Half-sib progenies evaluation in velvet grass

Velvet grass (Holcus lanatus L.) has a great potential use for winter pasture in subtropical regions due to its good establishment, persistence, high forage production, cold resistance, palatability and tillering ability. The objective of this work was to estimate genetic variability and heritability of agronomic traits and to identify superior progenies. Sixty half-sib progenies were assessed for natural height at the vegetative stage (NH), plant diameter (diameter), heading date (cycle), final height (after flowering) (FH), and tiller number (tillers). Difference among progenies was observed for all traits. Heritability estimates were 38%, 32%, 92%, 57% and 64% for NH, diameter, cycle, FH and tillers, respectively. The highest genetic gain estimate was 30.77 % for the tiller number. There is genetic variability for all the traits and gains after selection among halfsib progenies can be expected.


INTRODUCTION
Velvet grass (Holcus lanatus L.) is a grass from the temperate areas of Europe and Asia and from Canary Islands and currently it is widely distributed in temperate regions around the world (Watt 1978, Pitcher andRusso 2003).
It is a cold season, allogamic plant (Watt 1978), which in Brazil behaves as an annual or biannual (Araújo 1956, Oliveira et al. 2001, Pitcher and Russo 2003).It is a competitive species that tolerates a wide range of edaphoclimatic factors (Watt 1978).This species adapts mainly to humid lands (Klitsch 1965) and can survive during short periods of drought (Pitcher and Russo 2003), also being a species resistant to cold (Oliveira et al. 2001, Pitcher andRusso 2003).
Its adaptation to soil pH variation is wide, having as optimal pH conditions a range from 5.0 to 7.5 (Pitcher and Russo 2003).Velvet grass prefers soils with high organic matter content (Araújo 1956) and responds well to fertilization (Oliveira et al. 2001).The adaptation in poor soils and the high competitiveness of velvet grass are due to its radicular system, which is well developed and aggressive (Watt 1978).
It has been considered to be equivalent to Italian ryegrass at intermediary fertility conditions and moderate stocking rate (Rumball 1980).In Uruguay, it surmounts the Italian ryegrass in dry matter production in the critical periods of autumn, winter and the beginning of the spring season, with similar quality (Bemhaja 1993).
The species was introduced by the veterinarian Charles Vincent at Ponta Grossa, the State of Paraná, Brazil,

NOTE NOTE NOTE NOTE NOTE
Half-sib progenies evaluation in velvet grass in 1912.It adapted perfectly to the edaphoclimatic conditions of the State of Rio Grande do Sul, in which it was introduced in 1917 at the Posto Zootécnico de Viamão, where it started to disseminate to the rest of the State.Currently, it occurs spontaneously in pasture areas (Araújo 1956).
According to Oliveira et al. (2001), the species produces forage of intermediary quality; however, it has high palatability.Good acceptance of velvet grass forage by the cattle was described by Araújo (1956).In the experiments performed at the Embrapa Temperate Agriculture, velvet grass has shown a protein content equal to the tall fescue (Festuca arundinacea) and superior to Bromus catharticus (Gomes and Reis 2000).
Velvet grass has high tillering ability and produces forage earlier than other perennial temperate species (Araújo 1956, Oliveira et al. 2001).This grass evidenced a perfect adaptation to the environment and also a good productivity and may be used as pasture or hay (Araújo 1956).Other advantages of this species are the intense flowering and high seed production in the establishment year (Araújo 1956, Moraes et al. 2000).Besides the forage production, velvet grass might be used in soil erosion control and the recovery of soils with low fertility and acidity.
Due to all these traits, velvet grass is a very interesting winter pasture for regions with subtropical climate and, among the winter perennial grasses, the best adapted to Southern Brazil.However, there are no commercial varieties in the country and no seed is available for farmers.
The first step for the development of velvet grass varieties is to know the existing variability in the most important traits as well as the genetic parameters of its populations.Therefore, the objective of this work was to estimate genetic variability and assess the heritability for traits of agronomic interest in a velvet grass population collected in the State of Rio Grande do Sul, Brazil, through the evaluation of half-sib progenies, and to identify superior progenies for each trait.

MATERIAL AND METHODS
The experiment was accomplished at the Low Lands Experimental Station of Embrapa Temperate Agriculture in the municipality of Capão do Leão (lat 31º 52' 00" S, long 52 º 21' 24" W, alt 13.24 m asl), State of Rio Grande do Sul, Brazil, during 2003.Sixty half-sib progenies collected from a population in the municipality of Bagé, State of Rio Grande do Sul, Brazil were studied.The Uruguayan commercial cultivar La Magnolia was used as a check.
Progenies were sown in trays on May 22 nd , and trays were kept in the greenhouse.On July 10 th , the seedlings were transplanted to the field.The experimental design was a randomized block, with three replications.Each plot was represented by one line with ten plants.Spacing among lines and among plants within lines was 0.3 m.
On September 29 th , in the vegetative stage of the plants, the first assessments were made, measuring the natural height (NH, cm) and plant diameter (diameter, cm).Heading started at October 8 th , and was evaluated weekly.Heading date (cycle) was measured as days from sowing to the emission of the first panicle.After flowering, two traits were assessed: final height (FH, cm), from the ground level until the insertion of the flag leaf, and tiller number (tillers): counting undertaken on November 27 th .Heading date and final height were assessed in all plants and the other traits in three plants per plot.
Analyses of variance were performed on a plot mean basis with SAS software (SAS Institute 1994).Progeny means were compared by the Scott-Knot test, with Genes software (Cruz 2001).Comparisons with the check variety were made by the t test.Variance components were estimated by the analysis of variance method using the software SAS (SAS Institute 1994).Additive genetic variances were estimated as indicated for half-sib progenies (Hallauer and Miranda Filho 1988): where σ hsp is the variance component associated to the effects of half-sib progeny.
Narrow-sense heritabilities were estimated by the expression: where σ 2 is the variance component associated to the error and r is the number of replications.
The selection gain was estimated for a scheme involving the selection among progenies and recombination through remnant seeds and is shown in absolute and percentual values.
Simple phenotypic correlations were estimated among all traits.

