Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-14T10:00:36.033Z Has data issue: false hasContentIssue false
  • English
  • Français

Genetic variation and correlation among yield and quality traits in cocksfoot (Dactylis glomerata L.)

Published online by Cambridge University Press:  01 August 2007

A. JAFARI and
H. NASERI
A. JAFARI*
Affiliation:
Research Institute of Forests and Rangelands, Tehran, Iran
H. NASERI
Affiliation:
Islamic Azad University, Brojerd, Iran
*
*To whom all correspondence should be addressed. Email: aajafari@rifr-ac.ir
Rights & Permissions [Opens in a new window]

Summary

The objective of the present research was to study the genetic variability for total dry matter (DM) yield, tiller number, heading date and three quality traits, namely content of digestible dry matter (DDM), water-soluble carbohydrate (WSC) and crude protein (CP), in cocksfoot (Dactylis glomerata L.). Twenty-five parents were randomly chosen from a genetically broad-based population, and their respective half-sib (HS) families were generated. Clonally-propagated parents and their HS family seeds were grown as individual plants using a randomized complete block design with two replications in Alborz Research Center, Karaj, Iran, during 2002–04. The results of combined analyses over 2 years showed significant variances between clonal parents for all traits except CP. In the HS generation, between-family variances were only significant for tiller number, heading date and WSC. Clone×year (S2GY) and family×year (S2FY) interactions were significant for all traits except for WSC in HS families. The estimates of broad-sense heritability (h2b) were moderate to high for all traits (h2b=0·37–0·69), except CP. Narrow-sense heritability (h2n) estimates from analyses of progenies and from regression of HS progenies on parents (h2op) were moderate, relatively the same values as h2b for heading date, tiller number and WSC, which suggested that additive genetic variance was the main component controlling these traits. For DM yield and DDM, h2n and h2op estimates were low, whereas h2b estimates were moderate, which suggested that both additive and non-additive gene effects played an important role in the genetic regulation of these traits. Genetic correlations among CP with both WSC and DDM were generally negative, whereas WSC was positively correlated with DDM and tiller number. The genetic correlation among DM yield with DDM was weak and inconsistent and, in general, negative. DM yield had negative and positive correlation with heading date and tiller number, respectively. It was concluded that there was significant variation and moderate heritability for most traits in the cocksfoot populations evaluated to improve yield and quality traits. Selection for high WSC is a means to improve quality in general. The data also indicate that response to combined selection for both DDM and DM yield should be possible. Selection for DDM alone could result in reduction in yield.

Information

Type
Crops and Soils
Information
The Journal of Agricultural Science , Volume 145 , Issue 6 , December 2007, pp. 599 - 610
Copyright
Copyright © Cambridge University Press 2007

INTRODUCTION

The botanical composition of rangelands is variable in Iran. Cocksfoot (Dactylis glomerata L.) is one of the main perennial grasses that naturally grow in temperate pasture and rangelands in northern and western Iran. It is used for grazing and hay production. Cocksfoot grows at altitudes of 500–2900 m (Rechinger Reference Rechinger1970) having more than 300 mm annual participation (Niaky Reference Niaky1995). The improvement of total annual yield, persistency, disease resistance and extended grazing season are important objectives in most herbage breeding programmes. However, data from animal nutrition studies show the need to focus more attention on nutritive value in selection programmes. Wheeler & Corbett (Reference Wheeler and Corbett1989) and Smith et al. (Reference Smith, Reed and Foot1997) ranked forage traits in terms of their nutritional value for live weight gain and dairy production, respectively. Improved digestibility and increased water-soluble carbohydrates (WSC) content were the two most important criteria on each of the two lists.

There are much published data which show that digestibility is a major factor affecting intake (Cooper Reference Cooper, Butler and Bailey1973) and animal performance (Connolly et al. Reference Connolly, do Valle Ribeiro, Crowley and Gilsenan1977). Carlier (Reference Carlier, Reheul and Ghesquière1994) concluded that WSC are completely digestible and have an important role in ruminant animal nutrition, as they are a primary source of the readily available dietary energy necessary for efficient microbial fermentation in the rumen. Beever & Reynolds (Reference Beever, Reynolds, Mannetje and Frame1994) have shown that an adequate supply of soluble sugars is essential for good fermentation and protein utilization in the rumen, leading to improved feed efficiency and animal performance. When considered as a separate characteristic, crude protein (CP) content was ranked as moderate or low priority in terms of quality objectives (Wheeler & Corbett Reference Wheeler and Corbett1989; Smith et al. Reference Smith, Reed and Foot1997). However, with regard to the important interaction between WSC and CP in the efficiency of protein metabolism as discussed above, it is clear that the combined evaluation of both characteristics is desirable in relation to selection for improved nutritional value in herbage.

From a plant breeding perspective, the possibility of improving forage quality by selection is a very attractive objective. The extent to which this is possible depends on the type of genetic control of the component characters and their interrelationship with other factors such as yield, disease resistance and persistence. To improve grass varieties for yield and quality traits, knowledge of genetic parameters is important for choosing an efficient selection strategy. Plant breeders have used the estimation of genetic variance and its additive components for selection proposes in perennial species because most breeding methods available to the forage breeder make little use of non-additive genetic variation (Nguyen & Sleper Reference Nguyen and Sleper1983).

The published data for cocksfoot suggest that genetic variation is present for digestibility (Cooper Reference Cooper1962; Christie & Mowat Reference Christie and Mowat1968; Frandsen Reference Frandsen1986) and its improvement by selection should be possible. However, there is some evidence that the inheritance of digestibility in forage grasses is not purely additive but that dominance is present (Marum et al. Reference Marum, Hovin, Marten and Shenk1979; Beerepoot et al. Reference Beerepoot, Bouter, Dijkstra, Reheul and Ghesquière1994). By comparison with other traits, there is little information on genetic control of WSC in cocksfoot. Cooper (Reference Cooper1962) and Grusea & Oprea (Reference Grusea, Oprea, Reheul and Ghesquière1994) reported that genetic effects of WSC were additive. However, contrasting results were reported by Humphreys (Reference Humphreys1989a, Reference Humphreysb) in perennial ryegrass, who found that WSC behaved as a complex polygenic trait that was controlled by mainly non-additive gene effects. For CP in cocksfoot, Cooper (Reference Cooper1962) and Shenk & Westerhaus (Reference Shenk and Westerhaus1982) found relatively high estimates of heritabilities (h 2n=0·55) and (h 2n=0·39–0·64). For dry matter (DM) yield, both additive (Cooper Reference Cooper1962; Frandsen Reference Frandsen1986; Annicchiarico & Romani Reference Annicchiarico and Romani2005) and dominant variance (Casler Reference Casler1998; Jafari Reference Jafari1998) have previously been reported.

