Materials and Methods

Open-pollinated seeds were collected from four V. stamineum plants growing on deep sandy soil in an open forest in southeastern Alachua County, Florida, in Aug. 2010. The plants were 2–4 m tall with 3–8 major canes and were estimated to be 10 to 40 years old. The plants selected as seed sources were vigorous and upright, produced large, juicy berries, had high yields, and retained the ripe berries well on the plant. The skins of the ripe berries were black to purple and the pulp ranged in color from white/green to dark purple depending on the plant from which the berries were taken. Seeds were extracted and dried. In Nov. 2010, the dry seeds were submerged in 0.2% aqueous colchicine for 7 d and then planted without rinsing under intermittent mist on the surface of Canadian peat in a greenhouse at the University of Florida in Gainesville. Several thousand seedlings resulted, and in Jan. and Feb. 2011, ≈400 were selected because they had thick hypocotyls and were transplanted to trays of peat. Three months later, 50 plants that had vegetative traits suggesting polyploidy were transplanted to pots of peat and maintained until they flowered. Pollen size was examined for each plant in early Spring 2012. Pollen was mounted in 45% acetic acid/55% water and examined at ×250 with a light microscope. Four plants had pollen that was conspicuously larger than the others. Additional seeds and seedlings from the same wild population were treated with colchicine in 2011 and 2012, and nine additional V. stamineum seedlings were found that produced categorically larger pollen. These plants were used in crosses with highbush cultivars during the three seasons of 2013 through 2015 (Table 1).

Table 1.Number of seeds or seedlings obtained during 3 years by pollinating tetraploid highbush cultivars with pollen from 13 tetraploid Vaccinium stamineum plants.

View Fullscreen

Crosses were made in a bee-proof greenhouse in Gainesville using plants growing in 10 L pots. The Cyanococcus plants, which were used as females, were emasculated before anthesis, and the stigmas were immediately covered with pollen brought by thumbnail from V. stamineum plants growing in the same greenhouse.

In 2012, two clones of V. darrowii (diploid Cyanococcus) were pollinated with pollen from diploid V. stamineum. From 649 pollinated flowers, numerous seedlings were produced and 200 were transplanted to flats of peat and were grown in a greenhouse for 4 months, after which they were transplanted to a field nursery at the University of Florida Plant Science Unit in Citra, FL. After 9 months in the field (May through January), 165 plants were identified as hybrids based on vegetative morphology. Compared with highbush seedlings, the hybrids were very weak, had yellow/bronze leaf color instead of dark green, had heavy pubescence on the petioles and stems of new shoots, and had areolate rather than reticulate leaf venation. By the end of the first growing season in the field, highbush seedlings had produced flower buds but the hybrids had not.

In 2013, intersectional (Cyanococcus × Polycodium) crosses were made at both the diploid and tetraploid level. The diploid Cyanococcus female parents were two V. fuscatum plants propagated from the central Florida peninsula near Sebring, and the V. stamineum plants that supplied the pollen were diploids selected from the forest in southeastern Alachua County. Female parents for the tetraploid crosses were 13 highbush blueberry cultivars, and the pollen came from four V. stamineum plants identified as tetraploids in 2012. The number of flowers pollinated for each cross, the number of berries that matured, and the number of fully developed seeds per berry were recorded. Seeds were extracted and dried. In November of the year in which the berries were harvested, the seeds were sprinkled onto the surface of Canadian peat in an unheated greenhouse under intermittent mist. The following January and February, seedlings were transplanted to trays of Canadian peat and grown until early May in a greenhouse, when they were transplanted to a field nursery at 45 cm × 15 cm spacing. About 10,000 seedlings from highbush × highbush crosses and 1000 V. stamineum seedlings grown from seed gathered from the forest were planted in the same nursery at the same time and were used for comparisons.

