Materials.

We generated 13 families using US 1896 (= V. padifolium US 908 × V. corymbosum US 1825) as either a female or a male. The details of the origin of US 1896 have been described previously (Ehlenfeldt and Polashock, 2014). Of the 13 families, two used US 1896 as a female and 4x V. corymbosum (‘Bluecrop’ and ARS 98-15, respectively) as males. Eleven families were generated using US 1896 as a male in crosses using commercial cultivars and selections as females. These cultivars were Biloxi, Bluecrop, Brunswick (lowbush), Cara’s Choice, Duke, Hannah’s Choice, Legacy, Northsky (half-high), Sharpblue, Sweetheart, and TH 622. The details of these crosses were described previously (Ehlenfeldt and Polashock, 2014). All of the parents listed are named cultivars except for ARS 98-15 and TH 622. ARS 98-15 is a breeding selection of the pedigree ‘Duke’ × ‘Nelson’. The selection TH 622 is a sibling of the southern highbush cultivars Camellia (NeSmith and Draper, 2007) and Gupton (Stringer et al., 2012) and comes from the cross of MS 122 × MS 6. The cross MS 122 × MS 6 was notable for productive, vigorous plants with large, pale blue fruit.

To confirm that the plants evaluated were true hybrids, we evaluated a subsample of 34 genotypes using a genotyping by sequencing (GBS) approach (Elshire et al., 2011). DNA extraction and GBS library construction were as detailed in Daverdin et al. (2017). The libraries were sent to Novogene Corporation Inc. (Sacramento, CA) and sequenced on the Illumina (San Diego, CA) HiSeq platform (PE150). GBS sequencing data were demultiplexed using STACKS process_radtags, and contigs were generated using the default STACKS pipeline denovo_map.pl (Catchen et al., 2011). BWA-mem and samtools were used to perform read alignment and call single nucleotide polymorphisms (SNPs) against contigs generated by the STACKS de novo pipeline (Li, 2011; Li and Durbin, 2009). Homogeneous SNPs were filtered out, SNPs unique to each parent were cataloged, and SNPs from offspring matching parent-type SNPs were subsequently binned and counted. In all cases, the maternal parent is known, so the focus of the data analyses was to confirm that all progeny inherited at least some SNPs from the putative pollen parent.

Progeny from all crosses were grown for 9 months in the greenhouse and were planted to the field in June 2015 (Table 2). Families were planted in three adjacent rows spaced at 2 feet within rows and 8 feet between rows, with rows averaging 650 feet long. For this evaluation, families were planted in consecutive order with no replication. Chatsworth, NJ, is located at 39.8° N latitude and 74.5° W longitude. New Jersey has a moderate climate, with cold winters and warm, humid summers. Chatsworth’s average midsummer high (July) is 30.5 °C; the average wintertime low (January) is –5.0 °C. The planting was made on Berryland soil and maintained using standard growing practices. Evaluations were made in 2018.

To assess general field performance, plants were evaluated for survival, vigor, height, bloom density, percent bloom, 50% ripe date, fruit size, and fruit color. Limited evaluations were made for fruit firmness and flavor/volatiles.

• • Survival was assayed at the first evaluation date, 18 Apr. 2018. Any plant that had resumed growth no matter how poor was considered a survivor.

• • Vigor was evaluated in mid-June. Rating was done on a scale of 0 to 3, where 0 = struggling, 1 = fair vigor, 2 = average vigor, and 3 = above average vigor.

• • Height was determined at the end of the summer season (22 Aug. 2018) by measuring each plant to the nearest 5 cm.

• • At the time of first inspection, bloom density data were also taken. Flowering intensity was rated with subjective notations of no flowers, almost no flowers, light flowering, average flowering, and vigorous flowering, but for the purposes of this report, flowering intensity was reduced to a binary (0–1) scale with 0 = no flowers or almost no flowers, and 1 = categories from light flowering to vigorous flowering.

• • Bloom timing was compared by measuring percent bloom at a single date, 17 May 2018.

• • A 50% ripe date was determined by evaluating/rating the percent of ripe fruit weekly from 20 June to 26 July. To calculate a 50% ripeness date, the dates bracketing a potential 50% value were converted to ordinal dates and linearly interpolated to determine a 50% ripeness date. Values were rounded to the nearest whole day. After averaging across clones to derive a family mean, ordinal dates were back-converted to calendar dates.

• • Fruit color was rated on a 1 to 5 scale, where 1 = light blue, 2 = light-medium blue, 3 = medium blue, 4 = dark blue, and 5 = any one of several shades including very dark blue-black, dusky black, or black. Commercial cultivars would be expected to rate as 2 or lower. Values were averaged across families, and for reporting purposes, average color values were additionally back-converted to subjective color descriptions.

