Introduction

The commercial strawberry, Fragaria × ananassa Duchesne in Lamarck, originated about 250 years ago when a few North American clones of F. chiloensis (L.) Miller and F. virginiana Miller accidentally hybridized in European gardens (Darrow 1966). Thomas A. Knight began the systematic breeding of strawberries in England in 1817, but had at his disposal only a small number of native and cultivated clones. Likewise, North American genetic improvement began in the mid-1800s with a restricted group of European F. × ananassa cultivars, South American F. chiloensis and North American F. virginiana. The cultivars originating from this background played the predominant role in most public and private breeding programs for the next 100 years.

While impressive breeding progress has been made using this narrow germplasm base, other horticulturally useful genes likely are available in native populations of Fragaria as both octoploid species have extensive geographical ranges that encompass a broad range of biotic and abiotic stresses (Hancock et al. 2004; Staudt 1999). Contained within the wild germplasm is a wide range of interesting flavors and aromas, unusual resistance to heat, drought and salinity, almost a continuum of photoperiod sensitivities, and tolerance to a wide variety of diseases and pests (Hancock 1999; Hancock et al. 1990; Luby et al. 1991).

We have spent over a decade cataloging horticulturally useful traits in native populations of strawberry (Hancock et al. 2001a, b; Serçe and Hancock 2002, 2005a; Serçe et al. 2002) and utilizing that variability (Hancock et al. 2001b, c; Hancock et al. 2002, 2003). One of our goals has been to reconstruct F. × ananassa using elite native octoploid clones (Hancock et al. 1993). To this end, we have collected and evaluated native Fragaria genotypes from all across North America. We examined over 600 genotypes of F. virginiana ssp. glauca from the northern Rocky Mountains (Hokanson et al. 1993; Luby et al. 1992; Sakin et al. 1997), and nearly 2,100 genotypes of F. virginiana ssp. virginiana from the central United States and Ontario, along with a few representatives from Alaska, Alberta, New York, Pennsylvania, and western North Carolina (Dale et al. 1993; Luby and Stahler 1993; Luby et al. 1991). From this group, we selected 15 elite genotypes of diverse geographic origin that appeared to be well adapted to the climate around the Great Lakes of the USA and Canada.

Based on our own evaluations and previous published reports, we also assembled an elite group of nine North American F. chiloensis genotypes that possessed horticulturally important traits and represented a considerable portion of the geographical range of this species in North America (Hancock et al. 2001a, 2003, 2005). We crossed the elite groups of F. chiloensis and F. virginiana with one another and evaluated the hybrid progeny. Herein, we describe the variation for several important horticultural traits in these reconstructed F. × ananassa populations including winter injury, photoperiod sensitivity, flowering date, fruit size, female fertility, and disease resistance and relate these findings to our previous evaluations of crosses of the F. virginiana parents with F. × ananassa cultivars (Hancock et al. 2002).

Materials and methods

Fifteen wild F. virginiana selections were included in the study: Eagle 14 (PI 612492), Frederick 9 (PI 612493), Hemlo 2, High Falls 22, and Montreal River 10 (PI 612497) from Ontario, LH 10-6 from Wyoming, LH 28-1, LH 30-4 (PI 612501), LH 39-15, and LH 40-4 from Montana, RH 18 from New York, RH 23 (PI 612498) and RH 30 (PI 612499) from Minnesota, N8417 from Alberta, and RH 43 (N 8688, PI 612496) from Alaska (Hancock et al. 2001a). The LH selections, N8417, and RH 43 represent F. virginiana ssp. glauca, while the others are F. virginiana ssp. virginiana. The RH selections were previously determined to be day-neutral, resistant to black root rot caused by unknown pathogens in Minnesota and at least partially hermaphroditic, with variable anther size and quality. The LH selections had among the largest fruit of all F. virginiana ssp. glauca evaluated, were day-neutral and were partial hermaphrodites. High Falls 22, Montreal River 10, RH 23, and RH 43, also have unusually large fruit for native genotypes. All the Ontario genotypes were previously shown to be field-resistant to powdery mildew [Sphaerotheca macularis (Wallr.:Fr.) Jacz. F. sp. Fragariae Peries] and leaf scorch [Diplocarpon earliana (Ellis & Everh.) Wolf]. Montreal River 10, Eagle 14, and High Falls 22 are short-day plants, while Hemlo 2 and Frederick 9 are day-neutral. Eagle 14 and Montreal River 10 are fertile hermaphrodites, while High Falls 22 is a female and Frederick 9 and Hemlo 2 are males.