Several hundred seedlings from the tetraploid intersectional crosses, which had been transplanted to the field nursery in early May, were examined in mid-Nov. 2014 to eliminate plants that did not appear to be highbush × V. stamineum hybrids. About 95% of the plants appeared to be the intended hybrids. The hybrids were much less vigorous than either parent. The leaves of most hybrid plants were yellow-brown in color, in contrast to the dark green leaves of the highbush and the grey-green leaves of V. stamineum. The petioles, leaves and stems of the hybrids were much more pubescent than those of highbush seedlings. None of the presumed hybrids had produced visible flower buds by mid-November, whereas most highbush seedlings in the same nursery had conspicuous flower buds.

On 20 Nov. 2014, 50 of the most vigorous highbush cultivar × tetraploid V. stamineum seedlings were dug, potted into 10 L pots, and placed in a walk-in cooler without light, where they were kept for 40 d. They were then placed in a sunny outside location for 10 d, after which they were moved to a bee-proof greenhouse with ambient light and temperature maintained between 2 °C and 27 °C. During the next 40 d, 23 more plants from the tetraploid crosses that looked like they might flower were dug from the field nursery and moved directly into the greenhouse. Between 27 Jan. and 20 Mar., 46 of the 73 putative hybrid plants in the greenhouse flowered. One of the 73 plants died, 3 were judged not to be intersectional hybrids based on flower morphology (Fig. 1) and absence of anther awns and 23 did not flower. The number of flowers produced per flowering plant ranged from 3 to 500. Flower and inflorescence morphology were examined for each plant. The relative length of the calyx, stamens, and style were measured, and the flowers were examined at ×30 for the presence of anther awns and calyx pubescence.

Potential male fertility was assessed for each plant that flowered by noting the amount of pollen that could be obtained by twirling the flower, corolla aperture down over a dark surface. The volume of pollen shed by highly fertile highbush cultivars such as Emerald and Farthing was used as a standard for “high” shed. Pollen was mounted in 45% acetic acid and examined at ×250 under a light microscope. Blueberry pollen is shed in tetrads of microspores. Several hundred pollen tetrads were examined for each plant to estimate microspore fertility. The percentage of individual microspores that appeared to be well developed and potentially viable was recorded. The diameter of complete tetrads was measured for 40 F₁ hybrids, for 4 highbush cultivars, for 5 tetraploid V. stamineum clones, and for 25 diploid V. stamineum clones. Ten well-developed tetrads were measured for each plant. Some F₁ hybrids did not have enough tetrads with 4 well-developed spores to permit accurate assessment of tetrad size.

Four types of crosses (Table 2) were made in the greenhouse using F₁ plants from the highbush × tetraploid V. stamineum crosses. Highbush flowers were pollinated with pollen from F₁ hybrids; flowers of F₁ hybrids were pollinated with pollen from one of three sources: a different F₁ hybrid, a highbush cultivar, or a tetraploid V. stamineum plant. The parents were selected to avoid inbreeding; every seedling had four distinct clones as grandparents. Percent fruit set, number of ripe berries harvested, and number of well-developed seeds produced by each cross were recorded.

Table 2.Results of various interspecific crosses in Vaccinium.

View Fullscreen

The following berry features of the F₁ hybrids were assessed: berry weight, internal flesh color, berry detachment (easy shattering or good retention of ripe berries on the bush), whether the pedicel remained with the berry or stayed on the bush during harvest, and skin bitterness. Berry firmness was estimated subjectively by repeatedly squeezing berries between thumb and forefinger and noting the rate at which firmness declined. Berry acidity, sweetness, and aroma were judged by eating one berry from each hybrid that produced fruit. Only one berry was sampled from each plant because saving the hybrid seed was given higher priority than evaluating flavor. Degrees Brix, an estimate of percent soluble solids in the juice, was measured using a hand refractometer using fresh-squeezed juice from ripe berries for several V. stamineum plants growing in the forest and for several tetraploid V. stamineum plants and highbush blueberry cultivars growing in the greenhouse.

On 31 Aug. 2015, 20–40 leafy softwood cuttings from each of 14 of the most vigorous F₁ hybrids were stuck in a peat-perlite medium under mist to determine their rooting ability. After 50 d (19 Oct.), the cuttings were removed from mist and placed in a sunny greenhouse. The rooted cuttings for each clone were counted 22 d later (10 Nov.).

Results

Intersectional crosses using pollen from diploid V. stamineum.

