Existence of Genetic Differences in Pellicle Peelability in Japanese Chestnut (Castanea crenata Sieb. et Zucc.) Cultivars and Selections with Difficult-peeling Pellicles

Abstract

The pellicle of the Japanese chestnut (Castanea crenata Sieb. et Zucc.) is difficult to peel, with the exception of the recently developed ‘Porotan’, whose pellicle peelability (PP) is high and controlled by a single major gene. The objective of the present study was to identify any genetic differences in PP in Japanese chestnuts with difficult-peeling pellicles. We detected a significant genetic difference in PP (evaluated as the nut surface area that peeled away without scraping with a hand knife after deep frying in cooking oil) among five Japanese chestnut cultivars grown in Tsukuba, Japan, using two trees in 2 years, and among three cultivars grown in five locations using a single tree in 1 year. We evaluated the PP of 32 Japanese chestnut cultivars/selections and one wild clone (Shibaguri-37) using a single tree in 3 years to quantify the variation. The broad-sense heritability of mean values over the 3 years was estimated as 0.67. Shibaguri-37 had the highest PP. The suggested new genes controlling the variation in PP have high potential in terms of breeding strategy for easy-peeling pellicles as an alternative to the major gene of ‘Porotan’, the use of which is likely to result in inbreeding.

Introduction

The Japanese chestnut (Castanea crenata Sieb. et Zucc.) is well adapted to Japan’s climate, and has been an important source of food and timber for the Japanese people since ancient times (Sawamura, 2006). Today, it is cultivated commercially for nuts because of its large nut size and high eating quality (Pereira-Lorenzo et al., 2012). Chestnut production in Japan totalled 21000 t in 2013 (FAOSTAT, 2016).

Many cultivars of Chinese chestnut (Castanea mollissima Bl.) and European chestnut (Castanea sativa Mill.) cultivars have a pellicle that is easy to peel (hereafter termed an easy-peeling pellicle, or EPP). In contrast, with the exception of the recently developed ‘Porotan’ (Saito et al., 2009), Japanese chestnut cultivars have a pellicle that is difficult to peel (hereafter, a difficult-peeling pellicle, or DPP), even after heating (Kikuchi, 1948; Miller et al., 1996; Pereira-Lorenzo et al., 2012). The pellicle of the Japanese chestnut is often scraped away by hand using a knife, but this is laborious and costly.

The Japanese chestnut breeding program started in 1947 at what is now the Institute of Fruit Tree and Tea Science, National Agriculture and Food Research Organization (NIFTS). Developing new EPP Japanese chestnut cultivars while maintaining large nut size and high productivity has been an important target. Breeders had tried to combine the EPP of the Chinese chestnut with the large nut size, high eating quality, and high productivity of the Japanese chestnut (Pereira-Lorenzo et al., 2012; Sawamura, 2006). However, EPP offspring selected from interspecific hybrids had small nuts and low productivity. In addition, their pellicles were difficult to peel when the trees were pollinated by neighboring Japanese chestnuts (Tanaka and Kotobuki, 1992a). Thus, interspecific breeding failed to give the desired result.

Another possible breeding strategy is to cross Japanese chestnut cultivars if a genetic difference in DPP is present among them, but this must first be ascertained. However, it is difficult to evaluate pellicle peelability (PP) by the conventional evaluation method, which requires roasting of the nuts for 15 to 30 min (Tanaka and Kotobuki, 1992a; Tanaka et al., 1981), which is time-consuming and expensive and limits the evaluation to 20 genotypes a day. Recently, Shoda et al. (2006) developed a high-temperature oil peeling (HOP) method by which PP is evaluated after deep frying in cooking oil. The HOP method shortens the heating time to only 2 min, thereby enabling breeders to evaluate more than 150 genotypes a day. Using this method, breeders at NIFTS discovered an exceptional EPP trait of a selection, which was released as ‘Porotan’ in 2006 (Saito et al., 2009).

The area of ‘Porotan’ cultivation increased rapidly after its release, reaching 158 ha in only 6 years. The EPP trait of ‘Porotan’ is controlled by a single recessive major gene (Takada et al., 2012), and a molecular marker linked to this locus has been developed (Nishio et al., 2013). This has enabled the selection of seedlings with the EPP trait by means of marker-assisted selection, making it possible to increase the efficiency of developing new EPP Japanese chestnut cultivars. However, few genotypes carry the EPP allele, and using these genotypes results in inbreeding within a narrow gene pool. Inbreeding depression, which reduces tree vigor and productivity, is common in cross-pollinated crops, including the Japanese chestnut. Breeders have attempted to identify other genes related to EPP from other genotypes.

