-6A-derived progeny than to WI 7012A-derived progeny. Nevertheless, the GD among backcross progeny, regardless of parental donor, was similar. Such genetic affinities among advanced backcross progeny are likely attributable to the progressive level of inbreeding (backcrossing and selfing) that occurs during line development. Tarter et al. (2004), in fact, argued that, in maize (Zea mays L. ssp. mays), intensive inbreeding and phenotypically based backcrossing within elite lines derived from exotic sources (i.e., comparable to the C. sativus parents used herein) could result in loss of exotic alleles. Appropriately designed MAS in cucumber during backcrossing has increased selection efficiency even after intense initial recurrent phenotypic selection (Fan et al. 2006). Thus, MAS in cucumber could be employed as a tool to mitigate loss of exotic alleles during inbreeding after relatively wide crossing (e.g., C. hytivus × C. sativus). The relatively high estimated level of gene diversity, Shannon’s information index, and polymorphic loci defined herein lend support to this contention.

Table 2 Genetic distance (Nei 1973) among cucumber parental lines and their derived backcross populations as measured by 24 molecular markers (SSR, SCAR, and SNP)
Full size table

Phenotypic diversity

For exotically derived germplasm to be employed in agriculture it must possess acceptable commercial attributes such as high yield and quality. The inbred parents employed herein were chosen based on their unique and complementary architectural (e.g., determinate/indeterminate, sequential fruiting, and multiple lateral branching) and fruit quality (e.g., fruit dimensions and interior quality) characteristics (Table 3). The horticultural evaluation of the cross-progeny developed herein provides a preliminary assessment of their commercial potential. Parents and backcross progeny were assessed for their relative vegetative vigor (Fig. 1), LB, DF, fruits per plant, and fruit L:D and E:T. Nevertheless, although parental line differences were detected for these traits, several selected and nonselected cross-progeny populations were similar (P < 0.05). Such results indicate the need for the development of breeding strategies that consider trait correlations, epistatic interactions, and the strength of marker–trait associations (Fazio et al. 2003a; Robbins et al. 2008; Robbins and Staub 2009).

Table 3 Least-square means (ls means) and least significant differences (LSD) for several traits as evaluated in cucumber parental lines and their derived backcross populations in the open field at Hancock, WI, 2007
Full size table

The strength and direction of fruit yield and quality correlations utilized herein has been well documented in a wide range of genetic backgrounds (Kupper and Staub 1988; Serquen et al. 1997; Cramer and Wehner 1999; Cramer and Wehner 2000b; Fazio et al. 2003a). For instance, lateral branch number was positively correlated (r = 0.58 to 0.42) with the number of fruit per plant in one processing cucumber population derived from the multiple branching line H-19 (Fazio 2001). Likewise, significant, positive correlations between yield and lateral branch number have been detected in a diverse array of cucumber populations (Fredrick and Staub 1989; Cramer and Wehner 1999, 2000a). This correlation, in fact, often increases during continued selection (from r = 0.67 to 0.82), which led Cramer and Wehner (2000b) to suggest that efforts to improve yield in cucumber should focus on increasing lateral branch number. Selection could be effective for this trait, since it is controlled by relatively few additive genes (~4–5; Wehner 1989; Serquen et al. 1997; Fazio et al. 2003a) that possess varying degrees of heritability [narrow-sense heritability (h 2) from 0.00 to 0.61] depending on the population exploited (Robbins and Staub 2009, 2008).

To mitigate negative correlations among yield and quality components in backcross progeny, complementary parental lines were used. In some cases, introgression of complementary parental characteristics resulted in backcross populations with unique characteristics (i.e., C. sativus- and C. hytivus-derived parents shared inferior and superior traits) (Table 3). For instance, even though a C. sativus parent (e.g., WI 7023A) was inferior for number of lateral branches (~1.00) and fruit yield (~13), it was superior for days to flowering (~20) and fruit quality [e.g., L:D (~3.2) and E:T (0.64) ratios]. As a result, derived backcross progeny (either Sel or NSel) often possessed commercially acceptable characteristics (i.e., yield and fruit quality) (Table 3). These findings are further supported by Delannay and Staub (2010b) who also found that backcross progeny derived from the complementary parents (i.e., WI 7023A and WI 7012A) exhibited more commercially acceptable traits than either parent alone. Likewise, the C. sativus inbred line WI 9-6A (P3) used herein contributed superior yield and quality component traits [i.e., LB (~3), L:D (3.8), E:T (0.57), and cumulative FN (37)] but inferior DF (25) to its progeny. When used as a recurrent parent in advanced backcrossing (BC2; {[(P1 × P2) × P2] × P3}) to introduce complementary characteristics (BC3P3 and BC2P3S1), progeny generally possessed an array of commercially acceptable attributes. Positive outcomes of MAS in cucumber, however, are not always predictable (Robbins and Staub 2009), and, thus, judicious use of marker–trait associations that employ appropriate breeding schemes (e.g., a combination of phenotypic and genotypic selection) is key to obtaining positive gain from selection.

Gain from selection for yield and quality traits in cucumber can be realized by altering plant architecture during selection. For instance, gynoecious, determinate, multiple lateral branching phenotypes with sequential fruiting habit have been proposed for increasing cucumber yield in once-over mechanical harvest operations (Staub et al. 2008; Cramer and Wehner 2000b). The multiple branching and sequential fruit habit is present in C. sativus var. hardwickii, C. hystrix, C. hytivus, and the C. hytivus-derived line WI 7012A used herein (Staub and Kupper 1985; Chen et al. 1997). These characteristics and the vigorous vegetative growth associated with WI 7012A make it attractive to breeding programs whose focus is yield improvement. Even though WI 7012A possesses some poor fruit quality attributes (i.e., predominant warts, blossom-end taper), the fruit of the WI 7023A and WI 9-6A-derived cross-progeny examined herein was commercially acceptable (Table 3). Moreover, some of these progeny possessed the plant architecture (i.e., gynoecious, determinate, multiple branching phenotypes) and yield attributes (e.g., early concentrated fruit set) required for mechanical harvest operations (data not shown).