Pollinating flowers of the diploid section Cyanococcus species, V. darrowii and V. fuscatum with pollen from diploid V. stamineum produced large numbers of seedlings. Seedling numbers given for these crosses in Table 2 are minimum estimates, because only a fraction of the seedlings that germinated were retained and counted. Nearly all of the seedlings that survived to be transplanted to the field nursery were very weak in the nursery compared with highbush seedlings growing nearby. One exceptional V. darrowii × V. stamineum seedling was vigorous, and when backcrossed to a highbush cultivar, produced 100 seedlings, many of which appeared to have some V. stamineum features.

The two crosses of tetraploid highbush × diploid V. stamineum gave only 0.02 hybrid seedlings per pollinated flower along with a few dozen seedlings that resembled highbush blueberry and did not appear to be hybrids with V. stamineum. Several of the hybrid seedlings were quite vigorous when grown in pots of peat. They were highly variable in phenotype, some showing unusual leaf and flower forms, but they were not studied further.

Intersectional crosses using pollen from putative tetraploid V. stamineum.

The crosses were made in 3 years, using 13 different V. stamineum plants, which were believed to be tetraploid, as pollen sources (Table 1) and many different highbush cultivars as seed parents. Large numbers of seedlings were obtained from most of the crosses. Seedlings from the 2013 crosses were in the field nursery long enough to allow separation, based on morphology, of the likely intersectional hybrids (more than 95%) from seedlings from self-pollination of the highbush parent or pollination from other Cyanococcus sources. After 6 months of growth in the field, seedlings from the 2014 crosses appear to have about the same percentage of true hybrids.

Phenotype of the tetraploid hybrid seedlings in the field nursery.

At the end of the first growing season in the field nursery, the 840 hybrid seedlings from the 2013 crosses were far less vigorous than seedlings of the parent species, which were planted in the same nursery at the same time. Leaves of nearly all hybrids had a bronze-yellow-green color compared with dark-green leaves on the highbush and grey-green leaves on seedlings of V. stamineum. Leaf petioles, leaf veins, and the stems of new vegetative growth were much more pubescent on the hybrids than on highbush seedlings. Leaf venation in the hybrids tended to be areolate, as in V. stamineum, rather than reticulate as in highbush blueberry. On 20 Nov., near the end of the first growing season in the field, none of the hybrids had obvious flower buds, whereas nearly all of the highbush seedlings growing nearby had conspicuous flower buds. The 1800 hybrid seedlings from the tetraploid intersectional crosses made in 2014, which were planted in a separate but nearby nursery on the same research farm, averaged much higher in vigor than the hybrids from the year before, but were still less vigorous, on average, than seedlings of either parent. Vigor was highly variable among the intersectional hybrids, with much variation both among and within F₁ families.

Phenotype of 69 selected hybrid seedlings (highbush cultivar × tetraploid V. stamineum) in the greenhouse in 2015.

Seventy-three hybrid seedlings from various highbush × tetraploid V. stamineum crosses were selected for further study and for use as parents in backcrosses and intercrosses. These plants were selected because they were the most vigorous of the hybrid seedlings and appeared to be the most likely to flower. Of the 73 selected plants, one died in the greenhouse and three were rejected because flower and inflorescence morphology when they flowered indicated that they might not be the intended hybrids. The number of plants used to assess each phenotypic character varied. Some plants did not flower and could not be rated for flower and berry characteristics. Some plants produced too few berries to permit assessment of berry qualities, either because they produced few flowers or because the flowers had been sacrificed for pollen to make crosses.

Vegetative vigor of the potted F₁ hybrids was low as they grew, flowered, and developed fruit in the greenhouse, partly due to an infestation of azalea white mite (Eotetranychus clitus). The hybrids seemed much more susceptible to these mites than either parent species. Discoloration of the leaves and leaf drop due to mites and possibly other causes continued during most of the time between flowering and berry ripening, and probably reduced berry size and berry quality on the hybrid plants. After the crop was harvested and the mites were controlled, some of the potted hybrid plants became quite vigorous, and leaf color became healthy green on some.

