Effect of seed rate and row spacing in seed production of Festulolium

Abstract

Festulolium (×Festulolium) is a cross between the two species fescue (Festuca L.) and ryegrass (Lolium L.) and is a promising forage and seed crop. To stimulate the production of Danish organic festulolium seeds a three-year field experiment was performed from 1999 to 2002 in a ryegrass-type festulolium, Paulita, and in a fescue-type festulolium, Hykor. The objectives were to examine the influence of row spacing (12, 24, and 36 cm) and seed rate (8, 12, or 16 kg ha⁻¹) on plant establishment, development, and seed yield. Observations of autumn and spring in-row plant densities indicated satisfactory plant establishment in all combinations of seed rate and row spacing. The number of reproductive tillers was in the range from 800 to 2200 m⁻² in Paulita and from 500 to 1300 m⁻² in Hykor. Row spacing had an effect on the number of reproductive tillers and in both cultivars the highest number was achieved at 12-cm row spacing. Seed yields in the Italian ryegrass type averaged from 1050 to 1150 kg ha⁻¹ and in the tall fescue type from 650 to 800 kg ha⁻¹. Doubling row spacing from 12 to 24 cm had no effect on seed yield in Hykor, while a further increase of row spacing to 36 cm showed a decrease in seed yield. No effect of row spacing on seed yield was observed in Paulita. Neither of the two types was affected in seed yield by seed rate. The fact that row spacing in both types can be increased to 24 cm without having a concomitant negative effect on seed yield implies that mechanical weed control is an optional management technique. Therefore our results have important implications in/for organic grass-seed production.

Keywords:

• Conventional seed production
• fescue
• organic seed production
• reproductive tillers
• ryegrass
• seed yield

Introduction

Festulolium (×Festulolium) is a cross between the two species fescue (Festuca L.) and ryegrass (Lolium L.). The crosses combine in different ways the persistence of fescues with high forage quality of the ryegrasses (Wit, 1959; Wacker et al., 1984; Deleuran & Boelt, 2000; Casler et al., 2002; Kopecký et al., 2006).

It is recognized that a well-established crop with high and regular coverage of soil surface optimizes competitiveness of the seed crop against weeds. If seed production can be performed at wider row spacings than 12 cm, there is potential for mechanical weed control with 1–3 harrowings or 1–2 hoeings (Boelt, 2003) and hence organic production. Festulolium is an interesting crop in organic production systems due to high potential as a forage crop (Deleuran & Boelt, 2000). In 2007 more than 25% of the seed-production area with festulolium was grown organically.

Festulolium is, under Danish conditions, predominantly grown for seed in one year. In 2006, 95% of the fields were 1st fields, 2.1% were 2nd fields, and 3% were 3rd fields (The Danish Plant Directorate, 2007).

In perennial ryegrass, optimum row spacing is 24 or 36 cm, while a further increase to 50-cm row spacing reduces seed yield (Borm, 1998). In trials by Deleuran et al. (2008) no seed-yield reduction was found in tetraploid perennial ryegrass, when increasing row spacing from 12 to 48 cm. In tall fescue row spacings of 12.5 and 25 cm were superior to those of 37.5 and 50 cm (Wander, 1996).

Optimum seed rates have been intensively studied in field trials for various grasses. In perennial ryegrass 3–6 kg ha⁻¹ are needed to obtain high seed yields if the seed weight is 1.5 mg (Nordestgaard, 1977). Four to 6 kg seeds ha⁻¹ (depending on seed weight) are needed to obtain optimum plant density of diploid Italian ryegrass (Lolium multiflorum L.) varieties. In tetraploid varieties, which usually have higher seed weight, optimum seed rate will be 6–9 kg ha⁻¹ when established in a spring cover crop (Nordestgaard, 1984). In tall fescue, 4 kg seeds ha⁻¹ was found to be sufficient to achieve a high seed yield compared to that from 8 kg seeds ha⁻¹ (Wander, 1996).

The objectives of the present experiment were to examine the influence of row spacing and seed rate and their interaction on plant establishment, development, and seed yield in two festulolium cultivars, Paulita and Hykor. Both cultivars were chosen because of their high potential as a forage crop and their concomitant potential use in forage-seed mixtures (Deleuran & Boelt, 2000).