Knowledge of correlation between traits of interest is useful in designing an effective breeding programme for a crop. Despite the increased emphasis on quality characteristics, total DM yield and seasonal yield distribution are of primary interest in herbage breeding. Consequently, the study of the potential for improvement in quality characteristics should be combined with the analysis of yield and its interrelationship with quality components. The extent to which various quality characters are correlated in forage grasses has been studied by a number of investigators (Frandsen Reference Frandsen1986; Humphreys Reference Humphreys1989c; Marum et al. Reference Marum, Rognli, Aastveit, Aastveit, Reheul and Ghesquière1994; Jafari et al. Reference Jafari, Connolly and Walsh2003a). In general, the correlation between CP and DM yield was negative and negative relationship between yield and digestible dry matter (DDM) was also frequently found. WSC was positively correlated with DDM, whereas the relationship between WSC and DM yield was inconsistent (Brown & Blaser Reference Brown and Blaser1970; Jafari et al. Reference Jafari, Connolly and Walsh2003a; Sanada et al. Reference Sanada, Takai and Yamada2004). Cocksfoot has an important role in grassland productivity and any improvement in its herbage yield and quality would be very beneficial in terms of animal productivity. The present research project was conducted because relatively little breeding work has been done on this species and the information on its breeding behaviour, especially under the climatic conditions of Iran, is scanty. The objectives of the study were: (1) to estimate genetic variability and heritability for DM yield and quality traits, (2) to examine relationships among yield and quality traits and (3) to predict genetic gain from one cycle of selection.

MATERIALS AND METHODS

The cocksfoot genotypes used in the present study were derived from domestic accessions that were collected from temperate pasture and rangelands in northern and western Iran as follows: Uremia, Ardabil, Karaj, Sari and Gorgan. Five accessions were collected with a range of ear emergence of 7 days. Five seeds were taken randomly from each accession and sown in compost. The resulting 25 seedlings were vegetatively propagated to give six clones of each. The polycross consisted of 25 genotypes arranged randomly in six clonal replicate blocks and was established in the polycross nursery at the Research Institute of Forests and Rangelands, Karaj, Iran, in September 2001. The ear emergence date was recorded twice a week in May 2002. At harvest, seed from the clonal replicates of each genotype was bulked. Seed of HS families were sown for progeny test. From each HS family, eight seedlings were established in compost.

At the same time, eight clonal propagations were made from tillers of each parental genotype, planted under the same conditions as the HS seed progenies. The vegetative propagules of the parents, together with HS progeny seedlings, were transplanted to the field in October 2003. Two experiments were established using randomized complete-block designs with two replications for both parents and progenies. Spaced plants were established in rows 500 mm apart, with 400 mm spacing within rows. Fertilizer application rates were 50 or 100 kg nitrogen (N) and phosphorus (P)/h at sowing. Application of nitrogen was continued at 50 kg/h for the second and third years. The field was irrigated once a week during summer. Due to the dry conditions after transplanting, some seedlings died; therefore, only five out of eight plants per plot were evaluated. No measurements were taken in the establishment year. In 2004 and 2005, the plants were harvested three times. Before the first harvest, ear emergence date was measured as the number of days from 21 March to the stage at which three flowering shoots were visible. At harvest, fertile tillers were counted on spaced plants in the first and second cuts of both years. At the first cut of spaced plants, fertile tillers were assessed visually: the number of stems per plant was classified into five groups as 1 (1–10 stems), 2 (11–20 stems), 3 (21–30stems), 4 (31–40 stems) and 5 (more than 40 stems per spaced plant). The distribution of data for tiller number was non-normal, especially in the second cut. To normalize the data, they were transformed based on normal score with a mean of 2·5.

In each harvest, plants were cut, weighed, dried at 70°C for 24 h, and reweighed to determine DM yield, then ground with a Retsch Impeller-type mill (1 mm screen). DM yields were measured for three cuts per year. Quality traits (DDM, WSC and CP) were estimated in the first and the second cuts for each year using near infrared spectroscopy (NIR). Details of the methodology and calibrations of NIR are given by Jafari et al. (Reference Jafari, Connolly and Walsh2003b).

Statistical analysis

Data were collected for DM yield, morphological and quality traits. For each year, data were analysed for total annual DM yield (three harvests) and average annual quality value (two harvests). Data were also subjected to a combined analysis of variance across years using a split-plot-in-time design with years as sub-plots (Steel & Torrie Reference Steel and Torrie1980). Expected mean squares (EMS) were based on a random effects model for blocks, years, parents and HS families. Variance and covariance components were used to estimate heritabilities and genetic correlations. Broad-sense (h 2b) and narrow-sense (h 2n and h 2op) heritabilities were estimated from analyses of parents, HS families, and regression of offspring on one parent, respectively (Nguyen & Sleper Reference Nguyen and Sleper1983; Falconer & Mackay Reference Falconer and Mackay1996).

For individual years,

h_{\rm n}^{\setnum{2}} \equals {{\sigma _{\rm G}^{\setnum{2}} } \over {\sigma _{\rm G}^{\setnum{2}} \plus {{\sigma _{\rm e}^{\setnum{2}} } \over r}}}\comma \qquad h_{\rm n}^{\setnum{2}} \equals {{\sigma _{\rm F}^{\setnum{2}} } \over {\sigma _{\rm F}^{\setnum{2}} \plus {{\sigma _{\rm w}^{\setnum{2}} } \over r}}}\comma \qquad h_{{\rm OP}}^{\setnum{2}} \equals 2b \equals {{V_{\rm A} } \over {V_{\rm P}}}.

