The Use of elliottii Chapmn. in Breeding Highbush

P.M. Lyrene

Horticultural Sciences Department, University of Florida, Gainesville, FL 32611

Subject category: breeding and genetics.

Abstract. Eighty vigorous V. elliottii seedlings were selected from an 8-km stretch of Perone Creek, in southwest Alabama near Silverhill. Four ramets of each were obtained by rooting softwood cuttings taken in August 2013. These were grown in pots of peat in a greenhouse until March 2014. They began to in early February, and on March 1 they were placed outside the greenhouse for open pollination. Nineteen tetraploid southern highbush selections, growing in 20-liter pots, which had been chilled in a cooler, were placed in a bee-proof greenhouse in

February 2014. on these were emasculated before they opened, and approximately 200 flowers per plant were pollinated with pollen from one or more of the potted V. elliottii selections. From 4301 pollinated flowers, only 78 seeds judged to be germinable were obtained. By contrast, crosses between tetraploid highbush cultvars gave 6720 seeds from 349 pollinated flowers and crosses between two V. elliottii clones gave 8000 seeds from 380 flowers. The 80 V. elliottii clones placed outside for pollination received few or no bee visits during flowering, which continued during March and April, and set was less than

1%. Mean weight of from several hundred V. elliottii plants sampled at five forested locations in northwest Florida and southwest Alabama ranged from

184 mg to 341 mg depending on the location. Two sample sites in Alabama were along Perone Creek, and were sites from which plants had been propagated for the crossing experiments. Two V. elliottii plants with fastigiate growth habit were found, one at Perone Creek and one along the Suwannee River in Hamilton

County, Florida.

Index words. , Mayberry, Elliott’s blueberry, blueberry breeding, interspecific hybridization, pollination ecololgy.

Introduction

Vaccinium elliottii, in section Cyanococcus, is a diploid, , crown- forming blueberry species, 2-4 m high, native in the southeastern from southeastern Virginia, south to the northern Florida peninsula, and west into

Louisiana and Arkansas (Camp, 1945). It is my opinion that most or all of the plants collected from the Florida peninsula south of Gainesville under the name V. elliottii are not V. elliottii, but are products of hybridization and introgression between V. darrowii and diploid V. fuscatum. During the summer, some of these hybrids superficially resemble V. elliottii, but during the winter they are far more evergreen, during flowering, they lack the distinctive short styles of V. elliottii, and they usually occur as minor components of much larger populations that include both parent species and their hybrids, many of which look quite unlike V.elliottii. It is quite inconceivable that a single plant of V. elliottii, should it appear in this part of peninsular Florida, could reproduce or maintain its integrity as a species when surrounded by numerous plants of V. darrowii, diploid V. fuscatum, and hybrids of the two.

V.elliottii has small berries, usually black but sometimes glaucous, with thin skins, small seeds, and a pleasant flavor, with a balance of tartness and sweetness similar to that of highbush .

Although Vander Kloet’s phenetic (Vander Kloet, 1980) combined various diploid, tetraploid, and hexaploid under the name V. corymbosum, I prefer the phylogenetic treatment of Camp for several reasons. A phylogenetic taxonomy, because it attempts to incorporate the evolutionary history of the taxa, better delimits the gene pools available to breeders. Differences in chromosome number are a fundamental factor restricting gene flow in Vaccinium, both in nature and in a breeding program, and differences in chromosome number are given high priority in Camp’s classification of the highbush blueberries but are afforded little taxonomic significance by Vander Kloet. The sampling method on which Vander Kloet based his taxonomy (Vander Kloet, 1980) gave undue weight to rare natural hybrids at the expense of the far more numerous plants of the parent species that were present at the locations he sampled. Camp’s taxonomy, much better than Vander Kloet’s, corresponds to what I have observed during 35 years of studying Vaccinium in the forests of Florida and in greenhouse crosses.

The allozyme data of Bruederle and Vorsa (Vorsa, 1996) supported maintenance of V. elliottii as a distinct species as did the studies of Uttal (1987) of the genus Vaccinium in Virginia. The short style of V. elliottii flowers, which places the stigma at the same level as the distal end of the anthers and well inside the corolla tube (Lyrene, 1994; Vander Kloet, 1998), provides a simple key character for separating V. elliottii from diploid highbush (V. fuscatum, V. atrococcum).

Homoploid interspecific crosses are easy to make in Vaccinium section

Cyanococcus. Crosses in Vaccinium between diploids and tetraploids encounter a strong triploid block. With great effort, triploid hybrids can be produced, and the frequency of tetraploid hybrids varies greatly, depending on the frequency with which the diploid parent produces 2n gametes.