Leaves on the hybrids tended to be lighter green in color than leaves of highbush. Hybrid leaf venation tended to be areolate rather than reticulate, with veins defining small raised islands on the upper leaf surface. Stems on new shoots were highly pubescent on most hybrid seedlings, and the leaf petioles were much more densely pubescent than on highbush. The bark pattern on 1-year-old wood was intermediate between highbush and V. stamineum.

The F₁ hybrids did not have visible flower buds at the time they were placed in the greenhouse. However, on most of the plants, new growth flushes soon emerged that had leaves and flowers on the same raceme. In these traits the hybrids resembled V. stamineum. It should be noted, however, that the plants were removed from the cooler in midwinter, when short days and cool nights (as low as 5 °C in the greenhouse on some nights) might have induced flowering even on new highbush growth flushes. Furthermore, a few of the F₁ hybrid plants that had been left in the field began to form highbush-type flower buds in mid-February on wood of the previous growing season. The seedlings dug in November were only 1-year old from seed. They had grown with marked lack of vigor and may have been too juvenile to initiate flower buds in the fall. Still, the long peduncles and pedicels of the hybrids when they flowered and the presence of leaves on the peduncles was suggestive of the V. stamineum flowering habit (Fig. 2).

View Figure

• Download as Powerpoint
• Download Figure

The mean diameter of well-developed pollen tetrads in diploid V. stamineum, estimated from 20 flowering plants, was 49 μm. Taking this value as 100% for comparisons, mean tetrad diameter for tetraploid V. stamineum was 125%, the same as the mean for five tetraploid highbush cultivars. Mean tetrad diameter was estimated for each of 40 F₁ hybrids based on 10 tetrads per plant. The median value for this population of means was 117%, with values for individual plants ranging from 92% to 150%. Reliability of the estimates for the F₁ plants was reduced by the fact that these plants produced very few tetrads in which all four microspores were fully developed. These tetrad sizes suggest that most, if not all of the F₁ hybrids were tetraploid or near tetraploid, as would be expected from crosses between tetraploid parents.

Fertility of selected F₁ hybrid seedlings (highbush cultivar × tetraploid V. stamineum).

Nearly all of the hybrids that flowered shed copious amounts of pollen when the flowers were twirled between thumb and index finger, corolla aperture downward. Percent viable pollen was estimated visually under a ×250 microscope for 44 of the hybrid plants that flowered in the greenhouse (Table 3). Several hundred pollen tetrads were examined for each plant. One plant was estimated to have 95% good microspores, but the median plant had only 30% good microspores, and a few plants were nearly male sterile. When pollen from F₁ plants was used to pollinate flowers of highbush cultivars, numerous hybrid seedlings were obtained (Table 2), but the crosses gave far fewer seeds than would be expected from highbush × highbush crosses.

Table 3.Number of plants of various types having pollen in each of 11 pollen viability classes.

View Fullscreen

Berry set and seed production were surprisingly high when the F₁ intersectional hybrids were used as female parents in backcrosses either to highbush cultivars (7.00 good seeds per pollinated flower) or to V. stamineum tetraploids (average 5.01) (Table 2; Fig. 3). When F₁ hybrids were used as pollen parents to pollinate highbush cultivars or other F₁ hybrids, seed set averaged only 1.5 seeds per pollinated flower (Table 2). Darwin (1859) and subsequent workers (Stevenson, 1965) noted that female fertility is normally higher than male fertility when wide hybrids are used in backcrosses.

View Figure

• Download as Powerpoint
• Download Figure

The tetraploid F₁ hybrids, when backcrossed to either parent species, produced far more seed than the original highbush × tetraploid V. stamineum crosses (Table 2). The same was true with highbush × tetraploid V. arboreum (section Batodendron) hybrids (Lyrene, 2013). This is somewhat surprising, since both parents of the initial crosses were fully fertile members of successful species, whereas one parent in the backcrosses was an intersectional hybrid, in which reduced fertility might be expected.

Other phenotypic characteristics of highbush × tetraploid V. stamineum F₁ hybrids brought to the greenhouse for use in crosses.