Combined across 2 years,

\openup3\eqalign{\hskip 38pt h_{\rm b}^{\setnum{2}} \equals {{S_{\rm G}^{\hskip 1\setnum{2}} } \over {S_{\rm G}^{\hskip 1\setnum{2}} \plus {{S_{{\rm GR}}^{\hskip 1\setnum{2}} } \over y} \plus {{S_{{\rm GY}}^{\hskip 1\setnum{2}} } \over r} \plus {{S_{\rm e}^{\hskip 1\setnum{2}} } \over {ry}}}}\comma }\cr \tab\qquad h_{\rm n}^{\setnum{2}} \equals {{S_{\rm F}^{\hskip 1\setnum{2}} } \over {S_{\rm F}^{\hskip 1\setnum{2}} \plus {{S_{{\rm FR}}^{\hskip 1\setnum{2}} } \over y} \plus {{S_{{\rm FY}}^{\hskip 1\setnum{2}} } \over r} \plus {{S_{\rm w}^{\hskip 1\setnum{2}} } \over {ry}}}}\comma \qquad h_{{\rm OP}}^{\setnum{2}} \equals 2b \equals {{V_{\rm A} } \over {V_{\rm P} }}\comma

where: b, r and y=regression coefficient, number of blocks and years, respectively. S 2G, S 2e, S 2F and S 2w=estimate of genetic, non-genetic, between- and within family variances, respectively. S 2GR, S 2GY, S 2FR and S 2FY=variance component due to parents×blocks, parents×years, family×block and family×year interaction effects, respectively.

Standard errors (s.e.) of h 2b and h 2n were computed as described by Dickerson (Reference Dickerson and Chapman1969), whereas s.e. (h 2op)=2×s.e. (b).

The reference unit to which all of these estimates apply was that of parents and family mean. Heritability estimates were obtained assuming a diploid inheritance model without epistasis and non-inbred parents chosen at random from the parent varieties. Negative variance components were considered to be zero in calculating heritabilities. Expected genetic gains per cycle of selection for parents and HS families were calculated for a selection intensity of 0·2 (standardized selection differential=1·4), based on Nguyen & Sleper (Reference Nguyen and Sleper1983), as follows:

R_{\rm P} \equals ih_{{\rm op}}^{\setnum{2}} \times \sqrt {S_{{\rm ph}_{{\rm P}} }^{\hskip 1\setnum{2}} } \comma \qquad R_{\rm F} \equals iS_{\rm F}^{\hskip 1\setnum{2}} \times \sqrt {S_{{\rm ph}_{{\rm F}} }^{\hskip 1\setnum{2}} } \comma

where: R P and R F=expected gains for selection of clonal parents and HS families, respectively. h 2op and S 2F=narrow-sense heritability and genetic variance among HS families, respectively. S_{{\rm ph}_{{\rm P}} }^{\hskip 1\setnum{2}} and S_{{\rm ph}_{{\rm F}} }^{\hskip 1\setnum{2}}=phenotypic variance among parents and HS families, respectively.

The genetic correlations between traits measured for both parents and HS families were estimated from variance and covariance components of analysis. For parents, this is an estimate of the correlation of total genotypic effects (r g). Because the variance and covariance of HS families are a measure of additive gene effects, the correlation based on these statistics is an estimate of the additive genetic correlation, i.e. correlation of the breeding values (r a) (Falconer & Mackay Reference Falconer and Mackay1996). Estimates of the non-genetic (error/environment) correlation coefficient (r e) were also calculated for parents. Estimates of correlation were made for each year separately and the two years combined:

\openup3\eqalign{\tab r_{\rm g} \equals {S_{{\rm G}\lpar xy\rpar} \over \sqrt {S_{{\rm G}_{\lpar x\rpar }}^{\hskip 1\setnum{2}} \cdot S_{{\rm G}_{\lpar \hskip 1y\rpar }} ^{\hskip 1\setnum{2}} }} \comma \qquad r_{\rm a} \equals{S_{{\rm F}\lpar xy\rpar } \over {\sqrt {S_{{\rm F}_{{\lpar x\rpar }} }^{\hskip 1\setnum{2}} \cdot S_{{\rm F}_{{\lpar \hskip 1y\rpar }} }^{\hskip 1\setnum{2}} } } }\comma \cr\tab \qquad\qquad\quad r_{\rm e} \equals{ { S_{{\rm E\lpar }xy{\rm \rpar }} } \over { \sqrt {S_{{\rm E}_{{\lpar x\rpar }} }^{\hskip 1\setnum{2}} \cdot S_{{\rm E}_{{\lpar \hskip 1y\rpar }} }^{\hskip 1\setnum{2}} } } }\comma}

where: S 2G, S 2F, S 2E, S G(xy), S F(xy), S E(xy)=estimate of genetic variances, between-family variances, error variance, genetic covariance, between-family covariance and error covariance components, respectively.

When the mean square of one or both traits was negative, no genetic correlation was calculated. Approximate standard errors of r g and r a were calculated as described by Becker (Reference Becker1984).

The phenotypic correlation (r p) between two traits was then calculated as follows:

r_{\rm p} \equals {{{\rm MP}_{{\rm \lpar X\comma Y\rpar }} } \over {\sqrt {{\rm MS}_{{\rm \lpar X\rpar }} } {\rm MS}_{{\rm \lpar Y\rpar }} }}\comma

where MP(X,Y) is the progeny or clone mean cross product for the characters X and Y and MS(X) and MS(Y) are the progeny or clone mean squares for the trait X and the trait Y, respectively.

RESULTS

Descriptive statistics for DM yield, morphological and quality traits derived from analysis of parents and HS families across years are summarized in Table 1. Estimates of components of genetic variance (S 2G) and error variance (S 2e), derived from analysis of parents, and between-HS families (S 2HS) and within-HS families (S 2w), derived from analysis of progenies and corresponding heritabilities values (h 2b and h 2n), are summarized in Table 2.

Table 1.

Summary of descriptive statistics for each trait derived from analyses of parents and HS families across 2 years for yield and quality traits

Table 2.

Estimates of components of genetic variance (S2G), error variance (S2e) and broad-sense heritability (h2b) derived from analysis of parents and between-family variance (S2F), within family variances (S2w) and narrow-sense heritability (h2n) derived from analysis of HS families for yield and quality traits for individual years (±s.e.)