Hybrids between cultivated tetraploid highbush blueberry and diploid

V. elliottii have been produced in several breeding programs. The largest efforts have been in North Carolina and Florida. Traits of interest in V. elliottii include early flowering, short fruit development period, early ripening, and an ability to thrive on upland soils low in organic matter. V. elliottii grows well on sandy and sandy-clay soils where rabbiteye blueberries grow well but highbush cultivars do not. This tolerance to upland sites is probably due to greater drought tolerance in V. elliottii compared to the more swamp-adapted highbush species.

The breeding program at North Carolina State University investigated V. elliottii as a source of genes for early ripening, tolerance to upland soils, and resistance to stem blight (Botryosphaeria dothidea) (Ballington et al., 1997 and

2006). Ballington et al. (2006) reported that of 631 elite southern highbush genotypes that had been selected in the North Carolina breeding program, 105 involved only V. elliottii and V. corymbosum. Of these, 22 were F1, 73 were BC1, and 10 were BC2 selections. Ballington et al. (2009) noted that “unfortunately, it has been difficult to produce BC1 hybrids to V. corymbosum with fruit color and firmness that merit release as improved cultivars.” In addition, in North Carolina, blueberry progenies with V. elliottii in their background seemed to be especially susceptible to Blueberry Red Ringspot Virus or were especially attractive to the insects that vector the disease (Ballington et al., 1997). ‘Carteret’, a vigorous and productive 2005 release from the North Carolina breeding program, is a BC1 from

V. corymbosum x V. elliottii to V. corymbosum. In the Florida breeding program, in 1978 and 1979, approximately 7,000 flowers of tetraploid highbush selections were pollinated with pollen from V. elliottii plants propagated from southwest Alabama and northeastern Florida. From

330 berries that were obtained, 175 seedlings were grown, of which 25 were selected as hybrids based on vegetative, flower, and fruit characteristics after 2 years in a field nursery (Lyrene and Sherman, 1983). Of 18 plants whose chromosomes were counted, 2 were triploid, 11 tetraploid, and 5 had higher chromosome numbers. One tetraploid F1 hybrid, FL81-15, was backcrossed to various tetraploid highbush selections. In 2000 and 2001, the three earliest-ripening selections among 200 clones in 15-plant Florida test plots were full-siblings from a cross in which FL81-15 was both the paternal grandparent and the maternal great- grandparent. In 2008, one of these, FL98-406, was released as the early-ripening , ‘Snowchaser’. This cultivar has proved to be highly vigorous and early ripening, producing a high-quality (although somewhat small) berry with excellent flavor. The success of ‘Snowchaser’ in very low-chill environments and under tunnels renewed our interest in the use of V. elliottii in breeding.

Materials and Methods

It is rare to see a heavy crop of ripe berries on V. elliottii in north Florida and adjacent southwest Alabama. The plants flower very early, and in most years much of the crop is lost to freezes between mid-January and early April. Berries that survive the freezes are highly attractive to birds, and it requires a bumper crop to feed the birds and leave enough berries to allow evaluation of plant-to-plant variation in berry quality. To circumvent this problem, in August 2013, I collected softwood cuttings from each of 80 V. elliottii plants, selected for high vigor and plant health, from cut-over longleaf pine forests at three locations along Perone

Creek, west of Silverhill in southwest Alabama. Five cuttings were taken from each plant and were rooted in a peat-perlite mix in a greenhouse at the University of

Florida in Gainesville. Rooting success was over 90%, and 4 cuttings of each genotype were potted into 10-liter pots of peat, 2 cuttings per pot, making 160 pots in total. The plants were kept in a greenhouse until March 1, 2014, and after flowers began to open, the plants were placed outside for cross-pollination. They were moved inside on frost nights, which were few, and some were moved temporarily into the greenhouse from time to time to provide pollen for use in crosses. Otherwise, the plants were left outside to permit pollination by bees.

In January, 2014, nineteen tetraploid southern highbush culivars and advanced selections that had been growing in 20-liter pots for 2 or 3 years were brought into a bee-proof greenhouse after having been chilled in a cooler. Between

60 and 682 flowers on each plant (depending on the availability of time and flowers) were pollinated with pollen from one or more of approximately 30 different V. elliottii genotypes growing in the pots. Highbush flowers were emasculated before they opened and a heavy load of pollen extracted minutes earlier from newly-opened V. elliottii flowers was applied immediately by thumbnail. Berries were collected when ripe. The first 20 ripe berries from each cross (if there were that many) were opened by hand, and the number of well- developed seeds per berry was determined. The remaining berries were opened, and whatever seeds they contained were carefully separated from the flesh and skins using water. Percent fruit set and total number of apparently-viable seeds were recorded for each cross. Two V. elliottii x V. elliottii and two highbush x highbush crosses were made for comparison.