The group of native North American F. chiloensis clones included BSP 14, RCP 37, and Pigeon Point (PI 551728) from California, HM1 (PI 612489) from Oregon, FRA 0319, FRA 0596, FRA 0688 (PI 612487), FRA 0883, and FRA 1267 (PI 612488) from British Columbia (Hancock et al. 2001a). Pigeon Point is female, FRA 0319, FRA 0596, and FRA 0688 have large fruit (particularly FRA 0688), are female, and are probably mildew resistant. FRA 1267 is mildew susceptible, but has large fruit and is a very fertile hermaphrodite (Dale et al. 1993). FRA 0883 is a partial hermaphrodite with mildew resistance. BSP 14 and RCP 37 are male and resistant to aphids, two-spotted mites, red stele, leaf spots, and powdery mildew. HM1 is female and has an unusually high fruit number for F. chiloensis.

Hybridizations were attempted for all interspecific combinations between the 14 F. virginiana and nine F. chiloensis selections and seeds from 100 combinations were obtained. Each F. virginiana selection was successfully crossed with four to nine of the F. chiloensis parents. In most crosses, except those involving the F. chiloensis BSP 14 and RCP 37, and F. virginiana, High Falls 22, the F. virginiana genotype was used as the male parent.

Seedlings of each family were germinated and grown to the four-leaf stage and approximately 1,100 plants from 98 families were subsequently set in May 1996 at Becker, Minnesota and 1,500 plants from 92 families at Simcoe, Ontario. The plants were set at 1.2 × 1.2 m spacing and their runners were trained by cross cultivation into a 0.6 × 0.6 m square. In Minnesota, up to four plants of each family were set in each of four blocks in a randomized complete block design with non-contiguous plots (Libby and Cockerham 1980) whilst in Ontario, a nested design was employed in which all plants from a single family were planted adjacent to one another, but families were randomized in the field.

The presence of new flowers was also recorded at 1 or 2 week intervals through 16 September to determine whether the genotypes were short-day plants that flowered only in the spring, or day-neutral that flowered in both the spring and mid-summer. A genotype was considered a day-neutral if it flowered in the spring and again in the summer between 15 July and 26 August (approximately 13.5 h photoperiod) in Minnesota and 12 August–19 September in Ontario. Genotypes that fruited in the spring and then not again until after the above dates were considered short-day types.

When at least 30–50% of the fruit of each plant were ripe, the width of the largest primary fruit in each plot was measured, and subjective estimates were made for each plant on the average number of flowers per inflorescence (1—only primaries, 2—primaries and secondaries, 3—primaries through tertiaries, and 4—primaries through quaternaries), and the percentage fruit set (1–10 representing 10% intervals).

In Minnesota, we also recorded the date when each plant began spring bloom as the number of as days after 30 April. Plants were not protected with mulch over the previous winter so winter injury was rated for each plot based on proportion of plants killed during the winter and vigor of surviving tissue using a scale from 0 = severe injury with ≥90% of plants dead; 1 = ∼90–70% of plants dead with greatly reduced vigor on surviving crowns; 2 = ∼60–40% of plants dead with reduced vigor on surviving crowns; 3 = ∼40–10% of plants dead with reduced vigor on surviving crowns; 4 = up to 10% crown mortality with only slightly reduced vigor on live crowns; to 5 = no injury. When the plants were fruiting, the average percentage of ovules set as filled achenes was visually estimated on a scale from 1 to 10 representing 10% intervals. In August 1997, disease incidence was estimated on a 1–5 scale for leaf scorch and powdery mildew, where 1 = 0–10% of the leaf surface was affected, while 5 = 80–100% of the leaf surface was affected. In Ontario, in late June, plant vigor was rated on a scale from 1(low vigor) to 3 (high vigor) and the average number of inflorescences per crown estimated ranging from 1 to 3.

