Euphytica DOI 10.1007/s10681-016-1649-7

Inbreeding depression in cassava for productive traits Juan Paulo Xavier de Freitas . Vanderlei da Silva Santos . Eder Jorge de Oliveira

Received: 6 August 2015 / Accepted: 25 January 2016 Ó Springer Science+Business Media Dordrecht 2016

Abstract Understanding inbreeding in cassava can guide breeders to explore its effects. Therefore, the aim of this study was to evaluate the effects of inbreeding depression in cassava, as well as to select transgressive individuals. Five elite cassava varieties were self-pollinated (Cascuda, BRS Formosa, Fe´cula Branca, Mani-Branca, and BRS Mulatinha), and the S1 families were evaluated in an augmented block design with six repetitions. The traits evaluated were fresh root yield (RoYi), above ground yield (AGYi), starch yield (StYi), harvest index (HI), dry matter content (DMC), and plant height (PlHe). The inbreeding depression varied widely between families; it was high in BRS Formosa, with averages of 19.38 % (RoYi), 1.68 % (AGYi), 18.18 % (HI), 0.47 % (DMC), 17.54 % (StYi) and 3.5 % (PlHe). Except for the S1 family of BRS Formosa, the additive effects (l ? a) were the most important, ranging from 69.95 % (RoYi) to 98.20 % (AGYi). In contrast, the contribution of heterozygous loci (d) was most relevant to J. P. X. de Freitas Universidade Federal do Recoˆncavo da Bahia, Campus Cruz das Almas, Cruz das Almas, BA 44380-000, Brazil e-mail: [email protected] V. da Silva Santos  E. J. de Oliveira (&) Embrapa Mandioca e Fruticultura, Rua da Embrapa, Caixa Postal 007, Cruz das Almas, BA 44380-000, Brazil e-mail: [email protected] V. da Silva Santos e-mail: [email protected]

RoYi, HI, and StYi, with averages of 30.05, 23.07, and 27.82 %, respectively, although these effects were more pronounced in S1 derived from BRS Formosa and Mani-Branca. Therefore, the exploitation of inbreeding effects in cassava can contribute to the selection of plants with better agronomic performance in order to obtain cassava inbred with high genetic and agronomic potential for use per se or as parents to produce new hybrids. Keywords Genetic effects  Breeding  Selection  Inbreeding  Manihot esculenta Crantz

Introduction Currently, cassava (Manihot esculenta Crantz) is grown in many tropical and subtropical regions of the world, and is a major source of calories in several African countries (Schons et al. 2009). Moreover, cassava has strong economic appeal, especially in agro-industrial production systems. Industrial cassava requires genotypes with high dry-matter content in the root as their main characteristic, while for human consumption, more attention should be given to cooking quality and/or starch characteristics (Oliveira et al. 2014). Despite its economic and social importance, the average yield of cassava in Brazil is 13.61 t ha-1 (FAO 2014), which is below the potential crop production of

123

Euphytica

79.03 t ha-1 under experimental conditions (Oliveira et al. 2012). Several factors may explain the low average cassava yield in Brazil, such as the use of marginal land with low fertility, low use of agricultural inputs, use of propagation material with low physiological and sanitary quality, pests and diseases, and use of unfit varieties with low productive potential for cultivation in certain climate conditions. Cassava is a crop that uses clonal propagation but involves seed production, which in nature is through outcrossing with a certain level of self-pollination (Silva et al. 2003). Therefore, breeding methods that are commonly applied for typical allogamous plants are used for cassava as well. The most widely used method of generating new cassava varieties is intraspecific hybridization between elite varieties and subsequent selection based on phenotype (mass selection) to search for superior individuals within the F1 segregating families (Fukuda et al. 2002; Ceballos et al. 2004, 2007a). In these cases, breeders take advantage of hybrid vigor, which, despite being used in several crops, is still a poorly understood phenomenon. Its main explanatory hypotheses have been associated with dominance, overdominance, and epistasis effects (Troyer 2006). In comparison to other dietary sources of great economic and food expression, such as corn, rice, and wheat, few methodological advances have been proposed or used for cassava breeding, due to its long growth cycle. Simple mass selection in self-pollinated families allows the exploration of additive, dominance, and epistatic genetic effects (Rojas et al. 2009), considering that the best individuals can be vegetatively propagated. Despite the simple mass selection allow to capitalize all genetic effects by clonal propagation in cassava, it is necessary to seek alternatives in order to improve the use of genetic effects on cassava to maximize the selection gains, as an example of inbreeding exploitation. Inbreeding is the effect of natural or artificial crossing between related individuals or individuals with some degree of kinship with each other; it is associated with increasing homozygosity and, in many cases, inbreeding depression for some traits (Keller and Waller 2002). Genetically, inbreeding depression is caused by increasing the frequency of homozygous (often deleterious) genes whose expressions are repressed in their heterozygous form. On the other hand, inbreeding has been identified as a strategy to