Material and methods

Field experiments were conducted in a conventional cropping system at Aarhus University, Research Centre Flakkebjerg, Denmark (55 19′00″ N, 11 24′00″ E), on a sandy loam soil which contains ~17.5% clay (<2 µm), 25.2% silt (2–63 µm), 55% sand (>63 µm), and 2.3% organic matter. Two (separate) experiments were sown, using the festulolium cultivar Paulita (Italian ryegrass type), and the festulolium cultivar Hykor (tall fescue type). The experimental design was a two-factorial block design with four replicates. Each experiment was sown in the spring of 1999, 2000, and 2001 with spring barley as cover crop.

Three seed rates (8, 12, or 16 kg seeds ha⁻¹) and three row spacings (12, 24, or 36 cm) were tested. Sowing date, nitrogen as calcium ammonium nitrate (N) application date, N-application rate, swathing date, and combine harvesting date of each grass seed crop are shown in Table I. To avoid interaction between crop plant density and competition from weeds, weeds were controlled during crop establishment with herbicides according to good experimental and agricultural practices for conventional cropping systems.

Table I. Dates for sowing, dates and rates for nitrogen (N) applications, and dates for swathing and harvest of the grass seed crops.

	Sowing	N-Application	N Rate (kg ha^−1)	Swathing date	Harvest date
Paulita	17 Apr. 99	20 Mar. 00	100	no swathing	26 Jul. 00
	07 Apr. 00	28 Mar. 01	100	20 Jul. 01	27 Jul. 01
	22 Apr. 01	25 Mar. 02	100	no swathing	15 Jul. 02
Hykor	17 Apr. 99	07 Oct. 99	40
		20 Mar. 00	100	13 Jul. 00	20 Jul. 00
	07 Apr. 00	04 Oct. 00	40
		28 Mar. 01	100	20 Jul. 01	27 Jul. 01
	22 Apr. 01	10 Oct. 01	40
		25 Mar. 02	100	no swathing	15 Jul. 02

Establishment year

For each experiment, corresponding grass and spring barley were sown in rows using a drill equipped with two separate seeding boxes (Villadsen et al., 2002). Spring barley was sown at 120 kg ha⁻¹ in rows 12 cm apart and displaced from festulolium, regardless of row spacing of grass. N at 90 kg ha⁻¹ was applied at sowing to spring barley. The trials did not have any attacks of pests or diseases. Spring barley was harvested with a trial combine and the straw removed immediately after harvest. The stubble was cut to 8–10 cm height and 40 kg N ha⁻¹ was applied to Hykor. In-row plant density was recorded at termination of autumn growth using a variable scale based on visual judgement where 0 equals ‘no plants’ and 100 equals ‘full in-row ground cover’.

Seed-production year

In the seed-production year, plant density was recorded at initiation of spring growth using the same visual variable scale as used in the previous autumn. At the initiation of spring growth 100 kg N ha⁻¹ was applied to both cultivars. One plant sample per plot was cut within an area of 0.25 m² prior to seed harvest and was used to count the number of reproductive tillers. Lodging at flowering and at harvest was recorded using a variable scale based on visual judgement where 0 equals ‘no lodging’ and 100 equals ‘full lodging’. Before shedding, an 8 m×2.5 m area per plot was swathed or harvested directly (in years with full lodging at crop maturity) with a trial combine harvester, and seeds were air-dried to 12% moisture before determining seed yield. The seeds from each plot were cleaned using a commercial cleaning machine. After seed drying and cleaning, one sample from each treatment was analysed for purity using international standardized methodology (ISTA, 1996), and seed yield expressed as 100% clean seed was calculated. Only results from the first seed-production year were recorded. Thousand-seed weight was calculated on the basis of eight individual counts of 100 seeds in each plot.