Analyses of variance (ANOVA) were conducted using the SAS GLM procedure (SAS, 1990). The variance among full-sib families was analyzed as a factorial design according to Comstock and Robinson (1948) and components of variance for general combining ability (GCA) and specific combining ability (SCA) were partitioned from the variance among full-sib families (Table 1). ANOVA for Minnesota data used a cross classified design for blocks and families. ANOVA for the Ontario data were conducted based on the nested design proposed by Gilbert (1967) that tests combining ability mean squares using an error mean square based on pooled variance among plants within families. This is a partially confounded design in which the term for SCA is expected to contain a portion of the environmental variation across the field in addition to the genetic interactions among parental pairs. Variables with percentage data were transformed using arcsine square root transformation to improve normality for ANOVA. However, the means were calculated using original data sets. Variance components were calculated for random factors using VARCOMP procedure. Estimations were considered as 0 when they were negative. Correlations were calculated based on full-sib family means to examine interactions between the various yield components (Bedard et al. 1971; Lacey 1973). Half-sib family means were estimated based on the means of full-sib families and a standard error of these means is presented based on the pooled deviations of full-sib family means. The nested design used in Ontario did not permit analysis of variance for proportion of day-neutral progeny in families. As the trait is of considerable interest in F. virginiana germplasm, however, we calculated the proportion of day-neutral progeny in each half-sib family as follows: the percentage for each full-sib family was calculated from the number of day-neutral progeny/total progeny. Then, half-sib family means were obtained for each parent. A pooled standard error of half-sib family means was calculated from the deviations of the means of the full-sib families in each half-sib family.

Table 1 Expected mean squares (EMS) and degrees of freedom (df) in the analysis of variance for Fragaria chiloensis × F. virginiana crosses
Full size table

Results

General combining ability was significant for all traits in at least one species except ovule set in Minnesota and flower number in Ontario (Tables 2 and 3). SCA was significant only for bloom date in Minnesota (Table 2). The F-test for SCA term was significant for all traits in Ontario (Table 3) though, as noted in the “Materials and methods”, this term is partially confounded with environmental variation. The GCA components of variance were significant for both F. chiloensis and F. virginiana for all traits except, for fruit diameter in Minnesota and % fruit set in Ontario, which were significant only for F. chiloensis parents.

Table 2 Mean squares from analysis of variance and components of variance (% of total variance in parentheses) due to general combining ability (σGCA) of Fragaria chiloensis and F. virginiana parents and specific combining ability (σSCA) for several horticulturally important traits in full-sib families from F. chiloensis × F. virginiana crosses grown in Minnesota
Full size table
Table 3 Mean squares from analysis of variance and components of variance (% of total variance in parentheses) due to general combining ability (σGCA) of Fragaria chiloensis and F. virginiana parents and specific combining ability (σSCA) for several horticulturally important traits in full-sib families from F. chiloensis × F. virginiana crosses grown in Ontario
Full size table