The morphology of the flowers on the 46 hybrid plants that flowered in the greenhouse was generally intermediate between those of the parent taxa (Figs. 1 and 2). In all of the plants judged to be F₁ hybrids, the flowers were open in the bud as in V. stamineum, and the styles extended outside the corolla tube well before anthesis. At anthesis, the tips of the anthers were located at or near the end of the corolla tube and the stigmas extended 2 to 4 mm beyond the end of the corolla (Fig. 2). In highbush flowers, the distal ends of the anthers are well inside the corolla tubes and the stigmas are located at or near the proximal end of the corolla tube (Fig. 1).

Anther awns are a key character separating section Cyanococcus species (which have no or very poorly developed awns) from section Polycodium, which has large and conspicuous awns (Camp, 1945). Three of the plants that had been dug as F₁ hybrids had no awns when they flowered. Two of these lacked other clear-cut deerberry features and were probably not from the intended cross. The third plant had some features suggestive of V. stamineum, but was not a typical F₁ hybrid and was not used in crosses. Anther awns were conspicuously present in all of the other 41 F₁ plants examined for this trait. The length and thickness of the awns varied considerably from plant to plant, but the awns of the hybrids were smaller and thinner than those of V. stamineum. In flowers of the median plant, the awns were about half as long as the anther sac.

Berry phenotype was assessed for all F₁ plants that produced five or more berries. Berries of most F₁ plants were smaller than those of either parent. Many of the V. stamineum plants in the forest had large berries. Thirty of the largest berries were harvested in the forest from each of two high-yielding plants on 14 July 2015. The average weight of these selected berries was 2.01 g per berry (sd 0.19) for one plant and 1.75 g (sd 0.14) for the other. Two of the tetraploid V. stamineum parent clones in the greenhouse had berries that averaged almost 2 g after hand crosses between the two clones. Among the F₁ plants in the greenhouse, the one with the largest berries averaged 1.59 g per berry for the first 20 berries harvested. Only about half of the F₁ plants had average berry weight over 1.1 g for the first 20 berries harvested. The smaller size of the berries on the F₁ hybrids compared with their parents probably resulted from their lower vigor and their problems with leaf mites and diseases.

Among the F₁ plants evaluated for pedicel attachment, 10 had pedicels that remained on the bush when the berry was harvested, which is characteristic of most highbush individuals, while 9 had pedicels that detached with the berry, which is characteristic of V. stamineum, and 10 had some berries detaching in each way.

Premature berry detachment, or shattering, is an important trait as berries must be retained on the plant long enough to permit the entire crop to be harvested in two or three picks without berry loss. Little information was obtained on this trait as highest priority was given to saving all the seed obtained from the hand crosses, so most of the berries were harvested when first ripe, before they had time to shatter. Berry retention on nine F₁ plants that were evaluated for this trait varied widely. From five plants, the ripe berries detached too readily; on four others, the berries remained firmly attached.

The skin of ripe berries of all F₁ plants was black or dark purple, which was also true of most of the original V. stamineum plants in the forest. The berries in some of the wild V. stamineum were glaucous due to surface wax, but the berry surface of the F₁ plants was not glaucous or only slightly glaucous.

Flesh color of the ripe berries was of interest, because undamaged berries of highbush blueberry and all other species in section Cyanococcus have green/yellow/white flesh, because the anthocyanins are limited to the skin. By contrast, the berries of some species in other Vaccinium sections have purple flesh and intense pigmentation throughout. Bilberry, V. myrtillus, has black to dark-purple flesh, which is considered desirable by consumers because of possible health benefits. In assessing the flesh color of the hybrids, care was taken to evaluate only berries that were not overripe and had no damage to the skin. The berries of the tetraploid V. stamineum parent FL13-872 have black skins and bright red flesh. Of the F₁ plants evaluated for flesh color, all 10 seedlings with V. stamineum FL13-872 as pollen parent had some internal pigmentation, the intensity varying from dark purple for two seedlings, bright red for two, pink for four, and greenish pink for two. Of the eight evaluated seedlings whose V. stamineum parent was FL13-869, seven had white/green flesh like highbush and one had purple flesh. All 14 evaluated seedlings from V. stamineum FL13-870 had green to white flesh. Of the tetraploid V. stamineum plants that were crossed with highbush, only FL13-872 and FL13-1155 had red to purple flesh; the other tetraploid V. stamineum clones used as parents had berries with green to white internal flesh color. The fact that flesh color varies from purple to red to yellow to green among V. stamineum plants in a single forest in Alachua County, FL, implies that one or a few qualitative genes control flesh pigmentation. If this is the case, internal red pigmentation from V. stamineum may be relatively easy to transfer into the highbush gene pool.