Statistical methods

The experimental design was a factorial design with two factors (seed rate and row spacing) with four replications. The effects of row spacing and sowing rate and their interaction were analysed using a linear mixed model where the effects were treated SSSas fixed effects. The full model was as shown in Equation (1):

where Y _ijkl is the dependent variable recorded at the i'th row spacing, j'th seed rate, at the k'th harvest year, and the l'th replicate. The Greek letter µ is the general mean, α is the fixed effect of row spacing, and β is the fixed effect of seed rate. Greek letters in brackets are interactions, where, e.g., (αβ ) _ij is the fixed effect of interaction between row spacing and seed rate. If the interaction between row spacing and seed rate was significant, the main effects α _i and β _j were excluded in order to let (αβ) _ij represent the combined effect. C is the random effect of year (k), D is the random effect of the interaction between row spacing (i) and year (k), E is the random effect of the interaction between seed rate (j) and year (k), F is the random effect of the interaction between row spacing (i), seed rate (j), and year (k), G is the random effect of replicate (l) nested within year (k), and H is the random effect of the interaction between row spacing (i), seed rate (j), year (k), and replicate (l). All random effects are assumed to be independently and normally distributed with mean zero and variance , , , , , and , respectively. To test the null hypothesis of no effect of the factors on variables recorded the denominator in the F-tests was calculated according to the random effects of the model. Means for combinations of row spacing and seed rate were separated by pair-wise comparisons at 5% level of significance, if the main effect was significant at that level. All analyses were performed using PROC MIXED within the Statistical Analysis System version 8.02, software package (SAS, 1999). The analyses were performed using PROC MIXED within Statistical Analysis System version 8, software package (SAS, 1999).

Results

The autumn and spring in-row plant density in Paulita and Hykor was in the range 74–88 (on a scale from 0 to 100, where 0 equals no plants and 100 equals full plant coverage). These figures indicate satisfactory plant establishment in seed production. In both cultivars row spacing and seed rate had a significant effect on autumn plant density and an interaction between row spacing and seed rate was found in Paulita (Table II). The spring in-row plant density in Paulita was significantly affected by both seed rate and seed rate and in Hykor only by row spacing (Table II).

Table II. Significance level for the fixed effects of row spacing, seed Rate, and their interaction on autumn plant variable, spring plant characteristics, seed yield, and number of reproductive tillers. The significance levels 0.001, 0.01, 0.05, and not significant have been indicated with ***, **, *, and ns, respectively.

Cultivar and factor	Autumn in-row density	Spring in-row density	No. reproductive tillers	Seed yield
Paulita
Row spacing (R)	**	*	*	ns
Sowing rate (S)	*	***	ns	ns
R*S	**	Ns	ns	ns
Hykor
Row spacing (R)	*	***	**	***
Seed rate (S)	***	***	ns	ns
R*S	ns	Ns	ns	ns

The number of reproductive tillers was between 800 and 2200 m⁻² in the Italian ryegrass type Paulita and from 500 to 1300 m⁻² in the tall fescue type Hykor. Row spacing had a significant effect on the number of reproductive tillers and in both cultivars the highest number was achieved at 12-cm row spacing (Tables II and III). No difference was observed in the number of reproductive tillers in either cultivar at 24- or 36-cm row spacing (Table III). Hykor had a significant, positive correlation (R=0.36***) between seed yield and number of reproductive tillers, while the number of reproductive tillers in Paulita was significant negatively correlated with both autumn in-row plant density (R= − 0.53***) and spring in-row plant density (R= − 0.51***) (data not shown).

Table III. Number of reproductive tillers per m² and seed yield (kg ha⁻¹) in Hykor and Paulita at three row pacings. Different letters within data for a cultivar indicate significant (P≤0.05) difference.

	Row spacing (cm)
		12	24	36
Number of reproductive tillers	Hykor	1236a	570b	506b
	Paulita	2078a	1062b	843b
Seed yield	Hykor	777a	755a	683b
	Paulita	1089a	1086a	1097a

Per cent lodging at flowering and at harvest was not significantly affected by row spacing or seed rate in any of the cultivars (data not shown).

Seed yields in the Italian ryegrass type Paulita averaged 1050–1150 kg ha⁻¹ and in Hykor 650–800 kg ha⁻¹. The highest yields in both cultivars were achieved in 2000, while there was no significant, mutual difference between yields in 2001 and 2002 (data not shown). Only Paulita showed a significant effect of calendar year on the seed yield. Average seed yield in Paulita was 1326, 1010, and 931 kg ha⁻¹ in 2000, 2001, and 2002, respectively. Only Hykor had a significant effect of row spacing on seed yield. Doubling row spacing from 12 to 24 cm had no significant effect on seed yield in the tall fescue type Hykor, while tripling row spacing to 36 cm showed a significant decrease in seed yield (Table III). No effect of row spacing was observed in Paulita (Table III). Neither Paulita nor Hykor was affected in the seed yield by seed rate.