In a previous study, there was a difference in PP among DPP cultivars of the Japanese chestnut (Tanaka and Kotobuki, 1992b). However, these were studies using samples from a plant of a genotype without any tree replications and yearly repeating. Therefore, it has not been clarified whether the variation was due to genetic or environmental factors. It has not been elucidated whether or not genetic variation exists in the PP among DPP Japanese chestnuts. The objective of the present study was to elucidate whether Japanese chestnut genetic resources other than ‘Porotan’ have genetic variation in PP or not, and if so, to estimate the magnitude of the genetic variation among years.

Materials and Methods

We used 34 genotypes, consisting of 18 cultivars, 15 selections, and 1 wild clone (Shibaguri-37) of Japanese chestnuts (Table 1), which did not possess a DNA marker (Nishio et al., 2011) linked with the recessive gene for EPP that is present in ‘Porotan’ with homozygotes.

Table 1

List of the 34 Japanese chestnut genotypes used in our experiments.

Expt. 1: Test for genetic difference using 5 cultivars grown in the NIFTS orchard over 2 years

We used five cultivars with DPP (‘Ichiemon’, ‘Ishizuchi’, ‘Shihou’, ‘Tanzawa’, and ‘Tsukuba’), and tested nuts from two trees per cultivar (genotype) that were growing in a cultivar testing and genetic resources conservation field at NIFTS (Tsukuba, Ibaraki; 36°02'56"N, 140°05'56"E) in 2011 and 2012 (Table 1). These cultivars were randomly chosen. The tree ages ranged from 9 to 12 years in 2011. The trees were grown following the standard culture techniques used in commercial chestnut production in Japan: they were pruned in the winter and then treated for pests and diseases. Nuts were harvested after the bur opened and stored at 5°C for one month, and 10 randomly selected nuts per tree were used for the evaluation of PP.

After the shells were removed, the nuts were fried in canola oil at 190°C for 2 min, following the HOP method (Shoda et al., 2006). The PP of each nut was then determined by means of hand-peeling with a paring knife; PP was scored by visual evaluation on the basis of the percentage of the surface area (peeling rate) that peeled away without scraping, on a scale graded in 10% increments, where 0 represents 0%, 5 represents <10%, 15 represents 10%–20%, 25 represents 20%–30%, 35 represents 30%–40%, 45 represents 40%–50%, 55 represents 50%–60%, 65 represents 60%–70%, 75 represents 70%–80%, 85 represents 80%–90%, and 95 represents 90%–100%. In the original HOP method, PP is quantified by the time required for peeling. However, the time is affected by nut size, which showed a large genetic difference among Japanese chestnut cultivars/selections. As we observed a large difference in the peeling rate without scraping among Japanese chestnut cultivars/selections in a preliminary experiment. We quantified PP as the peeling rate in this study.

The average peeling rate of the 10 nuts (APR) was used as continuous variable based on the central limit theorem (Snedecor and Cochran, 1972). The APR was arcsine-transformed to improve the normality of the error distribution (Snedecor and Cochran, 1972). The transformed-APR values were then tested by analysis of variance (ANOVA) with a genetic model for the phenotypic value of PP (P_1ijkl) for the kth nut of the jth tree of the ith genotype (cultivar) in the lth year:   

$$P_{1ijkl}={\mu }_1+g_{1i}+y_{1l}+t_{1ij}+(g{y}_1{)}_{il}+e_{1ijkl}$$

where subscript 1 represents the value in experiment 1, and:

μ₁ is the overall mean,

g_1i is the random genetic effect of the ith genotype,

y_1l is the random year effect of the lth year,

t_1ij is the random tree effect of the jth tree of the ith genotype,

(gy₁)_il is the genotype × year interaction between the ith genotype and the lth year, and

e_1ijkl is the residual of the kth nut of the jth tree of the ith genotype in the lth year.