In Minnesota, family means varied considerably for most of the traits (Table 4). Families from the British Columbia F. chiloensis parents (FRA 1267, FRA 0319, and FRA 0883) exhibited the least winter injury, while families from California F. chiloensis parents (BSP 14, Pigeon Point, and RCP 37) were the least hardy. The F. chiloensis parents BSP 14, FRA 0688 and FRA 0883 and the F. virginiana parents, Fredrick 9, Hemlo 2 and High Falls 22 produced families that exhibited the least leaf scorch. Powdery mildew infection was minimal in all families. Means for spring bloom date varied by approximately 5 days among F. chiloensis and F. virginiana families. The earliest families were derived from FRA 1267 and FRA 319 on the F. chiloensis side and Eagle 14, Hemlo 2, and RH 23 on the F. virginiana side. The latest families came from the F. chiloensis parents BSP 14, HM 1 and Pigeon Point and the F. virginiana parents LH 10-6, LH 30-4, Montreal River 10, and RH30. Estimates of fruit set ranged from 39 to 71% and ovule set ranged from 62 to 79%. Most families produced fruit with a mean diameter between 13 and 15 mm. The LH 30-2 families had the smallest fruit at 12.4 mm while the Montreal River 10 families produced the largest fruit at over 17 mm. Families having larger mean fruit diameter also had higher estimated fruit set and ovule set (r = 0.48 and 0.73, respectively, P < 0.001). Flower number was not significantly correlated with fruit set, ovule set or fruit diameter.

Table 4 Means of half-sib families from Fragaria chiloensis and F. virginiana parents for several horticulturally important traits in crosses of F. chiloensis parents with F. virginiana parents tested in Minnesota
Full size table

In Ontario, family means also varied considerably for most of the traits (Table 5). Most families were vigorous, but the families derived from the two F. virginiana ssp. glauca genotypes, N 8417 and RH 43, and most of the F. virginiana ssp. virginiana genotypes from Montana and Wyoming (LH 10-6, LH 30-4, and LH40-4) were less vigorous than the other F. virginiana genotypes. Most family means for fruit diameter ranged from 5.7 to 8.7 mm, but the F. virginiana genotypes, High Falls 22 and Montreal River 10, and F. chiloensis genotypes, Pigeon Point and RCP37 were larger and the F. virginiana genotype, N 8417 was smaller. For fruit set, family means for FRA1267, High Falls 22, and RH30 were greater than 50%, and for FRA883, Pigeon Point, and Frederick 9 and RH23, were less than 40%. For the number of inflorescences, the families of FRA 0883, Hemlo 2, High Falls 22, and RH30 had more than two per crown, whilst the families of N8417 and the LH series had less than 1.5 inflorescences per crown. The families of those genotypes that were more vigorous had larger fruits and more inflorescences (r = 0.50 and 0.55, P < 0.02 and 0.01, respectively) and those with more inflorescences had more flowers per inflorescence (r = 0.58, P < 0.01).

Table 5 Means of half-sib families from Fragaria chiloensis × F. virginiana parents for several horticulturally important traits in crosses of F. chiloensis parents with F. virginian a parents tested in Ontario
Full size table

The traits that were measured at both locations were significantly correlated (flower number, fruit set, and fruit diameter, r = 0.68, 0.56, and 0.53, P = 0.001, 0.01, and 0.02, respectively).

In Minnesota, all parental genotypes had at least one individual exhibiting the day-neutral flowering habit within their half-sib families whilst in Ontario one-third of the parental genotypes had no individuals express the trait. In most families, less than 10–15% of the plants were day-neutral; however, in the F. virginiana family of RH 43 over 20% of the plants expressed the trait. Families with an earlier spring bloom date tended to have a significantly higher proportion of day-neutral progeny (r = −0.39, P < 0.001) in Minnesota, but not in Ontario.

Discussion

Our results suggest that F. virginiana and F. chiloensis parents will need to be chosen carefully when F. × ananassa populations are reconstructed. This should be relatively easy as significant levels of GCA were found for all the production traits, except ovule set, and no significant negative correlations were observed between the various yield components (both within and between the locations). For F. virginiana, this confirms our previous conclusions (Hancock et al. 2002). The significant F-tests for SCA in the Ontario trial may suggest that some care is needed to match specific combinations of parents to maximize the genetic gains, however it may simply be a reflection of inflation due to the confounding of environmental effects with parental interaction effect in the nested design used at this location.