• • Fruit size was rated on a 1 to 4 scale, where 1 = very small, 2 = small, 3 = medium, and 4 = large. For fruit size, 0.5 increments were used as needed. Commercial cultivars would be expected to rate as 3 or greater. As was done for fruit color values, fruit size values were averaged across families, and for reporting purposes, average fruit size values were also back-converted to subjective descriptions.

To better understand the families we had generated, the evaluations rated the plants, calculated means, and determined standard deviations where meaningful. Additionally, not every evaluation could be made on every clone; for example, no 50% ripeness date could be generated for a plant that failed to flower, so observation numbers are listed parenthetically in both tables and text.

Genotyping by sequencing.

Of 952,013,696 total sequencing reads, 755,924,283 were available after de-multiplexing and filtering for matching barcodes. The STACKS de novo pipeline generated 737,894 contigs. After read alignment and SNP calling of reads back to the de novo generated contigs, the sum total of unique SNPs from all female parents was 16,990, whereas the male parent, US 1896, exhibited 1,583 unique SNPs (Table 1). On average, individual offspring matched ≈652 total unique SNPs across all varieties used as females in this study, whereas the bulk of the matching SNPs were identified from at least one parent in the pedigree (Table 1). Most of the unique SNPs could be attributed to the pollen parent, US 1896. This suggests that for all of the clones evaluated, the progeny are not selfs and are indeed true hybrids.

Table 1.

Unique parental single nucleotide polymorphisms (SNP) variants exhibited in a subset of parents and progeny selections (parents of a given selection are identified by gray shading).

View Table

Morphology.

In the BC₁ cross generation, ancestry was 25% V. padifolium and 75% cultivated germplasm (i.e., V. corymbosum, V. angustifolium, and lesser amounts of V. darrowii and V. virgatum) thus much of the morphology tended toward cultivated phenotypes (V. corymbosum/V. angustifolium). A casual observer of these families in the field would not recognize any significant differences from highbush. The offspring with ‘Northsky’ as a parent tended to have shorter, bushier, and more branching plant structure; offspring with ‘Cara’s Choice’ were intermediate in stature; offspring with TH 622 were more upright, robust plants. Bush structure appeared little different from comparable highbush type crosses; however, on larger plants, canes were more robust, but possibly also more brittle. Some, but not excessive, breakage was observed.

The two main features that generally distinguished these hybrids, were leaf morphology and variant shoot development. Regarding leaf morphology, although similar in size to cultivated germplasm, leaves were generally flatter and less textured than those of highbush, but venation appeared slightly more prominent against this flatter texture (Fig. 1). The parent US 1896 possesses a noticeably patterned leaf surface, and its leaves are relatively coarse; thus some of this morphology appears to have been inherited.

View Full Size

Leaf morphology of V. corymbosum ‘Bluecrop’ (left) and a backcross (BC₁) clone of V. padifolium, US 2445 (= TH 622 × US 1896) (right) at Chatsworth, NJ, 2018.

Citation: HortScience horts 55, 11; 10.21273/HORTSCI15279-20

• Download Figure
• Download figure as PowerPoint slide

Also unusual was the development of variegated, and sometimes totally chlorophyll deficient shoot development (Fig. 2A and B). This is believed to be the result of nuclear-cytoplasmic interactions of this exotic material with cultivated germplasm (Correns, 1909). In conventional crosses, variegation, although not frequent, is not unusual in combinations that mix early-maturing germplasm (generally containing greater lowbush ancestry) with midseason or late-season maturing germplasm. What was somewhat unusual in these offspring was the vigor of the totally chlorophyll deficient shoots derived from some crosses (Fig. 2B). Flowers on these hybrids did not appear notably different from V. corymbosum material.

View Full Size

Variegation and chlorophyll deficient shoots in backcross (BC₁) families of V. padifolium to 4x V. corymbosum/V. angustifolium material at Chatsworth, NJ, 2018: (A) US 2425 (= ‘Northsky’ × US 1896) and (B) US 2428 (= ‘Northsky’ × US 1896).

Citation: HortScience horts 55, 11; 10.21273/HORTSCI15279-20

• Download Figure
• Download figure as PowerPoint slide

Survival.