123

exploit the benefits of alleles in the homozygous state, a classic example is the use of inbred lines to obtain hybrid corn in the early 1900s, as well as to identify new phenotypes of interest, such as the waxy gene discovered in a S1 cassava clone (AM206-5) (Ceballos et al. 2007b). Waxy cassava starch has important applications in the food industry (Hannah 2000). Moreover, an S3-derived inbred from the cassava cultivar MCOL-1505 was used for the genome sequencing project in an attempt to minimize the number of heterozygous alleles that the assembler would have to deal with (Prochnik et al. 2012). The exploitation of inbreeding has recently been reported in cassava (Ceballos et al. 2004, 2007a; Sheela et al. 2008; Rojas et al. 2009; Kawuki et al. 2011). The results show that the estimates of inbreeding depression depend on the traits and S1 families evaluated. According to Rojas et al. (2009) and Kawuki et al. (2011), several S1 families showed that inbreeding depression may vary, on average, from 24.7 to 89.3 % for root yield and from 2.0 to 23.8 % for dry matter content in the roots. These results suggest that inbreeding depression in cassava varies with the progenitor and that it is possible to obtain genetic progress in cassava through inbreeding, once transgressive individuals can be crossed among themselves, followed by another selfpollinated generation and selection to consolidate the fixation of favorable alleles. However, this strategy has seldom been explored in cassava breeding programs, especially in germplasm originating from Latin America. The objectives of this study were to estimate inbreeding depression after one self-pollinated generation in five elite cassava varieties and assess the possibility of obtaining genetic gains by selecting transgressive individuals based on several productive traits.

Materials and methods Generation and development of S1 families Five elite cassava varieties widely cultivated in Brazil—two cultivars developed by Embrapa Cassava & Fruit (BRS Formosa and BRS Mulatinha) and three local varieties (Mani Branca, Fe´cula Branca, and Cascuda)—were used as progenitor (S0) to generate S1 families. The controlled self-pollinations were made

Euphytica

artificially in 2010, and the mature fruits were harvest from each progenitor and placed in paper bags for seed release. The S1 botanical seeds were not treated before germination. The seeds were stored for a period of 2 months to break their dormancy and then placed in plastic tubes containing 200 cm3 of substrate with topsoil, coconut fiber, and vermiculite, in the proportion of 2:1:1, to germinate. The S1 seedlings were planted in a field in 2011, in Cruz das Almas, Bahia, Brazil, following the crop management recommended for cassava (Souza and Farias 2006). At harvest time (12 months), the only adopted criterion for selecting the S1 plants was the capacity of stems to produce at least five cuttings to carry out a clonal evaluation. This minor restrictive selection should not influence the estimates of inbreeding depression, considering that over 90 % of the S1 plants were selected. Assessments of the progenitors and S1 families in the seedling stage The clonal evaluation trials (CETs) of the S1 families were planted in the experimental area of the Alianc¸a Cooperativa do Amido (Laje, Bahia, Brazil) using 20-cm cuttings. The experimental design was an augmented block design with 165 uncommon treatments and 5 common treatments (progenitor) in six blocks, using plots with five plants. The spacing was 0.9 m between rows and 0.80 m between plants. The crop management was performed according to the recommendations for cassava (Souza and Farias 2006). The numbers of S1 families evaluated from the Cascuda, BRS Formosa, Fe´cula Branca, BRS Mulatinha, and Mani Branca families were 30, 57, 13, 31, and 34, respectively. Agronomic evaluations were carried out on five plants per plot. The traits analyzed were: (1) above ground yield (AGYi, measured in t ha-1); (2) root yield (RoYi, measured in t ha-1), (3) harvest index (HI, in %, measured as the ratio between RoYi and the total biomass of the plant); (4) dry matter content of the roots (DMC, measured in % using the gravimetric method); (5) starch yield (StYi, measured in t ha-1, measured as the starch content 9 RoYi); and (6) plant height (PlHe, measured in meters). Dry matter and starch content of the roots were measured according to Kawano et al. (1987).