Potential commercial value of backcross progeny

The maturity of backcross progeny varied due to differences in earliness and plant architecture (Table 3). For instance, although DF varied from ~20 (BC3P2S) to ~25 (BC2P3S) among primarily indeterminate backcross populations (i.e., BC1 and BC2), determinate and indeterminate plants were observed in early generations (data not presented). Regardless of generation, however, determinate plant types possessing early flowering habit were the consequence of WI 7023A alleles [P2; determinate (de), gynoecious (F)]. Determinate gynoecious plant types typically flower earlier than indeterminate monoecious genotypes, and often provide a yield advantage over indeterminate genotypes in early (1–2) by not in late (4–6) harvests (Staub et al. 2008). Consequently, differences in cumulative six-harvest yield were detected among backcross populations, varying from ~18 fruits per plant in BC3P2 (Sel) to ~39 fruits per plant in BC2P3 (Sel). In fact, the fruit yields of progeny with WI 7012A (monoecious, vigorous indeterminate) in their pedigree were consistently moderate to high (Table 3). Plant vigor (vine length) of both parental line WI 7012A (P1; donor parent) (vigor score = 4; Fig. 2) and derived backcross progeny (Table 3; Fig. 3) were comparatively high (in relation to ‘Vlasset’), especially in some advanced generations [e.g., BC2P3 (Sel), BC3P3 (Sel)]. These observations recapitulate similar findings reported by Delannay (2009) during advanced backcrossing in a C. hytivus genetic background with different recurrent C. sativus parental market types (i.e., Beit Alpha and Long European greenhouse). It is likely that the vigorous vegetative habit consistent with C. hytivus-derived progeny provided a yield advantage during later harvest dates, where its vigor-associated alleles (i.e., multiple lateral branching, indeterminate habit) were complementary but different from those of C. sativus recurrent parents used (Robbins et al. 2008). This was dramatically evident when WI 9-6A was used as the recurrent parent, since WI 9-6A-derived backcross progeny [e.g., BC2P3 (Sel), BC3P3 (Sel), BC3P3 (Sel), and BC2P3S1 (Sel)] were more vigorous and at times higher yielding than BC2P2 progeny (Table 3; Fig. 3). These vigor and yield responses were likely potentiated by the use of WI 9-6A as an alternate parent in backcrossing [i.e., BCP2 × P3] given the increase in heterozygosity and polymorphism level detected (Table 2). Line WI 9-6A originated from C. sativus crosses where line H-19 contributed sequential fruiting in a multiple lateral branching background (Staub et al. 2008). The yield and quality marker–trait associations were employed during MAS herein (Fan et al. 2006; Robbins et al. 2008) and likely led to the strategic pyramiding of yield-related QTL alleles (Robbins et al. 2008) that contributed to the documented yield increases. Recent increased saturation of the historic RIL-based map (Gy-7 × H-19) with SSR markers defined relatively tight marker–trait linkages with de (1.4 and 4.2 cM) and vigor-associated traits [e.g., little leaf (ll) at 4.2 and 3.6 cM] (Weng et al. 2010). These linkages will assist in MAS where the goal is recovery of determinate germplasm with increased vigor (increased lateral branching associated with ll).

Fig. 3
figure 3

Fruit of Cucumis hytivus WI 7012A-derived inbred backcross lines and C. sativus L. line 9-6A. a Fruit irregular in shape and spiny, b nonuniform dark-green fruit at edible maturity with low E:T ratio (0.57), c uniform dark-green fruit harvested at edible maturity from a single plant, d nonuniform dark-green fruit harvested at edible maturity from a single plant

Full size image

The genotypic and phenotypic diversity of the backcross populations were significant for all the traits examined. Where direct comparisons could be made [i.e., BC2P2(NSel) versus BC2P2 (Sel), BC2P2S1 (NSel) versus BC2P2S1 (Sel)], marker-selected progenies were vigorous and early flowering (Table 3). Likewise, backcross progeny varied dramatically in morphology depending on the recurrent parent used (Fig. 2). Whereas multiple lateral branching (~3.5 laterals/plant) BC3P3 (Sel) progeny produced, as a cumulative average, 33 fruits/plant with mean L:D of 3.7, unilateral branching (~1.5 laterals/plant) BC3P2 (Sel) progeny produced comparatively fewer (~18 fruits/plant) and shorter (L:D of 3.2) fruit. Given the morphological diversity detected in these backcross populations and the potential benefit of using alternate recurrent parents (e.g., WI 9-6A) to increase diversity and improve population characteristics (e.g., plant vigor, LB, and fruit number and L:D), it is likely that the positive attributes of C. hytivus-derived WI 7012A can be capitalized upon during breeding. This assertion is supported by the fact that several backcross families [e.g., BC2P3 (Sel), BC3P3 (Sel), and BC2P3S1 (Sel)] lacked the negative attributes associated with WI 7012A (i.e., late flowering, and warty, oblong fruit). If progeny with positive attributes could be recovered in such advanced backcross families, then MAS might have utility for further line advancement (fruit yield and quality; Fazio et al. 2003b; Fan et al. 2006) and subsequent assembly of diverse, heterozygous, commercially acceptable hybrids.