Both genetic and between family variances were significant for all traits (P⩽0·01) in both years, except for between-family variances for WSC in the second year (Table 2). The estimates of heritabilities for the various traits were different (h b2=0·25−0·96 and h 2n=0·13−0·85; Table 2). The heritability estimates based on individual year were relatively high for more of the traits. This is expected, since heritability estimates based on one year are exaggerated if genetic×environment interaction variance is significant. The combined analyses across years and corresponding heritability values for parents and HS families are summarized in Table 3. Estimates of components of genetic variance (S 2G) were significant in all cases except CP. The between-family variances (S F2) were significant only for heading date, tiller number and WSC. Parent×year and family×year interaction effects were significant in all cases except WSC for HS families (Table 3).

Table 3.

Estimates of components of genetic variance (S2G), parents×blocks (S2GR) parents×years (S2GY), error variance (S2e), between-family variance (S2F), family×blocks (S2FR), family×years (S2FY) interaction effects and within family variances (S2w) derived from combined analysis of parents and HS families across 2 years for yield and quality traits (±s.e.)

Three estimates of heritability were derived from the data (h 2b, h 2n and h 2op) for individual years and combined across years; these are summarized in Table 4. The estimates of heritabilities based on combined analyses take account of genetic×environment interaction components and were, as expected, lower than those for individual years (Table 4). Based on combined analyses, h 2n and h 2op for DM yield were effectively zero (less than twice the corresponding s.e.; Table 4).

Table 4.

Summary of three estimates of heritability (±s.e.) broad-sense (h2b), narrow-sense (h2n) and parent-offspring (h2OP) heritability for yield and quality traits for parents and HS progenies analysis across 2 years

Expected genetic gains were calculated for all traits in parents and HS families. The expected genetic gains for phenotypic parental selection were higher than HS families except DM yield (Table 5).

Table 5.

Estimates of predicted selection response values per one cycle of selection for yield and quality traits for parents and HS progenies analysis across 2 years

The estimates of phenotypic (r p), genotypic (r g) and additive genetic (r a) correlations from analysis of both parents and HS families are summarized in Table 6 for yield and morphological traits, Table 7 for yield and quality traits and Table 8 for three quality traits. The genetic and between-families variances for CP were negative or not significant in parents and HS families (see Table 3) and in such cases no r g and r a correlations were calculated. Genetic correlations had large sampling variances and the approximate s.e. values were large. Only where the r g or r a value was greater than twice its s.e. was it considered to be significant. This approximate test was used previously for this statistic by Hill & Leath (Reference Hill and Leath1975). Estimates of the non-genetic (error/environment) correlation coefficient (r e) for parents were calculated (Tables 6–8). With some exceptions, the r e values were small and not significant (Tables 6–8). DM yield had strong negative and positive correlations (r g and r a) with heading date and tiller number, respectively. Heading date had strong negative r g correlation with tiller number (Table 7). All the estimates of r g and r a among DM yield with both DDM and WSC were generally negative but not significant. However, the values for HS families were lower than those for parents (Table 7). The correlation between DM yield and CP was inconsistent across years. No r g and r a correlations were calculated in the combined analysis, because of negative values of genetic and family variance components (Table 7).

Table 6.

Phenotypic (rp), genotypic (rg±s.e., ra±s.e.) and environmental (re±s.e.) correlation coefficients among yield and morphological trait estimates from individual years and combined analysis across 2 years for parents and HS families

Table 7.

Phenotypic (rp), genotypic (rg±s.e., ra±s.e.) and environmental (re±s.e.) correlation coefficients among yield, morphological and three quality traits: content of DDM, WSC and CP estimates from combined analysis of variance and covariance for parents and HS families across 2 years. P values >0·05 are not included

* Mean square (MS) of one or both traits was not significant.

Table 8.

Phenotypic (rp), genotypic (rg±s.e., ra±s.e.) and environmental (re±s.e.) correlation coefficients among content of DDM, WSC and CP estimates from individual years and combined analysis across 2 years for parents and HS families

* Mean square (MS) of one or both traits was not significant.

The correlations among heading date and quality traits were inconsistent, although any significant values between heading date and DDM were positive (Table 7). Tiller number had positive relationship with DDM, but its relationship with CP was inconsistent. Estimates of additive correlation r a between tiller number and WSC were also strongly positive and significant (Table 7). All the estimates of r g and r a between DDM and WSC were strongly positive and significant (Table 8). Genetic correlations among CP with both WSC and DDM were generally negative and some were significant. The r e correlation for WSC v. CP was strong and negative for the first year and combined across years (Table 8).

DISCUSSION

The variations for DM yield, morphological and quality traits were always wider in parents than in HS families for all traits except CP. These differences between two generations are expected, since variation among parents is controlled by both additive and non-additive (dominant) genetic variance, whereas variation among HS families is only controlled by additive variance (Falconer & Mackay Reference Falconer and Mackay1996). The mean of DM yield was relatively high for HS families. Jafari (Reference Jafari1998) similarly detected higher DM yield in HS families of perennial ryegrass than those for parental clones and suggested that plants grown from clonal propagated tillers are less resistant to environmental hazards than those grown from seeds. In contrast, the average values of heading date, WSC and DDM were relatively high in parents compared with HS families. Since DM yield had negative relationships with both WSC and DDM (Table 8), this result might be expected.