Data analysis Analyses of variance were performed on the averages of the five plants for each evaluated trait using the easyanova package (Arnhold 2013) for R software (R Core Development Team). The statistical model was: Yij ¼ l þ hi þ rj þ eij , where Yij = data from S1 genotype i in the repetition j, l = general mean, hi = fixed effect of the S1 genotype I, rj = random effect of the repetition j NID (0, r2rj ), and eij = experimental error NID (0, r2). The asymmetry of the data 3=2

was given by: c1 ¼ l3 =l2 , where l2 and l3 are the second and third central moments (Yau 2013). The negative and positive asymmetry indicates that the mean of the data values is less or high than the median, respectively. Therefore, the data distribution is leftskewed or right-skewed, for negative and positive asymmetry, respectively. Gardner’s method (Gardner 1965) was used to obtain the average genetic components of the families, based on the dominant-additive model, wherein the estimated expected means of the inbred lines randomly obtained from the families was l þ a ¼ 2S1  S0 , where l is the general mean and a is the estimated cumulative contribution of the homozygous loci to the family mean. The estimated contribution of the heterozygous loci (d) was obtained by d ¼ 2ðS0  S1 Þ. Estimates of inbreeding depression (ID), in percentages, were obtained using the following formula:ID ¼ ððS0  S1 Þ=S0 Þ  100, wherein S0 is the average of the original progenitor and S1 is the average of the population after one self-pollinated generation. In situations where the average of the S1 generation was above the average of S0, the ID was considered zero (0). The expected gains for the selection were obtained using the average of the top five individuals in each family compared to the averages of the S0 and S1 generations.

Results and discussion Analysis of variance Table 1 presents a summary of the analysis of variance for the performance of the progenitor and their respective S1 families for all traits. Significant differences were found among progenitor in all traits

123

Euphytica Table 1 Summary of the analysis of variance for the yield traits of cassava families in generations S0 and S1 Sources of variation

DFa

Traitb RoYi

AGYi

HI

3291.10**

DMC

StYi

PlHe

Parental

4

627.17**

5120.97**

162.83**

230.56**

6.48**

Generation

1

620.55**

66.00ns

3010.23**

27.36*

40.71**

0.09ns

Block

5

16.44ns

62.01ns

440.04ns

69.74ns

9.90ns

ns

ns

ns

Parental 9 generation Residue a

4 180

272.07** 53.38

96.92

405.23

9025.12

22,377.89

30.49 1289.31

101.30** 880.56

1.20** 0.35ns 11.75

DF degree of freedom

b

RoYi root yield, AGYi above ground yield, HI harvest index, DMC dry matter content, StYi starch yield, PlHe plant height; *, **: significant at 5 and 1%, respectively, according to F test; ns non-significant according to F test

(p \ 0.01). Moreover, significant differences between generations (S0 versus S1) were observed for RoYi, HI, and StYi (p \ 0.01), as well as DMC (p \ 0.05). Significant interactions between parent and generation were only observed for RoYi and StYi. These results show that there were important variations between the S0 and S1 generations evaluated in the cassava families, indicating to the possibility existence of inbreeding depression in these traits. In general, the average of RoYi, HI, and StYi were higher in the S0 generation, compared to S1 families. In contrast, the opposite situation was observed for DMC, in which the S1 generation showed the higher average, compared to S0 (Table 2). Estimates of inbreeding depression and genetic effects The inbreeding depression was negligible for some traits and families, such as for AGYi (Cascuda, ManiBranca, and BRS Mulatinha), DMC (Cascuda, Fe´cula Branca, and BRS Mulatinha), StYi (Cascuda and Fe´cula Branca), and PlHe (Cascuda, Mani-Branca, and BRS Mulatinha) (Table 2). In several cases, S1 families presented higher means, compared to the S0 generation. Similar results have been reported in the literature for some traits, such as the leaf area, seed weight, and harvest index in maize (Ahmad et al. 2005), in which four families showed zero inbreeding depression (negative values). It is still possible to identify such examples of productive traits in cassava, in which several families showed higher yields compared to the progenitor (Rojas et al. 2009). In contrast, the inbreeding depression of F2 okra families (facultative autogamy specie) was not significant for