Discussion

Row spacing in Paulita and Hykor can be increased to 24 cm without having a concomitant negative effect on seed yield. A drawback from doubling row spacing from 12 to 24 cm is a reduced number of reproductive tillers per area. Growing at 24-cm row spacing allows mechanical weed control and therefore results can be adopted in organic production systems.

In-row plant density in autumn was highest at wide row spacing in both cultivars. This indicates that a higher number of tillers in the row were present at the widest row spacings. This situation persisted until the onset of growth in the following spring and this was reflected in spring in-row plant density. The many tillers resulted in increased competition for especially light and space, and self-thinning most likely started to occur in the row. Self-thinning is a major feature of experimental and natural populations as individuals increase in size (Kays & Harper, 1974). Generally, as tiller production increases as a result of increasing seed rate or row spacing, the percentage of tillers which become fertile reduces (Meijer, 1984).

Perennial ryegrass is characterized by a strong compensational effect between seed-yield components, where a low number of reproductive tillers can be compensated for by a higher number of seeds per reproductive tiller and/or a higher seed weight (Hebblethwaite & Ivins, 1977; Nordestgaard, 1979; Young et al., 1996; Gislum & Boelt, 1998). This is supported by a number of experiments where no significant positive correlation is found between the number of reproductive tillers and seed yield in perennial ryegrass (Young et al., 1996; Gislum & Boelt, 1998). The results in the present experiment are in agreement with these findings as the decrease in number of reproductive tillers from doubling of row spacing to 24 cm did not have an impact on seed yield in Paulita. From the number of reproductive tillers, the average seed yield, and the missing significant correlation between the number of reproductive tillers and seed yield the ryegrass type Paulita showed characteristics similar to other ryegrass varieties.

The positive correlation between the number of reproductive tillers and seed yield in the tall fescue type Hykor is in good agreement with former experiments performed in tall fescue (Chastain & Grabe, 1989; Deleuran & Boelt, 2005) and in red fescue (Boelt, 1998). In red fescue and tall fescue Hampton & Fairey (1997) suggested a reproductive tiller number of 1500–3000 and 600–900 m⁻², respectively, to be required for economically feasible seed production. For tall fescue grown under Danish field conditions tiller numbers above 900 m⁻² still increase seed yield (Deleuran & Boelt, 2005). The average number of reproductive tillers at 12-cm row spacing was 1236 m⁻², which should be sufficient to obtain higher seed yield. However, at 24–36-cm row spacing the reproductive tiller number in Hykor at harvest was 500–600 m⁻², which is a rather small and insufficient number of tillers and in this case assumedly the limiting factor for the potential seed yield.

In tall fescue it was not possible to define an optimum N-application rate, defined as the N-application rate which gave the highest seed yield, when N-application rate was in the range 90–210 kg ha⁻¹ (Young et al., 1999), while Fairey & Lefkovitch (1998) found that soil-available N should be in the range 100–150 kg ha⁻¹ to maximize seed yield. Based on these findings a total of 140 kg N ha⁻¹ was applied in the present experiment. However, it has been recognized that optimum N-application rate is 187 kg ha⁻¹ to fully utilize seed-yield potential in Danish tall fescue seed production (Gislum et al., 2006). The reason for lower seed yield in Hykor than in Paulita might therefore be due to N-application rates below optimum. In particular, it is assumed that autumn N-application rates higher than 40 kg N ha⁻¹ will increase the number of reproductive tillers.

The present study shows that spring in-row plant density in Paulita was significantly affected only by seed rate and in Hykor only by row spacing. This effect on plant development was followed by a significantly higher number of reproductive tillers at 12-cm row spacing in both types. The three seed rates had no effect on seed yield in either Paulita or Hykor. Yet, the three row spacings had no effect on seed yield in Paulita, while the optimum row spacing in Hykor was 12 or 24 cm. The wider row spacings allow for mechanical weed control, and hence adaptation of the present results in organic production.

Acknowledgements

The authors thank the DJF technical team at research group crop ecology and product quality for their very skilled assistance. This study was funded by ICROFS (International Centre for Research in Organic Food Systems) http://www.icrofs.org.