In the wild V. stamineum population from which the parents came, the berries on all plants were firm and juicy. Berries on some plants had a crisp or crunchy texture. On a few of the wild plants, the berries were large and sweet (but with bitterness ranging from slight to medium) 10 to 14 d before the color changed from green to purple. One-hundred purple ripe berries picked from one such plant showed no decay or softening after 3 weeks storage in a paper bag at 22 °C, although the berries shriveled somewhat. Berry firmness, evaluated for 16 of the F₁ hybrids, ranged from high to very high compared with highbush blueberry cultivars such as Star, Emerald, and Farthing. Deerberries tend to have a tough skin, which can detract somewhat from their palatability but probably increases firmness and berry postharvest life.

Ballington et al. (1984b) found that deerberries grown in North Carolina from seeds obtained from wild populations in South Carolina and Florida had much higher soluble solids than the berries of highbush and rabbiteye cultivars with which they were compared. On 5 July 2012, Brix was determined for deerberries from the forest in southeastern Alachua County that was the source of V stamineum parents used in this study. Brix ranged from 15.1 to 21.1 for three randomly selected plants. Another plant, selected because it had berries that seemed very firm and sweet, had a Brix of 20.6 for ripe berries picked from the bush and 24.0 for firm berries gathered from the ground beneath the bush. The Brix of frozen berries from the highbush cultivar Jewel, determined at the same time, was 15.0. Brix of ripe berries from four V. stamineum plant tested from the greenhouse in 2015 ranged from 16.1 to 20.0, compared with 10.0 to 13.8 for three highbush cultivars growing in the same greenhouse.

Flavor components (bitterness, acidity, sweetness, and aromatics) were evaluated for 29 berries, one harvested from each of 29 F₁ plants in the greenhouse. Only one berry per plant was sampled for flavor because saving the hybrid seeds was given priority. Flavors were probably affected by the fact that the F₁ plants were not very healthy during the fruit development period due to mites and other factors, which turned many of the leaves brownish or yellow and caused some defoliation. There was some bitterness in the skins of nearly all berries, but much less than in V. stamineum, and the skins of seven berries were not bitter. The berries of only one plant had detectable bitterness in the flesh. Berries from two plants were considered bland. Berries from 15 plants were considered tart, but most of these were described as “tart but pleasant.” Berry flesh for 8 plants was described as “sweet.” Berries from three plants were described as “aromatic.” Flavor of berries from 10 plants seemed as good as or better than that of the highbush cultivars Star, Emerald, and Farthing. Ballington (personal communication) reported that NC2558 (¼ V. stamineum, ¼ V. darrowii, and ½ highbush), when backcrossed to the cultivar Echota, produced a seedling population with unusually bland fruit, but when crossed with NC967, the population segregated for fruit quality. The preliminary flavor evaluations from the present study indicate that berries with good to excellent flavor should be attainable after recombining highbush and deerberry genes.

Rooting of softwood cuttings from 14 F₁ intersectional hybrid plants.

Percentage rooting of the softwood cuttings was 83% or higher for eight of the clones and four other clones rooted between 56% and 74%. One clone, which defoliated shortly after the cuttings were stuck, rooted at 40%, and one clone produced not a single rooted cutting from 20 that were stuck, even though the cuttings retained their leaves and looked healthy during 50 d under mist.

Discussion

Hybrids between section Cyanococcus taxa and V. stamineum, were relatively easy to obtain, both at the diploid (V. darrowii and diploid V. fuscatum × diploid V. stamineum) and at the tetraploid (highbush cultivars × tetraploid V. stamineum) levels (Table 2). The average of 0.25 to 0.29 good seeds per pollinated flower obtained in the diploid crosses (Table 2) is far below the 20 to 30 good seeds typical of homoploid crosses in section Cyanococcus, but is more than enough to generate large hybrid populations with relative ease.