123

most agronomic traits. The only significant trait was plant height in one of the crosses (Aware et al. 2014). For traits in which significant reductions in the S1 generation were observed (RoYi, HI, and StYi; see Table 2), it was found that inbreeding depression was quite variable in the families from different varieties. For RoYi, inbreeding depression was not significant for the Cascuda and Fe´cula Branca families, but ranged from 15.90 to 55.20 % in the families derived from BRS Mulatinha and BRS Formosa, respectively. For HI, inbreeding depression was not significant in the Cascuda family, but varied from 15.33 to 25.92 % in the Mani-Branca and BRS Formosa families, respectively. For StYi, the inbreeding depression was only high in the Mani-Branca (20.91 %) and BRS Formosa (55.57 %) families. Although there are experimental difficulties in making comparisons between different studies, it turns out that the estimates of inbreeding depression in this study were discordant in magnitude (but not direction) with respect to the available estimates in the literature under different soil and climatic conditions. For instance, according to Rojas et al. (2009), the average inbreeding depression for RoYi was 64.0 %, ranging from 50.6 % (AM337) to 77.8% (AM320), while it ranged from 16.4 to 56.5 % for AGYi (average of 37.9 %), HI ranged from 16.6 to 43.0 % (average of 26.5 % and DMC ranged from 0.3 to 8.7 % (average of 5.3 %). In the latest study, Kawuki et al. (2011) found inbreeding depression of 61.21 % for RoYi, ranging from 24.7 % (TDMC30572) to 89.3 % (Bamunanika); 24.7 % for AGYi, ranging from 11.0 % (TDMC30572) to 58.0 % (MH95/0469); 33.83 % for HI, ranging from 4.7 % (95/SE-00036) to 77.0 % (Bamunanika); and 13.2 % for DMC, with a

Euphytica Table 2 Means of the productive traits evaluated in elite progenitors (S0) and families derived from self-pollinated (S1) for inbreeding depression (ID) and estimates of the contribution of the homozygous (l ? a) and heterozygous loci (d) Families from

Root yield (RoYi—t ha-1) S0

S1

Minimum

Maximum

ID

l?a

d

Cascuda

12.95(a)a

12.72(a)

3.99

24.83

1.78

98.19

1.81

BRS Formosa Fe´cula Branca

35.60(a)

15.95(b)

3.13

29.44

55.20

0.00

100.00

10.26(a)

9.92(a)

1.39

50.00

3.31

96.57

3.43

Mani-Branca

23.54(a)

18.67(b)

2.64

47.74

20.69

73.92

26.08

BRS Mulatinha

16.29(a)

13.70(b)

1.39

40.00

15.90

81.09

18.91

Average

19.73(a)

14.19(b)

19.38

69.95

30.05

Families from

Above ground yield (AGYi—t ha-1) S0

Cascuda

S1

Minimum

Maximum

ID

l?a

d

4.09(a)

5.45(a)

2.31

11.39

0.00

100.00

0.00

BRS Formosa Fe´cula Branca

14.32(a)

13.29(a)

3.89

31.94

7.19

92.25

7.75

3.25(a)

3.21(a)

1.11

10.76

1.23

98.75

1.25

Mani-Branca

20.43(b)

25.42(a)

9.72

56.53

0.00

100.00

0.00

BRS Mulatinha

10.82(a)

14.57(a)

3.24

48.89

0.00

100.00

0.00

Average

10.58(a)

12.39(a)

1.68

98.20

1.80

Families from

Harvest index (IC—%) S0

S1

Minimum

Maximum

ID

l?a

d

Cascuda

75.39(a)

70.22(a)

47.52

83.33

6.86

92.64

7.36

BRS Formosa Fe´cula Branca

71.65(a)

53.08(b)

18.03

72.88

25.92

65.02

34.98

76.23(a)

62.57(b)

35.42

95.36

17.92

78.17

21.83

Mani-Branca

53.94(a)

45.67(b)

10.16

61.50

15.33

81.89

18.11

BRS Mulatinha

61.58(a)

46.28(b)

8.16

77.42

24.85

66.94

33.06

Average

67.76(a)

55.56(b)

18.18

76.93

23.07

Families from

Dry matter content (DMC—%) S0

S1

Minimum

Maximum

ID

l?a

d 0.00

Cascuda

29.26(b)

32.14(a)

24.10

36.48

0.00

100.00

BRS Formosa Fe´cula Branca

33.37(a)

33.14(a)

26.47

37.34

0.69

99.31

0.69

29.30(b)

31.31(a)

27.46

37.98

0.00

100.00

0.00

Mani-Branca BRS Mulatinha

32.35(a) 33.04(a)

31.81(a) 34.74(a)

20.67 22.86

36.85 46.94

1.67 0.00

98.30 100.00

1.70 0.00

Average

31.47(b)

32.63(a)

0.47

99.52

0.48

Families from

Starch yield (StYi—t ha-1) S0

Cascuda

S1

Minimum

Maximum

ID

l?a

d

3.21(a)

3.57(a)

0.78

6.50

0.00

100.00

0.00

BRS Formosa Fe´cula Branca

10.24(a)

4.55(b)

0.93

8.95

55.57

0.00

100.00

2.54(a)

2.68(a)

0.39

16.53

0.00

100.00

0.00

Mani-Branca

6.60(a)