We assumed that all random effects had means of zero and were independent of each other. The distribution of the residual was not significantly different from a normal distribution at P = 0.05 by Kolmogorov–Smirnov one-sample test (Campbell, 1974), indicating that ANOVA was applicable. The total variance (σ_T1²) was divided into among-genotypes (σ_g1²), among-years (σ_y1²), genotype × year interaction (σ_gy1²), among-trees within the genotype (σ_t1²), and residual (σ_e1²).

Expt. 2: Test for genetic variation using 3 cultivars grown in 5 different locations in 2011

We used three cultivars with DPP (‘Ishizuchi’, ‘Kunimi’, and ‘Tanzawa’), with a single tree per cultivar, in the orchards of four prefectural institutes and NIFTS (Ibaraki Agricultural Center, Horticultural Research Institute, Kasama, Ibaraki (36°16'44"N, 140°19'14"E); Saitama Agricultural Technology Research Center, Kuki, Saitama (36°05'27"N, 139°37'59"E); Gifu Prefectural Research Institute for Agricultural Technology in Hilly and Mountainous Areas, Nakatsugawa, Gifu (35°29'07"N, 137°28'01"E); and Kumamoto Prefectural Agricultural Research Center, Institute of Fruit Tree Science, Uki, Kumamoto (32°38'40"N, 130°42'49"E) and NIFTS). The tree ages ranged from 11 to 12 years in 2011.

In 2011, we evaluated the PP of 10 randomly selected nuts per tree, as in Expt. 1. The transformed-APR values were tested by ANOVA with a model for the phenotypic values of the PP (P_2im) of the ith genotype (cultivar) at the mth location:   

$$P_{2im}={\mu }_2+g_{2i}+L_{2m}+e_{2im}$$

where subscript 2 represents experiment 2, and:

μ₂ is the overall mean,

g_2i is the random genetic effect of the ith genotype,

L_2m is the random location effect of the mth location, and

e_2im is the residual of the ith genotype at the mth location.

We assumed that all random effects had means of zero and were independent of each other. The distribution of the residual was not significantly different from a normal distribution at P = 0.05 by Kolmogorov–Smirnov one-sample test, indicating that ANOVA was applicable. The total variance (σ_T2²) was divided into among-genotypes (σ_g2²), among-locations (σ_L2²), and residual (σ_e2²).

Expt. 3. Estimation of genetic variability among 32 cultivars/selections and 1 wild clone using trees grown at the NIFTS orchard over 3 years

We used 33 Japanese chestnut genotypes (17 cultivars, 1 wild clone, and 15 selections), with a single tree per genotype grown at NIFTS, Tsukuba, under the same cultivation conditions as in Expt. 1. Their ages in 2003 ranged from 12 to 26 years. The genotypes and trees were randomly chosen based on their availability in the field.

In 2003, 2004, and 2005, we evaluated the PP of 10 randomly selected nuts per tree, as in Expt. 1. The transformed-APR values were tested by ANOVA with a model for the phenotypic value of PP (P_3il) of the ith genotype in the lth year:   

$$P_{3il}={\mu }_3+g_{3i}+y_{3l}+e_{3il}$$

where subscript 3 represents experiment 3,

μ₃ is the overall mean,

g_3i is the random genetic effect of the ith genotype,

y_3l is the random year effect of the lth year, and

e_3il is the residual of the ith genotype in the lth year.

We assumed that all random effects had means of zero and were independent of each other. The distribution of the residual was not significantly different from a normal distribution at P = 0.05 by Kolmogorov–Smirnov one-sample test, indicating that ANOVA was applicable. The total variance (σ_T3²) was divided into among-genotypes (σ_g3²), among-years (σ_y3²), and residual (σ_e3²).

Results

Expt. 1: Test for genetic difference using 5 cultivars grown in the NIFTS orchard over 2 years

The effect of cultivar (genotype) was highly significant (P < 0.01), indicating the existence of genetic differences in PP among the five cultivars. The effect of year, the genotype × year interaction, and the among-trees effect within a genotype were not significant (Table 2).

Table 2

ANOVA results for pellicle peelability of Japanese chestnuts in Expt. 1, using 5 cultivars (genotypes) with 2 trees per genotype over 2 years^z.