With the tetraploid crosses, highbush cultivar × tetraploid V. stamineum gave an average of 0.68, 0.77, and 0.60 good seeds per pollinated flower in the three crossing seasons 2013, 2014, and 2015, with a total of 12,225 flowers pollinated. Again, these numbers are well below what would be expected from highbush × highbush crosses or from crosses between two V. stamineum clones, but they permit large populations of hybrids to be grown. Moreover, far more seedlings were obtained from these intersectional homoploid crosses than were obtained by crossing tetraploid (highbush cultivars) and diploid (V. elliottii) members of the same section (Lyrene, 2014). This contrast shows that the triploid block is very strong in section Cyanococcus.

Lack of full compatibility in the intersectional crosses probably reflects the extent of evolutionary divergence between the two sections. Low pollen quality from the colchiploid V. stamineum plants seems not to have been a problem. Although autotetraploid plants obtained by treating diploid plants with colchicine often have reduced fertility compared with their diploid antecedents, in part because they produce frequent aneuploid gametes (Crane and Lawrence, 1952; Hagberg and Ellerstrom, 1959; Sanford, 1983), pollen tetrads from five of the tetraploid V. stamineum plants that were examined at ×250, apart from its large size, seemed normal for V. stamineum. One clone was estimated to have 70% viable microspores and estimated microspore viability for the other four clones was 90% to 95%. Similar high pollen quality was found in colchicine-produced tetraploid V. arboreum (Lyrene, 2011). At least two of the V. stamineum tetraploid clones used as parents were sectorial chimeras, with some branches producing conspicuously smaller flowers whose pollen was the normal size for diploids, but most of the pollen used in the crosses was from large flowers with large pollen. The tetraploid V. stamineum plants used as parents varied considerably in their ability to produce seeds when crossed onto highbush females (Table 1). At least two V. stamineum plants (not shown in Table 1) thought to be tetraploid based on pollen size gave few or no seeds when crossed onto highbush.

After reviewing many experiments in wide hybridization of plants, Darwin (1859) concluded that three measures of crossing success: the ease with which hybrids can be produced, the vigor of the resulting hybrids, and the fertility of the hybrids, are often not well correlated. This observation has been supported by many subsequent examples (Hadley and Openshaw, 1980; Stebbins, 1950). Some crosses that are very hard to make produce vigorous, fertile hybrids. Some crosses that are easy to make give weak or sterile hybrids. With the crosses reported here, the initial hybrids were easy to obtain, were low in vigor compared with their parents, but were surprisingly high in fertility, especially when used as female parents, both in intercrosses and in backcrosses to either parent.

After a comprehensive review of literature dealing with the rate of divergence of daughter taxa from a common ancestor, Levin (2012, 2013) concluded that hybrid sterility normally develops several million years before cross-incompatibility. In herbaceous taxa having the same ploidy level, hybrid sterility typically requires 4–5 million years to develop in subpopulations that have become isolated, and cross-incompatibility requires 8–10 million years. In trees, the waiting time for the development of sterility and cross-incompatibility are much longer, because the rate of molecular evolution in trees is roughly two to five times slower than in herbaceous plants (Levin, 2013). The extent of reduction in both cross-compatibility and hybrid fertility in the intersectional crosses reported here gives some indication of the amount of time that has passed since section Cyanococcus and section Polycodium diverged from their common ancestor.