5.22(b)

0.72

14.52

20.91

73.56

26.44

0.25

11.88

BRS Mulatinha

4.63(a)

4.11(a)

Average

5.44(a)

4.03(b)

11.23

87.35

12.65

17.54

72.18

27.82

123

Euphytica Table 2 continued Families from

Plant height (PlHe—m) S0

S1

Minimum

Maximum

ID

l?a

Cascuda

0.66(a)

0.81(a)

0.50

1.40

0.00

100.00

0.00

BRS Formosa Fe´cula Branca

1.20(a)

1.09(a)

0.62

1.58

9.17

89.91

10.09

0.84(a)

0.77(a)

0.42

1.50

8.33

90.91

9.09

Mani-Branca

1.50(b)

1.77(a)

1.30

2.43

0.00

100.00

0.00

BRS Mulatinha

1.13(a)

1.21(a)

0.72

1.83

0.00

100.00

0.00

Average

1.06(a)

1.13(a)

3.50

96.16

3.84

a

d

Means with the same letter in a row are not significantly different from each other (p \ 0.05)

range from 2.0 % (95/SE-00036) to 23.8 % (I92/ 00067). Therefore, the variation found in inbreeding depression is related to the nature of the trait and the degree of genetic heterozygosity in the S0 genotypes. However, in general, the inbreeding depression was high for productive traits, like RoYi and StYi, as observed in the present work. In general, high inbreeding depression is expected in cassava, considering the reproductive system of the species, which is predominantly carried out by crosspollinated. However, the average inbreeding depression in the cassava families for the productive traits was lower than that observed in typical outcrossing species, such as maize (Cleso et al. 2002; Ahmad et al. 2005). This can be explained by the existence of natural selfing (Silva et al. 2003), which eliminates deleterious alleles throughout the different generations of reproduction, as well as the lack of germination of seeds or seedling mortality with high genetic load, which lead to underestimated estimates of inbreeding depression (Bison et al. 2004). Furthermore, according to Rojas et al. (2009), S1 individuals represent an unbiased sample of partially inbred populations, but there is usually a small underestimation of inbreeding depression. According to Ceballos et al. (2007a), two key events occur when an elite clone is self-pollinated: (1) allelic combinations (genotypes) of individuals are broken down, resulting in phenotypic change for various agronomic traits; and (2) the homozygosity of the loci increase to facilitate the elimination of undesirable alleles that have hitherto been concealed by the genotype’s heterozygous state. Therefore, the use of self-fertilization in cassava breeding will reduce the genetic load and thus improving the predictability

123

of the crossings from the use of contrasting genotypes with high levels of homozygosity. From the perspective of genetic effects, estimates of inbreeding depression in cassava were lower for DMC, AGYi, and PlHe in relation to productive attributes, possibly because the genetic effects of dominance are more important for traits related to root yield in selfpollinated varieties, similar to what is observed in other species. Similar finds were reported for corn yield, whose average inbreeding depression was 43.1 and 49.1 % in the works of Ahmad et al. (2005) and Cleso et al. (2002), respectively. Cleso et al. (2002) also observed that inbreeding depression was higher in populations with a broad genetic base that was never exposed to inbreeding. On average, estimates of the contributions of the heterozygous loci (d) were more important for RoYi (average of 30.05 %, ranging from 1.81 to 100.00 %), HI (average of 23.07 %, ranging from 7.36 to 34.98 %), and StYi traits (average of 27.82 %, ranging from 0.00 to 100.00 %). However, the d effects, and consequently heterosis, are most important for these traits in families derived from the BRS Formosa and Mani-Branca varieties. Differences in d estimates are commonly associated with the genetic nature of the trait, in which major depression inbreeding is found in characteristics controlled by a greater number of dominant alleles, as well as in germplasm with high heterozygosity and populations with high genetic load, like unimproved germplasm. These hypotheses may explain the fact that inbreeding depression and estimates of the effects of the heterozygous loci were low in the present work, considering that the self-pollinated cassava varieties were improved varieties that had undergone several cycles of selection, which could resulted in reduced genetic load.