Estimates of components of genetic variance (S 2G) were significant in all cases except CP. The between-family variances (S 2F) were significant only for heading date, tiller number and WSC, which suggests that additive genetic variance was the main component controlling these traits. Parent×year (S 2GY) and family×year (S 2FY) interaction effects were significant in all cases except WSC in HS families (Table 2). Marum et al. (Reference Marum, Rognli, Aastveit, Aastveit, Reheul and Ghesquière1994) also found significant genotype×environment interactions for DDM in cocksfoot. In contrast, no significant genotype×environment interactions were found by Walters & Evans (Reference Walters and Evans1974) for DDM, Sanada et al. (Reference Sanada, Takai and Yamada2004) for WSC and Shenk & Westerhaus (Reference Shenk and Westerhaus1982) for CP. Buxton & Casler (Reference Buxton, Casler, Jung, Buxton, Hatfield and Ralph1993), in a review, concluded that genotype×environment interactions should be smaller for forage quality than for DM yield and that quality traits might be relatively stable across environments. However, the results of the present study indicate the presence of genotype×environment interactions for DDM and CP. When genetic×environment interactions are significant then evaluation prior to selection is more difficult. Ideally, more than one environment (e.g. years and locations) should be used to assess the breeding materials. The estimates of heritabilities based on combined analyses were lower than those for individual years (Table 4). Based on combined analyses, h 2n and h 2op for DM yield were effectively zero (Table 4). For DM yield, the results of h 2n estimates were similar to those of Casler (Reference Casler1998) and Nguyen & Sleper (Reference Nguyen and Sleper1983), but lower than those of Frandsen (Reference Frandsen1986) and Annicchiarico & Romani (Reference Annicchiarico and Romani2005), whose published data suggested that both additive and non-additive gene effects play an important role in the genetic regulation of DM yield. But in the present study, non-additive genetic variance is probably of greatest importance. For both heading date and tiller number, there was little difference between h 2b and h 2n, which suggests that genetic variations in these characteristics are controlled largely by additive gene action (Table 4).

Estimates of h 2b for DDM were moderate, whereas h 2n estimates were low and inconsistent, which suggest that genetic variation in this characteristic is controlled mainly by non-additive gene effects. This is in agreement with Beerepoot et al. (Reference Beerepoot, Bouter, Dijkstra, Reheul and Ghesquière1994) and Marum et al. (Reference Marum, Hovin, Marten and Shenk1979) reported that both additive and dominance gene effects influence DDM. But Frandsen (Reference Frandsen1986) and Sleper et al. (Reference Sleper, Drolsom and Jorgensen1973) found that genetic variance for DDM was mainly additive. Estimates of h 2b and h 2op for WSC were moderate to high, whereas h 2n estimates were low, suggesting that genetic variation in this trait is controlled by both additive and non-additive gene action. This is in agreement with Cooper (Reference Cooper1962) and Grusea & Oprea (Reference Grusea, Oprea, Reheul and Ghesquière1994), who concluded that for WSC in cocksfoot, gene action was additive. However, in perennial ryegrass, Humphreys (Reference Humphreys1989a, Reference Humphreysb) found that WSC behaved as a complex polygenic trait, which was controlled mainly by non-additive gene effects. There was no significant variation for CP in either generation; the estimates of h 2b and h 2n values were low for individual years and they were effectively zero across years. The estimates of h 2op based on regression analysis over 2 years, with two exceptions, were always higher than h 2n estimated from variance components. This is in agreement with Vogel et al. (Reference Vogel, Haskins and Gorz1980), who suggested that h 2op would be overestimated if the parents and offspring shared the same plot. Therefore, they proposed that the regression of offspring in one replication and parents in another replication would remove such bias.

The expected genetic gains for phenotypic parental selection were higher than those for HS families, except DM yield. Based on the present findings, clonal evaluation appeared to be adequate to select parents for highly heritable characteristics, such as morphological and quality traits. Similar results for DM yield, were obtained by Pavetti et al. (Reference Pavetti, Sleper, Roberts and Krause1994), who suggested that progress with selection for improved herbage and DM yield is likely to be slow.

The results of the correlation analysis indicated general agreement in both sign and magnitude between genotypic and phenotypic correlations. This suggests that for all traits, genetic and phenotypic correlations have similar effects. In a comparison of the estimates of r a and r g, the results showed that where r g or r a values were considered to be significant, the relevant generation had the same sign; otherwise they were not significant. With few exceptions, r e values were always small and not significant. This indicates that the phenotypic association for all other pair-wise combinations of traits measured on parental clones was due to genetic rather than environmental factors.

Genetic correlations between DM yield and heading date were strongly negative, similar to the findings of Martiniello (Reference Martiniello1998) and Kanapeckas et al. (Reference Kanapeckas, Tarakanovas and Lemežienë2005). Both r a and r g correlations were positive and significant for DM yield with tiller number. This result was not unexpected because tiller number is a yield component. Heading date had a strong negative r g correlation with tiller number. These results suggest that selection of early flowering accessions would lead to more tillers and increased DM yield in cocksfoot. All the estimates of r g and r a between DM yield v. DDM were generally negative but not significant. However, the values for HS families were lower than those for parents (Table 7). The negative correlation between DDM and DM yield has also been reported for cocksfoot (Brown & Blaser Reference Brown and Blaser1970; Marum et al. Reference Marum, Rognli, Aastveit, Aastveit, Reheul and Ghesquière1994). However, in contrast, some published data suggest that DDM is largely independent of DM yield (Frandsen Reference Frandsen1986; Humphreys Reference Humphreys1989b). The results from the current study do not agree with these conclusions and instead indicate a weak negative relationship between these traits. The r g estimates for DM yield v. WSC were consistently negative, but the r a values were inconsistent (Table 7). Brown & Blaser (Reference Brown and Blaser1970) and Sanada et al. (Reference Sanada, Takai and Yamada2004) in cocksfoot, and Jafari et al. (Reference Jafari, Connolly and Walsh2003a) and Marais et al. (Reference Marais, Goodenough, De Figueiredo and Hopkins2003) in ryegrass, found inconsistent relationships between these two traits, although significant values were generally positive. In the present study, the values of r a are in agreement with those reported in the above publications, but the r g obtained from analysis of parents indicate a weak negative relationship between these traits. The correlation between DM yield and CP was inconsistent for individual years, whereas, based on combined analysis because of negative values of genetic and family variance components, no r g and r a correlations were calculated (Table 7). Low and inconsistent relationships between these two traits have also been reported by Lamb et al. (Reference Lamb, Vogel and Reece1984) and Ray et al. (Reference Ray, Karn and Dara1996). In contrast, Jafari et al. (Reference Jafari, Connolly and Walsh2003a) and Humphreys (Reference Humphreys1989b) in perennial ryegrass obtained negative correlations between these two traits under sward conditions. The latter author suggested that this negative relationship may be inconsistent where nitrogen availability is low.