The cultivars in Expt. 1 had the following transformed-APR values: ‘Tsukuba’, 23.2; ‘Ichiemon’, 28.3; ‘Ishizuchi’, 32.1; ‘Shihou’, 39.6; and ‘Tanzawa’, 50.5. Thus, transformed-APR values differed by more than 2 times among the cultivars. The variance components were estimated as σ_g1² = 108.52, σ_y1² = 4.81, σ_gy1² = −2.35, σ_t1² = 5.86, and σ_e1² = 13.19. Because σ_gy1² was negative, we assumed it to be zero (Yamada et al., 1995), so the total variance (σ_T1²) was 132.38. Thus, genetic variance (σ_g1²) accounted for 82.0% of the total variance (σ_T1²), among-year variance (σ_y1²) for only 3.6%, among-tree variance (σ_t1²) for only 4.4%, and the residual (σ_e1²) for 10.0%. These results strongly indicated the existence of genetic differences in PP among the five cultivars.

Expt. 2: Test for genetic variation using 3 cultivars grown in 5 different locations in 2011

ANOVA revealed that the effects of both genotype and location were significant (P < 0.05; Table 3). The variance components were estimated as σ_g2² = 58.10 (29.9% of total), σ_L2² = 74.67 (38.5%), and σ_e2² = 61.31 (31.6%), resulting in σ_T2² = 194.08. These results indicated the existence of genetic differences in PP among the three cultivars.

Table 3

ANOVA results for pellicle peelability of Japanese chestnuts in Expt. 2, using 1 tree of 3 cultivars (genotypes) grown at 5 different locations in 2011^z.

Expt. 3. Estimation of genetic variability among 32 cultivars/selections and 1 wild clone using trees grown in the NIFTS orchard over 3 years

ANOVA revealed a significant effect among genotypes (P < 0.01; Table 4). The variance components were estimated as σ_g3² = 53.37, σ_y3² = −1.59, and σ_e3² = 80.62. Because σ_y3² was negative, we assumed it to be zero, so σ_T3² = 134.00. Thus, σ_g3² accounted for 39.8% of the total variance and σ_e3² for 60.2%.

Table 4

ANOVA results for pellicle peelability of the Japanese chestnuts in Expt. 3, using 1 tree of 32 cultivars/selections and 1 wild clone (33 genotypes) over 3 years^z.

This experiment was carried out without tree replications for each genotype, so the g₃ effect was the sum of the genetic effect and the environmental effect among trees within a genotype. Although the variance among trees within a genotype was not estimated, it was assumed to be negligible compared with the genotype effect (σ_g3² = 53.37), because σ_t1² in Expt. 1 was only 5.86. The broad-sense heritability neglecting the among-trees effect of transformed-PP based on the average value over the 3 years was estimated as σ_g3² / (σ_g3² + [σ_e3²/3]) = 0.67.

We calculated the 95% confidence interval for the genotypic value estimated from the transformed-APR value for 3 years for each of the 33 genotypes, and then back-transformed the values to the original scale. The APR values showed a continuous distribution, ranging from 13.7% (Hiratsuka-24) to 70.0% (Shibaguri-37) (Fig. 1). The mean peeling rate for the 33 genotypes was 41.4%, equal to that of ‘Autumun koron’. The APR of Hiratsuka-16 (41.5) and Hiratsuka-17 (28.9) were between the cross parents of ‘Tanzawa’ (54.4) and ‘Ibuki’ (28.1). Conversely, the APR of Hiratsuka-15 (17.4) was lower than the cross parents of ‘Tsukuba’ (32.4) and ‘Ibuki’, and that of Hiratsuka-19 (28.9) and Hiratsuka-23 (23.9) was lower than the cross parents of ‘Tsukuba’ and ‘Ishizuchi’ (46.6).

Fig. 1

Varietal differences in the average value of the peeling rate (APR) of nuts evaluated by using the high-temperature-oil peeling method (2003–2005). Bars show the 95% confidence interval based on the APR values from 3 years for each genotypes.

Discussion

Metaxenia exists in hybrids of the Japanese chestnut with the Chinese chestnut, in which the pollen parent affects the PP (Tanaka and Kotobuki, 1992a). Japanese chestnut pollen usually scatters 10 m or more by wind. The trees used in Expts. 1 and 3 were randomly planted in the field, so if metaxenia occurred, most of the effect may be included in the residual effect. If metaxenia affected PP through interaction with an adjacent tree, the detected genetic effect included the interaction effect.