Euphytica

Bison et al. (2004) found significant interactions between clones x generations in eucalyptus (p \ 0.01), demonstrating that inbreeding depression varies among clones. For instance, inbreeding depression for circumference at breast height (CBH) was pronounced for families derived from three clones and zero for another clone. In contrast, inbreeding depression for wood basic density (WBD) was stronger in different families than those observed for the CBH trait. Except for the family derived from the BRS Formosa variety, the additive effects (l ? a) were most important to all traits and families, with variation ranging from 69.95 % (RoYi) to 98.20 % (AGYi). In corn, Souza Sobrinho et al. (2002) observed that estimates of l ? a and d represented 29.4 and 70.6 %, respectively, of the average yield in F1 hybrids. In eucalyptus, the contributions of homozygous loci were considerably more expressive than those of the heterozygous loci, especially for stem diameter at breast height (Bison et al. 2004). The few reports on the prevalence of genetic effects in cassava indicate that the results of this study are similar with respect to the predominance of additive effects for DMC and HI, but discordant for RoYi (Calle et al. 2005). According to Calle et al. (2005), the sum of squares associated with the effects of general combining ability (GCA) and specific combining ability (SCA) is indicative of the importance of additive and non-additive effects in trait expression. Therefore, when evaluating a diallel among 10 elite cassava clones in savannah acid soils in eastern Colombia, these authors found that the effects of SCA accounted for 53 % of the variation related to fresh root yield, therefore having predominantly nonadditive effects. On the other hand, the SCA effects were less significant for harvest index (38 %), height of the first branching (33 %), dry matter content, plant type, and the severity of super elongation (\20 %), showing that additive genetic effects were more important. For the traits in which the l ? a effects are more important, there is a high potential of success in selecting best genotypes within segregating families, whose results may reflect the development of inbred lines with high genetic potential. Even for traits with low contribution of the homozygous loci (RoYi, HI, and StYi), there is the possibility of selecting plants for use in new selfpollinated cycles. For RoYi, the high estimates of l ? a were obtained for the Cascuda (98.19 %) and Fe´cula Branca (96.57 %) families, but due to the low

average of the trait, these families were not the best ones for advances in obtaining inbred lines. The ManiBranca and BRS Mulatinha families’ high RoYi values make them the most promising for obtaining inbred lines, with l ? a estimates of 73.92 and 81.09 %, respectively. For HI, high l ? a estimates were obtained from the Cascuda family (92.64 %), in which the HI values were above 70 %. In absolute terms, the plants derived from Fe´cula Branca family were also an excellent option for obtaining inbred lines, with high HI ([62 % in S1). For StYi, the Cascuda and Fe´cula Branca families presented the highest l ? a estimates (100 %), but with low starch yield in the S1 generation (3.57 and 2.68 t ha-1, respectively). In this case, the plants derived from the Mani-Branca and BRS Mulatinha families are more suitable for advances in the development of inbred lines, due to their high values for starch yield (5.22 and 4.11 t ha-1, respectively), with l ? a estimates [70 %. The average of the potential inbred lines after successive generations of self-pollinations will be equal to l ? a, once the effects of the homozygous loci are canceled on the segregant loci in early generations of selfing, such that the average of the inbred lines will only depend on the fixed loci set in the progenitor. In general, the l ? a estimates were high for the majority of the traits and cassava families analyzed, probably because of high frequency of favorable alleles. This fact certainly enables the continuity of selfing and selection of plants with better agronomic performance to obtain inbred cassava lines with high genetic and agronomic potential. On average, the asymmetry of the S1 cassava individuals only suggests a relatively symmetrical distribution of their frequencies for DMC. However, these values varied widely depending on the family (range -2.26 to 0.85). Furthermore, positive asymmetry (right-skewed) values for RoYi, AGYi, StYi, and PlHe indicate asymmetric distributions with long tails to the right, consolidating the presence of high numbers of S1 individuals with low agronomic performance. Nevertheless, negative asymmetry (left-skewed) was found in some families for RoYi and StYi. In contrast, the asymmetry of the individuals regarding HI was mostly negative or with small deviations from symmetry in some cases, indicating the presence of a high number of S1 plants with improved performance compared to the parent (Table 3). Similarly, Rojas

123

Euphytica Table 3 Number of S1 plants with superior (:) or lower (;) performance compared with the parental average (S0), mean values, and variation in the asymmetric distributions of cassava families for six agronomic traits Family from

RoYi

AGYi

HI

DMC

StYi

PlHe

:

;

:

;

:

;

:

;

:

;

:

;