The correlations between heading date and quality traits were inconsistent, although significant values between heading date and DDM were positive (Table 7). Hacker (Reference Hacker and Hacker1982), in a review, suggested that the correlations between heading date and quality traits are highly variable and depend on date of sampling and regrowth interval. Breese & Davies (Reference Breese and Davies1970) suggested that selection for high digestibility in cocksfoot was accompanied by faster growth rate and earlier heading date. The present results were in agreement with Quesenberry et al. (Reference Quesenberry, Sleper and Cornell1978), who found significant positive correlation between digestibility and ear emergence date. However, when their comparisons were made at the same morphological stage of growth, early flowering types were usually more digestible. In this investigation, the relationships between heading date and WSC were inconsistent, although Sanada et al. (Reference Sanada, Takai and Yamada2004) found positive correlation between heading date and WSC in cocksfoot.

Tiller number had positive relationship with DDM, but its relationship with CP were inconsistent. Estimates of additive correlation r a between tiller number and WSC were strong, positive and significant (Table 7). This positive correlation between DDM and tiller number was in agreement with results previously reported for spaced plants (Clements Reference Clements1973; Humphreys Reference Humphreys1989c). The positive estimates for spaced plants may be associated with the absence of competition under the present growth environment, where genotypes with large tillers (thicker stems) have high yield and such tillers also have high digestibility and low lignified vascular tissue (Ehlke & Casler Reference Ehlke and Casler1985).

All the estimates of r g and r a for the relationship of DDM with WSC were strongly positive and significant (Table 8). Since WSC is completely digestible, a positive correlation between these two parameters is expected and is in agreement with Jafari et al. (Reference Jafari, Connolly and Walsh2003a) and Humphreys (Reference Humphreys1989b) in perennial ryegrass. The r g and r p estimates between DDM and CP were negative and significant in the first year. But the r a values were inconsistent and not significant (Table 8). Reports from the literature for a relationship between these two traits are inconsistent. The present results are in agreement with Jafari et al. (Reference Jafari, Connolly and Walsh2003a) and Humphreys (Reference Humphreys1989c), who reported a strong negative relationship, whereas Frandsen (Reference Frandsen1986) and Marum et al. (Reference Marum, Hovin, Marten and Shenk1979) found positive correlations between the two traits. Radojevic et al. (Reference Radojevic, Simpson, St. John and Humphreys1994) concluded that relationships between these two traits are strongly influenced by environmental effects such as drought, light intensity and nitrogen level during the growing season. The relationships between WSC and CP were generally negative and some r g and r p values were significant (Table 8). The r e correlation was strongly negative in the first year and when combined across years, which suggested that the relationship between these traits is affected by environmental factors as well as by correlated genetic effects. Such environmentally induced effects have also been reported by Humphreys (Reference Humphreys1989c), who suggested that as growth increases with rapid uptake of nitrogen fertilizer, increase of CP and decrease of WSC content are environmentally induced effects.

It was concluded that genotype×year interactions were present for all traits except WSC in HS families, suggesting that more than one environment should be used to assess the breeding material. The results of analyses showed that the estimates of h 2b were always larger than either h 2n or h 2op, indicating that some non-additive variance was present in almost all analyses. For DDM and WSC, the h 2b estimates were relatively high, whereas the h 2n and h 2op estimates were low to moderate, indicating that both additive and non-additive effects were important in controlling the expression of this trait. With few exceptions, the value of h 2op was more than that of h 2n, indicating that narrow-sense heritability estimated from parents/offspring might be an overestimate. The h 2b estimates for DM yield were moderate, whereas the h 2n and h 2op estimates were negligible, indicating that non-additive effects were important in controlling expression of this trait. Thus, to improve DM yield, recurrent selection based on progeny testing should be effective. The h 2n and h 2op estimates for heading date and tiller number were relatively high, about the same magnitude as h 2b, which suggested that additive genetic variance was the main component controlling these traits and that response to selection would be likely.

The weak negative correlation between DDM and total DM yield indicates that combined selection for both DDM and DM yield should elicit a response. Selection for DDM alone could result in reduction in yield. Given the relationship between WSC, CP, DM yield and DDM, it is tempting to suggest that selection for high WSC is a means to improve quality in general. Beerepoot & Agnew (Reference Beerepoot, Agnew and Weddell1997) have argued that this simple approach may not result in improved herbage quality because of possible negative effects on rumen pH. There is, however, indirect evidence that higher WSC in ryegrass may result in improved animal performance. On the basis of the present results, it is suggested that increased WSC, when CP is in excess, would improve herbage quality.

The study was part of a project for developing improved grass varieties (code: 1-016-11-7901). We thank Dr M. H. Assareh, Head of Research Institute of Forests and Rangelands that funded this research and Dr H. M. Arefi, Head of Iranian Natural Resource Gene Bank that provided seed samples.