It is important that Expts. 1 and 2 showed significant genetic differences in PP among the DPP Japanese chestnut cultivars. In Expt. 1, the very large σ_g1² (108.52, 82.0% of total variance) indicated a large genetic difference among the five cultivars. In Expt. 2, σ_g2² (58.10) was smaller than σ_g1², and accounted for 29.9% of the total variance, probably owing to the choice of cultivars used in the experiment. Expt. 1 had enough replications of nut sample size, years, and trees to detect significant genetic differences among the cultivars. Expt. 2 had no tree replication within locations, so the location effect (L₂) consisted of both a true location effect and a true tree effect (environmental effect). Therefore, the genetic effect (g₂) was separated from the tree effect. The results in Expt. 2 showed a large location effect in PP, suggesting a large effect of climate on PP. In addition, the PP was possibly affected by the differences in cultivars planted around the tested tree as pollen parents.

PP showed high environmental variation in Expt. 3, and the genetic variance (σ_g3² = 53.37) in the test with 33 genotypes accounted for only 39.8% of the total variance; thus, the broad-sense heritability in this evaluation, based on 10 nuts from a single tree in 1 year was 0.40. However, based on 3 years of data, it increased to 0.67. This suggests that it is not easy to distinguish genetic differences in PP among the Japanese chestnut genotypes on the basis of a single year of data because of the large residual variance. The continuous variation in PP (Fig. 1) suggests quantitative gene effects. If PP is controlled by additive effects of polygenes or a number of oligogenes, those genes could be accumulated through crossing over generations among Japanese chestnut genotypes with relatively high PP values to produce offspring with high PP values. Evaluating PP using the HOP method revealed no effect of year or tree (Expts. 1 and 3). This provides an advantage in evaluating the variations in PP among varieties, since it permits statistical genetic analyses without the need to account for variation among years, and thereby allows researchers to combine data from multiple years (Yamada et al., 1995).

Generally, the risk of inbreeding is high in breeding of tree fruit and nut crops as a result of breeders’ efforts to accumulate additive genes that promote high fruit and nut weight and high eating quality within a narrow genetic base (Yamada et al., 2012), as done in Japanese chestnut breeding (Nishio et al., 2013; Takada et al., 2012). The suggested new genes controlling PP variation have high potential in breeding for EPP as an alternative to the major gene of ‘Porotan’, the use of which is likely to result in inbreeding. We used only one wild-clone (Shibaguri-37) in this study, but this does not represent the wild Japanese chestnut. The high PP of Shibaguri-37 suggests additional genetic variation in wild Japanese chestnut trees in Japan, coinciding with the report of Inoue et al. (2008). Therefore, it is important to explore genetic resources, including wild types. The APR of ‘Porotan’ is >90%, while that of genotypes with higher PP among the DPP genotypes is about 70%. The inheritance of the genes controlling genetic differences among DPP Japanese chestnuts is not yet known. However, if the cause is polygenes with additive effects, it will be effective to accumulate genes expressing higher PP in the breeding of new cultivars with practical PP. In addition, combining those genes and the recessive major gene of ‘Porotan’ will allow the development of new cultivars with higher PP than ‘Porotan’