Cascuda

8 |8|a

22 |14|

14 |10|

16 |12|

5 |5|

25 |16|

20 |9|

10 |9|

8 |8|

22 |13|

22 |14|

8 |8|

BRS Formosa Fe´cula Branca

0 |0|

57 |47|

16 |15|

41 |26|

1 |1|

56 |39|

29 |25|

28 |22|

0 |0|

57 |46|

13 |13|

74 |29|

4 |2|

9 |4|

3 |2|

10 |4|

4 |4|

9 |3|

9 |2|

4 |4|

4 |2|

9 |3|

3 |2|

10 |3|

Mani-Branca

1 |1|

33 |16|

20 |12|

14 |14|

4 |4|

30 |22|

20 |15|

14 |11|

2 |2|

32 |16|

31 |17|

3 |3|

BRS Mulatinha

12 |11|

19 |11|

17 |8|

14 |14|

6 |6|

25 |15|

26 |17|

5 |5|

15 |9|

16 |12|

20 |9|

11 |11|

Asymmetry

0.99

Min–max

-0.27

1.80

-1.31

0.15

-2.26

0.85

-0.26

2.72

0.20

a

1.37 2.29

0.70

-0.53

-0.18

1.37

0.67 1.48

The numbers of S1 plants within the modules differ significantly from the S0 generation, based on t tests

et al. (2009) and Kawuki et al. (2011) also observed positive symmetry deviations for RoYi, AGYi and PlHe, as well as negative deviations for DMC and HI. The numbers of individuals with significant performance above or below the average of the progenitor (S0) generation are presented in Table 3. Only the BRS Formosa family did not present S1 plants superior to its progenitor line for RoYi and StYi. On the other hand, the numbers of superior clones were highly variable among the families for the different traits. Nevertheless, these results indicate the possibility of selecting the best individuals within each cassava family to obtain genetic gains. Rojas et al. (2009) also noted the possibility of selecting S1 genotypes with better agronomic performance for root yield, compared with the progenitor plants, in two of eight families (AM336 and AM337) analyzed. Perspectives of cassava breeding There are few reports about estimates of inbreeding depression in cassava, and most of the works involve populations/varieties not grown in Brazil (Rojas et al. 2009; Kawuki et al. 2011). On the other hand, along with predicting the productive potential of inbred S1 families and determining the genetic variation present, the l ? a estimates guide breeders in making decisions on the most promising segregating populations to explore their additive and dominant effects. The improved performance of inbred cassava can have a positive impact in continuing the selfing and the subsequent extraction of inbred lines to be used per se or as progenitor lines for generating contrasting hybrids.

123

Moreover, by presenting high productive stability and superior agronomic traits, few cassava varieties occupy most of the area planted with this crop in Brazil. Therefore, it is important to know the variation in inbreeding depression when selfing these varieties to verify the feasibility of selecting superior plants, given the presence of a high number of favorable alleles for root yield and disease resistance. Another major advantage of the exploitation of selfing in cassava is the possibility of obtaining homozygous lines. These inbred lines can facilitate the conservation of M. esculenta germplasm, which is predominantly done in vitro or ex situ in the field. Both strategies have some issues, such as risks of loss due to environmental issues (in the field) and high maintenance costs (in vitro). Therefore, storage could be done by seeds, with full conservation of the genotypes, like autogamous species. Cassava propagation using seeds could also accelerate the cleaning of stems against viruses and other diseases in a practical, quick, and cost-effective way, and also allow an easy exchange of cassava germplasm between different countries and institutions. Another important aspect of the use of inbreeding in cassava is the discovery of natural or induced mutations that can bring significant benefits to the starch industry, as happened with waxy starch (Ceballos et al. 2007b), or even competitive agronomic advantages, such as resistance to pests and diseases. Traits that can bring competitive advantages in the starch modification process for many different industrial applications, as identified by starch grain analysis (shape, size, particle size, behavior against different chemical and

Euphytica

physical modification) or even through pasting properties, can result in significant economic gains. Acknowledgments The authors thank the Fapesb, CAPES and CNPq for the financial assistance and scholarship support.