The results of intersectional crosses of highbush cultivars with V. stamineum can be compared with previously reported crosses of highbush cultivars with V. arboreum (section Batodendron; Lyrene, 2011). In both cases, tetraploid forms of the diploid species were obtained using colchicine, and male fertility of these autotetraploids appeared to be only slightly reduced from that of the diploids based on examination of the pollen at ×250. Intersectional hybrids with highbush cultivars were much easier to obtain using V. stamineum pollen (0.71 seeds per pollinated flower based on 12,225 flowers compared with 0.09 seeds per pollinated flower based on 17,968 flowers for V. arboreum). Both types of crosses produced a high percentage of very weak F₁ plants, but a fraction of the V. arboreum F₁ hybrids were highly vigorous whereas even the most vigorous V. stamineum F₁ hybrids were less vigorous than either parent. With respect to F₁ male fertility, the two types of hybrids were quite similar. With V. arboreum, median microspore viability in F₁ hybrids, estimated by microscopic examination, was 40% based on 91 hybrids; with V. stamineum the median was 30% based on 44 hybrids. With V. arboreum, 2658 highbush flowers pollinated with pollen from F₁ plants averaged 1.24 seeds per pollinated flower and with V. stamineum hybrids, 3250 highbush flower pollinated with pollen from F₁ plants gave an average of 1.47 seeds per pollinated flower. The reciprocal crosses, however, showed a marked contrast between V. arboreum and V. stamineum hybrids. Vaccinium arboreum F₁ hybrids pollinated with highbush pollen gave an average of 0.81 seeds per pollinated flower (766 flowers), whereas V. stamineum F₁ hybrids pollinated with highbush pollen averaged 7.00 seeds per pollinated flower (1102 flowers). Overall, in the comparisons, the V. arboreum hybrids seemed to be superior in vigor and the V. stamineum hybrids higher in fertility.

Hybridization with highbush blueberry, followed by backcrosses of the F₁ hybrids to V. stamineum, could be the first steps in domesticating V. stamineum as a new crop. Traits needed from highbush are easy propagation from softwood cuttings, elimination of objectionable bitter components from the fruit, and good retention of ripe berries on the plant. Ballington (personal communication) found that F₁ hybrids between diploid V. corymbosum and V. stamineum were generally easy to root from leafy stem cuttings. Of the 14 F₁ highbush × V. stamineum hybrids tested in the present study, 8 rooted readily from softwood cuttings. The only hybrid for which no cuttings rooted behaved like V. stamineum, in that cuttings maintained their leaves well under mist but formed no roots. If backcrosses to V. stamineum produce some seedlings that root readily from cuttings, one major obstacle to domestication of V. stamineum will probably be easy to overcome. Bitterness was much reduced in the berries of the F₁ hybrids compared with the bitterness of V. stamineum berries. Some F₁ hybrids retained their ripe berries well on the bush. For these three traits, the phenotypes of the hybrids gave encouraging indications that desirable highbush characters could be transferred into the V. stamineum gene pool.

The potential to obtain useful traits from V. stamineum to improve highbush blueberry cultivars also received encouragement from the phenotypes of the F₁ hybrids. Red pigmentation in the pulp of berries from one family shows that this character might be easily transferred. The hybrids showed the ability to flower on new spring growth, and their flowering racemes were very long and open. Vaccinium stamineum is a late-ripening species (Ballington et al., 1984a; Sharpe and Sherman, 1971) and could be used to breed highbush varieties that ripen later in the growing season. Whether hybrid plants will have better adaptation to upland soils remains to be seen. Recovery of commercially useful highbush cultivars after introgression of genes from V. stamineum should be made easier by the fact that the form of V. stamineum used in these crosses has several characteristics resembling highbush blueberry cultivars now being cultivated. The berries are large and juicy, with high Brix. Vaccinium stamineum resembles highbush blueberry in plant size, shape, and architecture. The flowers shed large amounts of pollen and are capable of high yield.

The low vigor of the tetraploid F₁ hybrids in the 2014 field nursery could portend problems in the backcross and intercross generations. However, some of the F₁ hybrids that were dug for use in crosses grew with medium to high vigor in a soil-peat mix in 10 L pots in the greenhouse when mites were controlled. Vigorous plants are expected in the backcross generations, and a better understanding of their cultural requirements will probably enable identification of vigorous highbush–V. stamineum introgression lines.

One exciting aspect of any program of wide hybridization is the possibility of obtaining plants with surprising characteristics not seen in either parent. In segregating generations following wide hybridization, discovery of plants having novel characteristics that would never have been predicted from a study of the original parents of the cross is not uncommon in other taxa (Stebbins, 1950; Stevenson, 1965).