Literature Cited

• Campbell, R. C. 1974. Statistics for biologists. 2^nd Ed. Cambridge Univ. Press, Cambridge, UK.
• FAOSTAT. 2016. Food and Agriculture Organization of the United Nations. Rome, Italy. <http://faostat3.fao.org/browse/Q/QC/E>.
•  Inoue,  E.,  T.  Homma,  H.  Hamada,  N.  Shiina,  M.  Ebata,  Y.  Kawamata,  M.  Kurabayashi,  Y.  Sakurai,  M.  Sasaki,  H.  Nakanishi and  H.  Hara. 2008. Peeling easiness of nut pellicle in native Japanese (Castanea crenata Sieb. et Zucc.) and European (C. sativa Mill.) chestnuts evaluated by HOP method. Hort. Res. (Japan) 7 (Suppl. 2): 464 (In Japanese).
• Kikuchi, A. 1948. Chestnut. p. 231–245. In: Pomology 1 (In Japanese). Yokendo, Tokyo.
• Miller, G., D. D. Miller and R. A. Jaynes. 1996. Chestnuts. p. 99–123. In: J. Janick and J. N. Moore (eds.). Fruit breeding. Vol. III. Nuts. Wiley, Inc., New York.
•  Nishio,  S.,  N.  Takada,  T.  Yamamoto,  S.  Terakami,  T.  Hayashi,  Y.  Sawamura and  T.  Saito. 2013. Mapping and pedigree analysis of the gene that controls the easy peel pellicle trait in Japanese chestnut (Castanea crenata Sieb. et Zucc). Tree Genet. Genomes 9: 723–730.
•  Nishio,  S.,  T.  Yamamoto,  S.  Terakami,  Y.  Sawamura,  N.  Takada and  T.  Saito. 2011. Genetic diversity of Japanese chestnut cultivars assessed by SSR markers. Breed. Sci. 61: 109–120.
• Pereira-Lorenzo, S., A. Ballester, E. Corredoira, G. Bounous, R. Botta, G. L. Beccaro, T. L. Kubisiak, M. Conedera, P. Krebs, T. Yamamoto, Y. Sawamura, N. Takada, J. Gomes-Laranjo and A. M. Ramos-Cabrer. 2012. Chestnut. p. 729–769. In: M. L. Badenes and D. H. Byrne (eds.). Fruit Breeding. Springer Science + Business Media, New York.
•  Saito,  T.,  K.  Kotobuki,  Y.  Sawamura,  K.  Abe,  O.  Terai,  M.  Shoda,  N.  Takada,  Y.  Sato,  T.  Hirabayashi,  A.  Sato,  T.  Nishibata,  Y.  Kashimura,  T.  Kozono,  H.  Fukuda,  K.  Kihara,  K.  Suzuki and  M.  Uchida. 2009. New Japanese chestnut cultivar ‘Porotan’. Bull. Natl. Inst. Fruit Tree Sci. 9: 1–9 (In Japanese with English abstract).
• Sawamura, Y. 2006. Chestnut. p. 82–85. In: Jpn. Soc. Hort. Sci. (ed.). Horticulture in Japan. Shoukadoh, Kyoto.
•  Shoda,  M.,  N.  Takada,  T.  Saito,  Y.  Sawamura and  K.  Kotobuki. 2006. A method for quickly removing pellicles from chestnuts by deep frying cooking oil. Bull. Natl. Inst. Fruit Tree Sci. 5: 21–27 (In Japanese with English abstract).
• Snedecor, G. W. and W. G. Cochran. 1972. Statistical methods. 6th ed. Iwanami-shoten, Tokyo.
•  Takada,  N.,  S.  Nishio,  M.  Yamada,  Y.  Sawamura,  A.  Sato,  T.  Hirabayashi and  T.  Saito. 2012. Inheritance of easy-peeling pellicle trait of Japanese chestnut cultivar Porotan. HortScience 47: 1–3.
•  Tanaka,  K. and  K.  Kotobuki. 1992a. Comparative ease of pellicle removal among Japanese chestnut (Castanea crenata Sieb. et Zucc.) and Chinese chestnut (C. mollissima Blume) and their hybrids. J. Japan. Soc. Hort. Sci. 60: 811–819.
•  Tanaka,  K. and  K.  Kotobuki. 1992b. Analysis of histological and chemical factors related to the ease of pellicle removal in Japanese chestnut. J. Japan. Soc. Hort. Sci. 61: 1–6 (In Japanese with English abstract).
•  Tanaka,  K.,  K.  Kotobuki and  N.  Kakiuchi. 1981. Numerization of peeling easiness and role of phenolic compounds of the pellicle in the adhesion between the pellicle and embryo in comparison of Japanese (Castanea crenata Sieb. et Zucc) and Chinese (Castanea mollissima Blume) chestnuts. J. Japan. Soc. Hort. Sci. 50: 363–371.
• Yamada, M., E. Giordani and K. Yonemori. 2012. Persimmon. p. 663–693. In: M. L. Badenes and D. H. Byrne (eds.). Fruit Breeding. Springer Science + Business Media, New York.
•  Yamada,  M.,  H.  Yamane and  Y.  Ukai. 1995. Genetic analysis of fruit ripening time in Japanese persimmon. J. Amer. Soc. Hort. Sci. 120: 886–890.