References Ahmad S, Khan MS, Swati MS, Shah GS, Khalil IH (2005) A study on heterosis and inbreeding depression in sunflower (Helianthus annuus L). Songklanakarin J Sci Technol 27:1–8 Arnhold E (2013) Package in the R environment for analysis of variance and complementary analyses. Braz J Vet Res Anim Sci 50:488–492 Aware SA, Deshmukh DT, Thakare SV, Zambre SM (2014) Heteosis and inbreeding depression studies in okra (Abelmoschus esculentus (L) Moench). Int J Curr Microbiol Appl Sci 3:733–747 Bison O, Aguiar AM, Rezende GDSP, Ramalho MAP (2004) Inbreeding depression in Eucalyptus clones. Crop Breed Appl Biotechnol 4:459–464 Calle F, Perez JC, Gaita´n W, Morante N, Ceballos H, Llano G, Alvarez E (2005) Diallel inheritance of relevant traits in cassava (Manihot esculenta Crantz) adapted to acid-soil savannas. Euphytica 144:177–186 Ceballos H, Iglesias CA, Pe´rez JC, Dixon AGO (2004) Cassava breeding: opportunities and challenges. Plant Mol Biol 56:503–516 Ceballos H, Pe´rez V, Juan C, Iglesias F, Carlos A, Fregene M, Calle F, Jaramillo OG, Morante N, Lo´pez J (2007) The use of doubled-haploids in cassava breeding. In: Howeler RH (ed) Cassava research and development in Asia: exploring new opportunities for an ancient crop. In: Proceedings of the seventh regional workshop held in Bangkok, Oct 28– Nov 1 2002. Centro Internacional de Agricultura Tropical (CIAT), Cassava Office for Asia, Bangkok, pp 150–160 Ceballos H, Sa´chez T, Morante N, Fregente M, Dufour D, Smith AM, Denyer K, Pe´rez JC, Calle F, Mestres C (2007b) Discovery of amylose-free starch mutant in cassava (Manihot esculenta Crantz). J Agric Food Chem 55:7469– 7476 Cleso APP, Santos MX, Cruz CD, Parentoni SN, Oliveira E, ´ E, Carvalho HWL, Guimara˜es HG, Gama EEG, Silva A Ju´nior PAV (2002) Inbreeding depression of 28 maize elite open pollinated varieties. Genet Mol Biol 25:12–14 FAO (2014) FAOSTAT database. FAO, Rome. http://faostat. fao.org/. Accessed 26 Jan 2015 Fukuda WMG, Silva SO, Iglesias C (2002) Cassava breeding. Crop Breed Appl Biotechnol 2:617–638

Gardner CO (1965) Teoria de gene´tica estadistica aplicable as las medias de variedades, sus cruces y poblaciones afines. Fitotecnica Latinoamericana 2:11–22 Hannah LC (2000) Starch biosynthesis and genetic potential. In: Murphy CF, Peterson DM (eds) Designing crops for added value. American Society of Agronomy, Madison, pp 181– 199 Kawano K, Fukuda WMG, Cenpukdee U (1987) Genetic and environmental effects on dry matter content of cassava root. Crop Sci 27:69–74 Kawuki RS, Nuwamanya E, Labuschagne MT, Herselman L, Ferguson M (2011) Segregation of selected agronomic traits in six S1 cassava families. J Plant Breed Crop Sci 3:154–160 Keller LF, Waller DM (2002) Inbreeding effects in wild populations. Trends Ecol Evol 17:230–241 Oliveira EJ, Resende MDV, Santos VS, Ferreira CF, Oliveira GAF, Silva MS, Oliveira LA, Aguilar-Vildoso CI (2012) Genome-wide selection in cassava. Euphytica 187:263– 276 Oliveira EJ, Santana FA, Oliveira LA, Santos VS (2014) Genetic parameters and prediction of genotypic values for root quality traits in cassava using REML/BLUP. Genet Mol Res 13:6683–6700 Prochnik S, Marri PR, Desany B, Rabinowicz PD, Kodira C, Mohiuddin M, Rodriguez F, Fauquet C, Tohme J, Harkins T, Rokhsar DS, Rounsley S (2012) The cassava genome: current progress, future directions. Trop Plant Biol 5:88–94 Rojas MC, Pe´rez JC, Ceballos H, Baena D, Morante N, Calle F (2009) Analysis of inbreeding depression in eight S1 cassava families. Crop Sci 49:543–548 Schons A, Streck NA, Storck L, Buriol GA, Junior Zanon A, Pinheiro DG, Kraulich B (2009) Arranjos de plantas de mandioca e milho em cultivo solteiro e consorciado: crescimento, desenvolvimento e produtividade. Bragantia 68:155–167 Sheela MN, Radhika VS, John KS, Abraham K (2008) Variation in crude protein, dry matter and starch in inbred and backcross lines of cassava. J Root Crops 34:115–119 Silva RM, Bandel G, Martins PS (2003) Mating system in an experimental garden composed of cassava (Manihot esculenta Crantz) ethnovarieties. Euphytica 134:127–135 Souza LS, Farias RN (2006) Aspectos socioeconoˆmicos e agronoˆmicos da mandioca. Embrapa Mandioca e Fruticultura Tropical. Cruz das Almas Souza Sobrinho F, Ramalho MAP, Souza JC (2002) Alternatives for obtaining double cross maize hybrids. Rev Bras Milho Sorgo 1:70–76 Troyer AF (2006) Adaptedness and heterosis in corn and mule hybrids. Crop Sci 46:529–543 Yau C (2013) R tutorial with Bayesian statistics using OpenBUGS. Amazon Digital Services, Inc. 554

123