References

REFERENCES

Annicchiarico, P. & Romani, M. (2005). Genetic variation, heritability and genetic correlation for forage quality and yield traits of Mediterranean tall fescue germplasm. Plant Breeding 124, 99–101.CrossRefGoogle Scholar
Becker, W. A. (1984). Manual of Quantitative Genetics, 4th edn. Pullman, WA: Academic Enterprises.Google Scholar
Beerepoot, L. J. & Agnew, R. E. (1997). Breeding for improved herbage quality in perennial ryegrass. In Seeds of Progress (Ed. Weddell, J. R.), pp. 135–145. Occasional Symposium of the British Grassland Society, Vol. 31Google Scholar
Beerepoot, L. J., Bouter, W. & Dijkstra, J. A. (1994). Breeding for improved digestibility in perennial ryegrass. In Breeding for Quality: Proceeding of the 19th EUCARPIA Fodder Crops Section Meeting, Brugge, Belgium, 5–8 October 1994 (Eds Reheul, D. & Ghesquière, A.), pp. 237–245. Wageningen, The Netherlands: EUCARPIA.Google Scholar
Beever, D. E. & Reynolds, C. K. (1994). Forage quality, feeding value and animal performance. In Grassland and Society: Proceedings of the 15th EUCARPIA General Meeting of the European Grassland Federation, 6–9 June 1994 (Eds Mannetje, L.'t & Frame, J.), pp. 48–60. Wageningen, The Netherlands: Wageningen Pers.Google Scholar
Breese, E. L. & Davies, W. E. (1970). Selection for factors affecting nutritive value. In Jubilee Report of the Welsh Plant Breeding Station 1919–1969, pp. 33–35. Aberystwyth, Wales, UK: Welsh Plant Breeding Station.Google Scholar
Brown, R. H. & Blaser, R. E. (1970). Soil moisture and temperature effects on growth and soluble carbohydrates of orchardgrass (Dactylis glomerata L.). Crop Science 10, 213–216.CrossRefGoogle Scholar
Buxton, D. R. & Casler, M. D. (1993). Environmental and genetic effects on cell wall composition and digestibility. In Forage Cell Wall Structure and Digestibility (Eds Jung, H. G., Buxton, D. R., Hatfield, R. D. & Ralph, J.), pp. 685–714. Madison, WI: American Society of Agronomy.Google Scholar
Carlier, L. (1994). Breeding forage quality, feeding value and animal performance. In Breeding for Quality: Proceeding of the 19th EUCARPIA Fodder Crops Section Meeting, Brugge, Belgium, 5–8 October 1994 (Eds Reheul, D. & Ghesquière, A.), pp. 25–27. Wageningen, The Netherlands: EUCARPIA.Google Scholar
Casler, M. D. (1998). Genetic variation within eight populations of perennial forage grasses. Plant Breeding 117, 243–249.Google Scholar
Christie, R. B. & Mowat, D. N. (1968). Variability in in vitro digestibility among clones of brome grass and orchard grass. Canadian Journal of Plant Science 48, 67–73.Google Scholar
Clements, R. J. (1973). Breeding for improved nutritive value of Phalaris tuberosa herbage: an evaluation of alternative sources of genetic variation. Australian Journal of Agricultural Research 24, 21–34.CrossRefGoogle Scholar
Connolly, V., do Valle Ribeiro, M. & Crowley, J. G. (1977). Potential of grass and legume cultivars under Irish conditions. In Proceedings of the International Meeting on Animal Production from Temperate Grassland, Dublin, June 1977 (Ed. Gilsenan, B.), pp. 23–28.Google Scholar
Cooper, J. P. (1962). Selection for nutritive value. In Report of the Welsh Plant Breeding Station for 1961, pp. 145–156.Google Scholar
Cooper, J. P. (1973). Genetic variation in herbage constituents. In Chemistry and Biochemistry of Herbage, Vol. 2 (Eds Butler, G. W. & Bailey, R. W.), pp. 379–417. London: Academic Press.Google Scholar
Dickerson, G. E. (1969). Techniques for research in quantitative animal genetics. In Techniques and Procedures in Animal Production Research (Ed. Chapman, A. B.), pp. 36–79. New York: American Society of Animal Science.Google Scholar
Ehlke, N. J. & Casler, M. D. (1985). Anatomical characteristics of smooth bromegrass clones selected for in vitro dry matter digestibility. Crop Science 25, 513–517.Google Scholar
Falconer, D. S. & Mackay, T. F. C. (1996). Introduction to Quantitative Genetics, 4th edn. London: Longman.Google Scholar
Frandsen, K. J. (1986). Variability and inheritance of digestibility in perennial ryegrass (Lolium perenne), meadow fescue (Festuca pratensis), and cocksfoot (Dactylis glomerata L.). II. F1 and F2 progeny. Acta Agriculturae Scandinavica 36, 241–263.Google Scholar
Grusea, A. & Oprea, G. (1994). Variation and inheritance of quality of Dactylis glomerata varieties which were obtained by different breeding methods. In Breeding for Quality: Proceeding of the 19th EUCARPIA Fodder Crops Section Meeting, Brugge, Belgium, 5–8 October 1994 (Eds Reheul, D. & Ghesquière, A.), pp. 145–149. Wageningen, The Netherlands: EUCARPIA.Google Scholar
Hacker, J. B. (1982). Selecting and breeding better quality grasses. In Nutritional Limits to Animal Production from Pasture Proceedings of an International Symposium, Queensland, August 1981, Australia (Ed. Hacker, J. B.), pp. 305–326. Farnham, Surrey, UK: Commonwealth Agricultural Bureaux.Google Scholar
Hill, R. R. & Leath, K. T. (1975). Genotypic and phenotypic correlations for reaction of five foliar pathogens in alfalfa. Theoretical and Applied Genetics 45, 254–258.CrossRefGoogle ScholarPubMed
Humphreys, M. O. (1989 a). Water soluble carbohydrates in perennial ryegrass breeding. II. Cultivar and hybrid progeny performance in cut plot. Grass and Forage Science 44, 237–244.CrossRefGoogle Scholar
Humphreys, M. O. (1989 b). Water-soluble carbohydrates in perennial ryegrass breeding. III. Relationships with herbage production, digestibility and crude protein content. Grass and Forage Science 44, 423–430.CrossRefGoogle Scholar
Humphreys, M. O. (1989 c). Assessment of perennial ryegrass (Lolium perenne L.) for breeding. II. Components of winter hardiness. Heredity 41, 99–106.Google Scholar
Jafari, A. (1998). Genetic analysis of yield and quality in perennial ryegrass (Lolium perenne L.). Ph.D. thesis, Department of Crop Science, Horticulture and Forestry, University College Dublin, Ireland.Google Scholar
Jafari, A., Connolly, V. & Walsh, E. J. (2003 a). Genetic analysis of yield and quality in full sib families of perennial ryegrass (Lolium perenne L.) under two cutting managements. Irish Journal of Agricultural and Food Research 42, 275–292.Google Scholar
Jafari, A., Connolly, V., Frolich, A. & Walsh, E. J. (2003 b). A note on estimation of quality parameters in perennial ryegrass by near infrared reflectance spectroscopy. Irish Journal of Agricultural and Food Research 42, 293–299.Google Scholar
Kanapeckas, J., Tarakanovas, P. & Lemežienë, N. (2005). Variability, heritability and correlations of genetic resources in meadow fescue. Biologija 3, 10–14.Google Scholar
Lamb, J. F. S., Vogel, K. P. & Reece, P. E. (1984). Genotype and genotype×environment interaction effects on forage yield and quality of crested wheatgrass. Crop Science 24, 559–564.CrossRefGoogle Scholar
Marais, J. P., Goodenough, D. C. W., De Figueiredo, M. & Hopkins, C. (2003). The development of a Lolium mutiflorum cultivar with low moisture content and an increased readily digestible energy to protein ratio. Australian Journal of Agricultural Research 54, 101–106.Google Scholar
Martiniello, P. (1998). Influence of agronomic factors on the relationship between forage production and seed yield in perennial forage grasses and legumes in a Mediterranean environment. Agronomie 18, 591–601.CrossRefGoogle Scholar
Marum, P., Hovin, A. W., Marten, G. C. & Shenk, J. S. (1979). Genetic variability for cell wall constituents and associated quality traits in reed canarygrass. Crop Science 19, 355–360.CrossRefGoogle Scholar
Marum, P., Rognli, O. A., Aastveit, A. H. & Aastveit, K. (1994). Improved digestibility and protein content as breeding problems in Norwegian timothy (Phleum pratense L.) and cocksfoot (Dactylis glomerata L.). In Breeding for Quality: Proceeding of the 19th EUCARPIA Fodder Crops Section Meeting, Brugge, Belgium, 5–8 October 1994 (Eds Reheul, D. & Ghesquière, A.), pp. 137–141. Wageningen, The Netherlands: EUCARPIA.Google Scholar
Nguyen, H. T. & Sleper, D. A. (1983). Genetic variability of seed yield and reproductive characters in tall fescue. Crop Science 23, 621–626.CrossRefGoogle Scholar
Niaky, S. (1995). Land Grass Cover of Iran (in Persian). Ahwaz, Iran: Chamran University Press.Google Scholar
Pavetti, D. I. T., Sleper, D. A., Roberts, C. A. & Krause, G. F. (1994). Genetic variation and relationship of quality traits between herbage and seed of tall fescue. Crop Science 34, 427–431.Google Scholar
Quesenberry, K. H., Sleper, D. A. & Cornell, J. A. (1978). Heritability and correlation of IVDMD, maturity and plant height in Rhodes grass. Crop Science 18, 847–849.Google Scholar
Radojevic, I., Simpson, R. J., St. John, J. A. & Humphreys, M. O. (1994). Chemical composition and in vitro digestibility of lines of Lolium perenne selected for high concentrations of water soluble carbohydrate. Australian Journal of Agricultural Research 45, 901–912.CrossRefGoogle Scholar
Ray, I. M., Karn, J. F. & Dara, S. T. (1996). Heritabilities of nutritive quality factors and interrelationships with yield in tetraploid crested wheatgrass. Crop Science 36, 1488–1491.CrossRefGoogle Scholar
Rechinger, K. H. (1970). Flora Iranica. No. 70. Graz, Austria: Akademische Druck und Verlagsanstalt.Google Scholar
Sanada, Y., Takai, T. & Yamada, T. (2004). Genetic variation in water-soluble carbohydrate concentration in diverse cultivars of Dactylis glomerata L. during vegetative growth. Australian Journal of Agricultural Research 55, 1183–1187.CrossRefGoogle Scholar
Shenk, J. S. & Westerhaus, M. O. (1982). Selection for yield and quality in orchardgrass. Crop Science 22, 422–425.CrossRefGoogle Scholar
Sleper, D. A., Drolsom, P. N. & Jorgensen, N. A. (1973). Breeding for improved dry matter digestibility in smooth bromegrass (Bromus inermis Leyss.). Crop Science 13, 556–558.Google Scholar
Smith, K. F., Reed, K. F. M. & Foot, J. Z. (1997). An assessment of relative importance of specific traits for the genetic improvement of nutritive value in dairy pasture. Grass and Forage Science 52, 167–175.Google Scholar
Steel, R. G. D. & Torrie, J. H. (1980). Principles and Procedures of Statistics: A Biometrical Approach, 2nd edn. London: McGraw-Hill Book Company.Google Scholar
Vogel, K. P., Haskins, F. A. & Gorz, H. J. (1980). Parent-progeny regression in indiangrass: inflation of heritability estimates by environmental covariances. Crop Science 20, 580–582.CrossRefGoogle Scholar
Wheeler, J. L. & Corbett, J. L. (1989). Criteria for breeding forages of improved feeding value: results of a Delphi survey. Grass and Forage Science 44, 77–83.CrossRefGoogle Scholar
Walters, R. J. K. & Evans, E. M. (1974). Digestibility and Voluntary Intake of Be-6393 Cocksfoot. Report of Welsh Plant Breeding Station for 1973, pp. 43–44. Aberystwyth, UK: Welsh Plant Breeding Station.Google Scholar
Figure 0