RESULTS AND DISCUSSION
Significant differences among treatments occurred for all traits (Table 1).With regard to NH, the means of the progenies varied between 8.00 and 17.22 cm, and the population mean was 13.14 cm (Table 2).Progenies were separated in two groups according to the Scott and Knott test.Plant diameter varied between 14.17 and 33.78 cm, and two groups were formed (Table 2).These traits may be selected as estimates of early production.In a Brazilian population of Italian ryegrass, correlations with early dry matter production were 0.50 for natural height and 0.54 for diameter (Mittelmann et al. 2006).Castro et al. (2003) found an even higher value for the diameter x dry matter yield correlation (0.68).
Heading date varied between 155.6 and 177.7 days and the population mean was 164.1 days (Table 4).For this trait, five groups of progenies were formed.Late flowering is usually desirable, extending the period of utilisation of the pasture.Final height varied among 38.03 and 50.13 cm, with a mean of 44.44 cm, and two groups of progenies were formed (Table 3).Concerning the number of tillers, the Scott and Knot test has pointed to only one group, despite the variation of the means, which ranged from 12.89 to 52.89 tillers per plant, and the significance through the F test (Table 3).
The check variety was the Uruguayan cultivar La Magnolia.For all traits, the population mean was superior to the check, indicating a potential for the selection of more adapted and productive cultivars.Regarding the cycle, there is a huge possibility of selection for late genotypes as well as for early genotypes, because there were also progenies with significantly lower means than the check.
Estimates of variance components associated to progeny effect and error are shown on Table 4.The estimates of additive variance, were 4. 06; 17.62; 109.76; 17.61 and 132.84 for NH (cm), diameter (cm), heading date (days), final height (cm) and tiller number, respectively.
Among the studied traits, heading date has shown the highest heritability, of 92% (Table 4).This value indicates a low effect of the environment and that this trait is easy to select.It is possible that this population is segregating for a gene of large effect on the heading date.Final height and tiller number showed heritabilities that might be considered as intermediary values, while NH and diameter have low values (Table 4).It is important to consider that these heritability estimations are in a narrow sense, once the variance among half-sib progenies involves basically additive effects.In maize, the model plant to the study of allogamous species, average estimates of heritability of 56.9% for plant height, 57.9% for heading date and 71.9% for number of tillers were found (Hallauer and Miranda 1988).In meadow bromegrass (Bromus riparius Rehm.), Araújo and Coulman (2004) have found broad-sense heritabilities of 90% for height and 58% for diameter.
Estimated selection gains for one cycle of selection and 10% intensity indicated extremely favourable results, especially for tiller number, with a gain of 30.77% over the original population mean.The expected gains were of 8.28% for NH; 8.11% for diameter; 5.37% for heading date and 6.25% for final height (Table 4).Natural populations, which were not submitted to selection, usually show large Half-sib progenies evaluation in velvet grass genetic variability, allowing high gains in the first cycles of selection.Due to the high heritability, an extremely high genetic gain, of 8.81 days, is expected for heading date.There were associations among traits, except for NH and tiller number, and heading date and tiller number.In general, the correlation values were low (Table 5).The highest value found was among the plant diameter and tiller number (r = 0.43).The NH and diameter traits measured in the initial stage of the plant development with the objective of selecting for early dry matter production has shown a certain degree of association (r = 0.35), besides having a negative correlation with heading date, which might point to a tendency of delayed flowering progenies for having a slower initial development.Mittelmann et al. (2004) have found similar correlations, and considered that they are of low value and do not hind the selection for an extended cycle of utilization of the pasture.

Table 1 .
Analysis of variance for natural height (cm), plant diameter (cm), heading date (days), final height (cm) and tiller number in halfsib progenies of velvet grass *

Table 2 .
Means of 60 half-sib progenies of velvet grass for natural height (cm) and plant diameter (cm) Means followed by the same letter do not differ significantly through Scott & Knott test with 5% of significance.*iDiffers significantly from the check through the t test with 5% of significance.

Table 3 .
Means of 60 progenies of half-sibs for heading date (days), final height (cm) and tiller number Means followed by the same letter do not differ significantly through Scott & Knott test with 5% of significance.* Differs significantly from the control through the t test with 5% of significance.

Table 5 .
Simple phenotypic correlations between traits in a velvet grass population