The field plot was not uniform in its drainage or soil conditions, and these plants were not planted in replicates; thus, a uniform estimate of survival was not feasible. Nonetheless, the survival rankings across families appeared generally to represent the potential relative responses typical for the families (Table 2). Across all families (a total of 996 plants), survival averaged 28.5% (284 plants). Notably, for the two crosses with US 1896 as a female parent, no offspring survived (both were relatively small families). The reasons for this are unclear, but it is a distinct possibility that these were selfs of US 1896 (these were not genotyped). As noted, it was recognized that US 1896 had poor survival under New Jersey field conditions. Among families using US 1896 as a male, survival across families ranged from 0.0% (‘Brunswick’ × US 1896) to 62.0% (‘Northsky’ × US 1896). Of potential interest in these two contrasting families is that lowbush introgression via ‘Brunswick’ produced a family apparently totally unadapted to our test conditions (although sample size was small, 7 individuals). This cross had also been less successful in terms of fruit set and total seed generated initially (Ehlenfeldt and Polashock, 2014). In contrast, the family with lowbush-highbush introgression via ‘Northsky’ exhibited superior survival capability (62% survival, 62 individuals).

Table 2.

Survival, vigor ratings, and height measurement averages for V. padifolium backcross (BC₁) families to 4x V. corymbosum/V. angustifolium material at Chatsworth, NJ, 2018 (sample sizes denoted in parentheses).

View Table

Vigor.

A number of plants had cane dieback (presumably from cold damage), but often these plants were growing back from ground level or from a branch near ground level. On the rating scale of 0 to 3, vigor ratings ranged from a high of 2.4 ± 1.0 (39 individuals) in a family of ‘Duke’ × US 1896 to lows of 1.3 ± 1.5 (three individuals) for a small family of ‘Hannah’s Choice’ × US 1896, and similarly, 1.4 ± 1.3 (36 individuals) for the family of ‘Sharpblue’ × US 1896 (Table 2). The 2.4 value represents a family of average to above average vigor. The 1.3/1.4 values represent fair to average vigor.

Plant height.

Height was measured to the nearest 5 cm at the end of the summer season (22 Aug 2018) (Table 2). Average height ratings ranged from 40 ± 13 cm (62 individuals) for ‘Northsky’ × US 1896 to 70 ± 21 cm (31 individuals) for TH 622 × US 1896. The best plant in the TH 622 × US 1896 family had a height of 120 cm. The best plant in ‘Northsky’ × US 1896 had a height of 75 cm. A height of 140 to 170 cm might be considered ideal (approximately chest high to head high); a height of 100 cm would also be considered acceptable (approximately waist high).

Bloom intensity.

Bloom intensity was summarized on a 0 to 1 scale with 0 = plants with no flowers or almost no flowers and 1 = light flowering to vigorous flowering. This value essentially described how many plants within the family bloomed. Bloom ratings ranged from 0.2 (17 individuals) for the family of ‘Sweetheart’ × US 1896 (i.e., ≈20% total flowering) to 0.8 (33 individuals) for ‘Cara’s Choice’ × US 1896 (≈80% total flowering) (Table 3). Although many factors can influence flowering, for our purposes flowering, is considered at least a partial proxy for winter floral-bud hardiness, and plants with weak flowering were ultimately discarded.

Table 3.

Averaged bloom ratings, 50% ripeness date, fruit color, and fruit size ratings for V. padifolium backcross (BC₁) families to 4x V. corymbosum/V. angustifolium material at Chatsworth, NJ, 2018 (sample sizes denoted in parentheses).

View Table

Bloom timing.

Bloom timing should be examined with the proviso that for plants with light flowering, it was often hard to judge flowering percentages critically. As a result, the listed data reflect only clones for which flowering percentage could be reasonably estimated (Table 3); in a number of families, this resulted in observation numbers lower than those subsequently used for fruit color ratings and other factors. Bloom timing was examined on 17 May 2018, a date that was considered to be mid-late bloom, allowing relative ranking of families. Upon observation, it was apparent that a number of individual plants in several families had already reached 100% bloom before this date. These are noted separately from the plants that were not yet at 100% bloom and are listed as 100+. Bloom on 17 June 2018 ranged from 61% [61% (five individuals); 100+ (one individual)] for the family TH 622 × US 1896% to 90% [90% (two individuals); 100+ (three individuals)] for the family ‘Sharpblue’ × US 1896. As a check on these estimates, bloom percentage was also calculated including the 100+ values as exactly 100%. Although it shifted values slightly, rankings remained essentially unchanged (data not shown). On the basis of the observed bloom, it can be concluded that these materials flowered relatively early, and their bloom timing may be similar to that of southern highbush; thus, climate adaptation may be a concern in selecting for their further utilization.