Different F. virginiana and F. chiloensis parents should be selected in different environments so that the resulting offspring are appropriately adapted to local conditions. For example, the more northerly F. chiloensis parents from British Columbia produced more winter hardy offspring in our test in the cold, continental climate of Minnesota and the F. virginiana ssp. glauca and most of the F. virginiana ssp. virginiana genotypes from Montana and Wyoming were less vigorous in Ontario. The F. chiloensis or F. virginiana parent could affect mean spring bloom date by up to 5 days. Also, several parents produced progeny that were notably more resistant to the prevalent foliar diseases.

The degree of progress made in introducing genetic variation from F. virginiana into the cultivated strawberry gene pool will depend on the other parent used, either through reconstitution of F. × ananassa or backcrossing. In our previous study, the F. × ananassa parents used contributed little to the variability expressed (Hancock et al. 2002), whereas here F. chiloensis contributed as much variability as F. virginiana. This is perhaps to be expected as F. × ananassa has a very narrow genetic background (Sjulin and Dale 1987).

The genetics of day-neutral flowering in Fragaria is still being elucidated. In previous work, the environments where the segregating populations are tested and the evaluation methods greatly affect the expression of day-neutrality (Hancock et al. 2002; Serçe and Hancock 2003, 2005b). Genetic control has been proposed to be by a single locus earlier (Ahmadi et al. 1990); however, when we have crossed F. virginiana genotypes which may carry different day-neutral genes to F. × ananassa cultivars, many segregation ratios do not fit the single-gene model (Hancock et al. 2002). In other work where several F. virginiana and F. chiloensis genotypes were crossed to F. × ananassa cultivars which come from a diverse genetic background (Serçe and Hancock 2005b), none of the families fit the single gene model and several two-gene and polygene models fit better than the single locus ones. Shaw (2003) also recovered heterogeneities in segregation ratios when he tested the genetics of day-neutrality among a limited number of F. × ananassa x F. × ananassa crosses, but further analysis of many more families led him to suggest that day-neutrality was regulated by a single major gene (Shaw and Famula 2005). Most recently, Weebadde et al. (2007) found a major QTL for day-neutrality in a segregating population of short day ‘Honeoye’ × day-neutral ‘Tribute’ in Minnesota, Michigan and Maryland, and a number of minor site specific QTL.

In this set of data, F. chiloensis were strongly short day and suppressed the expression of day-neutrality in the F. virginiana. This set of F. virginiana parents produced on average 30% day-neutral offspring when crossed with short day F. × ananassa parents, depending on the test location (Hancock et al. 2002), they produced only 7.5% day-neutral progeny in crosses with F. chiloensis parents in this study. The results of these crosses are consistent with other studies suggesting day-neutral flowering in octoploid strawberry is likely under polygenic control (Hancock et al. 2002; Serçe and Hancock 2005b) and that short day F. × ananassa carry some of the genes required for day length insensitivity. Indeed our study suggests that the expression of day-neutrality is correlated with early flowering which has been previously reported by Barritt et al. (1982).

In conclusion, sufficient variability exists within F. chiloensis and F. virginiana that superior types of F. × ananassa might be produced through the intercrossing of elite clones of each species, followed by several generations of pedigree breeding. A key to success might be to use Chilean genotypes of F. chiloensis that have been shown to be much larger fruited than North American ones (Hancock et al. 2005). A reconstruction of F. × ananassa using elite clones of F. virginiana and F. chiloensis may be more efficient than simply backcrossing into the F. × ananassa background, as a much higher proportion of unique genes will be available for recombination in later generations.

We suggest a four-stage process of reconstruction: (1) Select elite clones of F. chiloensis and F. virginiana. We have performed a number of screens which can be used to select these genotypes (Hancock et al. 2001a, 2003, 2005). (2) Hybridize the elite selections of each species and select the superior progeny. This process will maximize the amount of beneficial genetic variability carried within each species for subsequent breeding. (3) Make diallel crosses between elite selections of the two species which contain complimentary traits. This process will generate new gene combinations and detect positive epistatic interactions. (4) Select the most promising genotypes from the various families and cross complementary genotypes amongst themselves. From their progeny, elite material can be selected for further breeding and/or varietal release.