From 28-30 April 2014 approximately 2 kg of ripe berries were harvested from hundreds of V. elliottii plants growing in open woods and along roadsides in west Florida and southwest Alabama (Table 3). Five locations were sampled, and each sample was a composite of berries from 20 or more plants. In addition, a separate sample was taken from one plant in the forest at Fairhope, Alabama which had a large crop of larger-than-average berries. Mean berry weight was determined for each sample site based on based on 5 subsamples, 50 berries per subsample.

Results and Discussion Surprisingly, the 160 pots of V. elliottii that flowering outside the greenhouse during March and early April attracted few or no bees. Bees were not brought to the site, and only naturally occurring insects were available. The number of flowers per plant averaged more than 100, but fruit set was less than 1%. Flowers from these plants shed abundant pollen when plants were brought into the greenhouse overnight to supply pollen for crosses the next day. The V. elliottii x V. elliottii crosses made in the greenhouse averaged 73% fruit set and produced about 8,000 well-developed seeds from 380 pollinated flowers. It is not known why the plants outside the greenhouse attracted so few bees, but they seemed incapable of setting fruit without cross pollination. Several hundred highbush plants that flowered during the same period in pots a few hundred meters away set full crops of seedy berries.

The unusual flower structure of V. elliottii, in which the short style places the stigma at the same level as the anthers, would seem to favor self-pollination and inhibit cross pollination, but the plants appear to be highly self-incompatible.

Placement of the stigmas and anthers deep inside the long, narrow corolla tube would seem to exclude many pollinating insects, except for sonicating bees such as the southeastern blueberry bee (Habropoda laboriosa) and the Bombus species.

Bees of both types are commonly seen on V. elliottii flowers in forested situations in north Florida. Flowering time for V. elliottii is normally January and February in north

Florida. The potted plants were delayed in flowering because they had been kept growing evergreen in a greenhouse through January, and they formed flower buds late. The flowers produced in the pots outside appeared healthy, and flowering continued from late January through March. Perhaps flowering was too late or the location was unsatisfactory to attract the native pollinators of the species.

Meanwhile, wild V. elliottii plants in the forests of north Florida and southwest Alabama produced a rare heavy crop of berries in April and May 2014.

Steady cold in January and February delayed flowering, and there were no widespread hard freezes during March or April. On April 30, about 10% of the berries were ripe with a heavy crop on a large population of V. elliottii plants growing in cut-over pine forests near Perone Creek west of Silverhill, Alabama and

40% were ripe on May 6. Fruit ripening in commercial peach and blueberry plantings located nearby was 7-14 days later than normal in 2014 due to the cool spring, and the V. elliottii ripening season was probably also later than normal.

From a total of 4301 highbush flowers pollinated in the greenhouse with V. elliottii pollen in 19 crosses, only 78 seeds were obtained that appeared to be viable, an average of only 1.81 good seeds per 100 pollinated flowers (Table 1, Table 2).

By contrast, the V. elliottii x V. elliottii crosses averaged 2100 seeds per 100 flowers and the highbush x highbush averaged over 1900. Thus, the relative success, in terms of the number of good seeds per pollinated flower from the interspecific crosses compared to the parental homoploid crosses was only .000898; pollination of 10,000 flowers in the heteroploid crosses would be expected to give fewer good seeds than pollination of only 9 flowers in homoploid crosses. Five of the heteroploid crosses produced neither fruit nor seed; two additional crosses produced a few berries but no seed (Table 2). The three most successful crosses produced about 5 good seeds per 100 pollinated flowers. Several of the highbush plants used as parents produced large numbers of seedless berries. In most cases, these berries were small compared to fully-seeded berries of the same cultivar, but a few highbush parents gave high set of large, seedless berries after pollination with V. elliottii pollen.

Success rate in the heteroploid crosses varied from one highbush parent to another. The reasons are unknown, but there are several possibilities. The V. elliottii pollen used on different highbush females varied as to source clones. If a single viable seed is not always enough to insure berry retention and development, then highbush clones that are capable of developing berries even when seedless might be better seed parents. Table 2 presents some evidence to support this idea.

Five crosses produced no fruit and therefore could not have produced seed. Two of the most successful highbush parents (in terms of number of viable seeds per 100 pollinated flowers) ripened berries from 60% or more of the pollinated flowers, most of the berries seedless. However, two other relatively successful parents, 12-

228 and 12-252 set only 10% and 4% of their berries, respectively.