Table 1.Summary of descriptive statistics for each trait derived from analyses of parents and HS families across 2 years for yield and quality traits

Figure 1

Table 2.Estimates of components of genetic variance (S2G), error variance (S2e) and broad-sense heritability (h2b) derived from analysis of parents and between-family variance (S2F), within family variances (S2w) and narrow-sense heritability (h2n) derived from analysis of HS families for yield and quality traits for individual years (±s.e.)

Figure 2

Table 3.Estimates of components of genetic variance (S2G), parents×blocks (S2GR) parents×years (S2GY), error variance (S2e), between-family variance (S2F), family×blocks (S2FR), family×years (S2FY) interaction effects and within family variances (S2w) derived from combined analysis of parents and HS families across 2 years for yield and quality traits (±s.e.)

Figure 3

Table 4.Summary of three estimates of heritability (±s.e.) broad-sense (h2b), narrow-sense (h2n) and parent-offspring (h2OP) heritability for yield and quality traits for parents and HS progenies analysis across 2 years

Figure 4

Table 5.Estimates of predicted selection response values per one cycle of selection for yield and quality traits for parents and HS progenies analysis across 2 years

Figure 5

Table 6.Phenotypic (rp), genotypic (rg±s.e., ra±s.e.) and environmental (re±s.e.) correlation coefficients among yield and morphological trait estimates from individual years and combined analysis across 2 years for parents and HS families

Figure 6

Table 7.Phenotypic (rp), genotypic (rg±s.e., ra±s.e.) and environmental (re±s.e.) correlation coefficients among yield, morphological and three quality traits: content of DDM, WSC and CP estimates from combined analysis of variance and covariance for parents and HS families across 2 years. P values >0·05 are not included

Figure 7

Table 8.Phenotypic (rp), genotypic (rg±s.e., ra±s.e.) and environmental (re±s.e.) correlation coefficients among content of DDM, WSC and CP estimates from individual years and combined analysis across 2 years for parents and HS families