Ballington (2001) reported several intersectional crosses of V. cylindraceum with both 2x and 4x germplasm of section Cyanococcus and other sections. Ballington generated BC₁ hybrids with V. corymbosum, several of which were grown under field conditions in North Carolina to observe their performance. These BC₁ hybrids used only V. cylindraceum germplasm from section Hemimyrtillus. The observed hybrids exhibited good fertility but were early flowering and often sustained frost damage.

Fifty percent ripeness dates.

Ripeness was rated weekly from 20 June to 26 July 2018. By the last evaluation date, only five plants had not completed ripening, and all five of these were at 90% ripeness or greater. These late ripening selections were distributed across four families. Fifty percent (50%) ripeness dates ranged from 26 June ± 5 (12 individuals) for ‘Duke’ × US 1896 to 5 July ± 4 (seven individuals) for TH 622 × US 1896. The cultivar Duke was rated at 60% ripeness on 20 June 2018. Thus, the family with ‘Duke’ might be considered relatively early ripening; however, most of these families would be considered, on average, to be midseason, and some clones were considerably later to ripen. For families of sufficient size, most standard deviation values were relatively equivalent.

Time and practical constraints did not allow specific assays for fertility, but ripeness estimates may be considered a partial/limited proxy for fertility because to estimate ripeness, it was necessary not only to have a parent with relatively good flowering but also to have a plant that subsequently set adequate fruit to allow a ripeness estimation. Pollination was not expected to be an issue because many seedling families of typical highbush were grown in adjacent evaluation rows.

Fruit color.

Fruit color was rated on a 1 to 5 scale with 1 being lightest and 5 darkest. Across families, fruit color tended toward darker than commercially standard values (2 = commercially acceptable). Our F₁ parent, US 1896, had relatively dark fruit and a mottled wax coating, and on the current scale, it would be considered 3 to 3.5. For offspring families, fruit color means ranged from 2.5 ± 0.8 (eight individuals; light-medium blue/medium blue) for the TH 622 × US 1896 family to 3.5 ± 0.7 (11 individuals; medium/dark blue) for the ‘Duke’ × US 1896 family. Although the family means were all greater than 2, five of 13 families had individual clones with acceptable commercial ratings of 2 (Table 3; Fig. 3A and B). It is worth noting that only the TH 622 × US 1896 family had any clones receiving a rating of 1 (light blue). As a southern highbush selection, the TH 622 parent had particularly pale blue fruit. In contrast, only the ‘Northsky’ × US 1896 family had any individuals receiving a rating of 5 (blue-black to black).

View Full Size

Examples of retained clones of backcross (BC₁) families of V. padifolium to 4x V. corymbosum material at Chatsworth, NJ, 2018: (A) US 2408 (= ‘Cara’s Choice’ × US 1896) and (B) US 2411 (= ‘Legacy’ × US 1896).

Citation: HortScience horts 55, 11; 10.21273/HORTSCI15279-20

• Download Figure
• Download figure as PowerPoint slide

Fruit size.

Fruit size was rated on a 1 to 4 scale with 1 smallest and 4 largest. Family means for fruit size ranged from 2.5 ± 0.5 (35 individuals; small) for ‘Northsky’ × US 1896 to 3.0 ± 0.3 (eight individuals; medium) for TH 622 × US 1896. Five families had individual clones that rated as high as 3.5 (medium-large). The parents of these families were ‘Duke’, ‘Legacy’, ‘Sharpblue’, TH 622, and, rather notably, ‘Northsky’.

Fruit quality.

Fruit quality was given only limited, subjective evaluation, and no clones were noted that had exceptional fruit firmness. Most clones appeared to have firmness comparable to traditional northern highbush. Similarly, fruit flavor/volatiles exhibited no perceptible unusual qualities; in fact, in early ripening stages, it was a concern that the fruit was undistinguished with regard to volatiles; however, when sufficiently ripened, fruit quality was also within normal ranges for northern highbush flavor.

Perhaps the most interesting question to pose is whether any clones exhibited unusual flowering patterns similar to V. padifolium. Vander Kloet and Dickinson (2005) labeled the flowering mode of V. padifolium as indeterminate. Manifestations of indeterminate flowering by their delineation include inflorescences developing without perenniation, ongoing flower differentiation within or along a floral axis even after initial bloom, and, as an outcome of this, high flower numbers. Our concern in this respect was monitoring mainly for nonspecific, continuous, or possibly a second flowering/cropping cycle. Thus far, under our growth conditions and cultural practices, no return bloom or similar manifestations have been observed in any of these clones, but such behavior may depend on maturity and specific seasonal conditions.