Overall, the number of well-developed seeds per pollinated flower in these heteroploid crosses was similar to what we obtained in similar crosses in 1978 and

1979 (Lyrene and Sherman, 1983). Because these success rates are so low, it would seem worthwhile to develop some tetraploid V. elliottii plants with chromosome- doubling methods (Perry and Lyrene, 1984) or to find V. elliottii clones that produce high rates of 2n gametes (Ortiz et al., 1992; Lyrene, et al., 2003). Although only a few tetraploid F1 plants are enough to produce thousands of backcross tetraploid seedlings, we would like to be able to introduce genes from a wider sample of V. elliottii genotypes and to be able to select from a larger F1 population the best plants for further breeding.

Literature Cited

Ballington, J. R., S. D. Rooks, W. O. Cline, J. R. Meyer, and R. D.

Milholland. 1997. The North Carolina State University blueberry breeding program – toward V. x covilleanum. Acta Hort. 446:243-247.

Ballington, J. R., S.D. Rooks, W. T. Bland, and A. D. Draper. 2006. The role of interspecific hybridization in the North Carolina State University Blueberry Breeding Program. Proc. 10th North American Blueberry Research and Extension

Workers Conf. Tifton, Georgia.

Camp, 1945. The North American blueberries with some notes on other groups of Vacciniaceae. Brittonia 5:203-275.

Lyrene, P.M. 1994. Variation within and among blueberry taxa in flower size and shape. J. Amer. Soc. Hort. Sci. 119:1039-1042.

Lyrene, P.M., N. Vorsa, and J.R. Ballington. 2003. Polyploidy and sexual polyploidization in the genus Vaccinium. Euphytica 133:27-36.

Lyrene, P.M. and W. B. Sherman. 1983. Mitotic instability and 2n gamete production in x V. elliottii hybrids. J. Amer. Soc. Hort.

Sci. 108:339-342.

Ortiz, R., N. Vorsa, L.P. Bruederle, and T. Laverty. 1992. Occurrence of unreduced pollen in diploid blueberry species, Vaccinium sect. Cyanococcus.

Theor. Appl. Genet. 85:55-60.

Perry, J.L. and P.M. Lyrene. 1984. In vitro induction of tetraploidy in

Vaccinium darrowi, V. elliottii, and V. darrowi x V. elliottii with colchicine treatment. J. Amer. Soc. Hort. Sci. 109:4-6. Uttal, L. J. 1987. The genus Vaccinium L. () in Virginia. Castanea

52:231-255.

Vander Kloet, S.P. 1980. The taxonomy of the highbush blueberry,

Vaccinium corymbosum. Can. J. Bot. 58:1187-1201.

Vander Kloet, S.P. 1998. The taxonomic status of Vaccinium elliottii, once more with feeling! p 17-23 in Proc. 8th NABREW Conf., Wilmington, N.C.

Vorsa, N. 1997. On a wing: the genetics and taxonomy of Vaccinium species from a pollination perspective. Acta Hort. 446:59-66.

Table 1. Comparative Success of Highbush x Highbush, V. elliottii x V. elliottii, and Highbush x V. elliottii Crosses in 2014

Type of Cross Number of Flowers Number of Seeds Seeds per 100 Flowers

Pollinated

Highbush x Highbush 349 6720 1925.50

V. elliottii x V. elliottii 380 8000 2105.26

Highbush x V. elliottii 4301 78 1.81

Highbush x V. elliottiiz 7000 175 2.50 zLyrene and Sherman, 1983 Table 2. Fruit set percent vs number of viable seeds obtained per 100 pollinated flowers for 14 tetraploid highbush x diploid V. elliottii crosses, 2014. Pollen for all crosses was from diploid V. elliottii.

Highbush parent flowers % set Viable seeds Viable seeds per

pollinated 100 flowers

05-603 682 100 16 2.34

Raven 115 100 1 0.87

06-161 99 92 5 5.05

02-22 545 70 10 1.83

06-340 262 60 14 5.34

12-64 60 55 1 1.67

12-275 144 15 5 3.47

Primadonna 203 13 0 0.00

04-245 259 12 1 0.39

12-228 204 10 10 4.90

01-234 373 9 3 0.80 12-215 206 8 4 1.94

12-252 217 4 8 3.69

06-399 181 1 0 0.00

Five crosses set no fruit and produced no seed. Table 3. Mean weight (mg per berry) of Vaccinium elliottii berries harvested from 6 locations in the western Florida panhandle and adjacent Baldwin County Alabama, April 28-30, 2014. Each location mean (except for Fairhope) is based on 5 samples, 50 berries each, from a composite of

20 or more plants.

Location Mean Standard deviation

Chipley, FL 227 14.5

Ponce de Leon, FL 341 7.0

Harold, FL 184 7.4

Cemetery, Silverhill, AL 241 11.4

Konicy site, Silverhill, AL 291 6.1

Fairhope, AL (one selected plant) 384 10.9