Autumn and spring management in seed crops of meadow fescue (Festuca pratensis Huds.)

Abstract

Different procedures for managing stubble and regrowth in meadow fescue (Festuca pratensis Huds.) seed crops were examined in two experimental series in southeast Norway. The first series investigated cutting and removal of stubble shortly after seed harvest in late July/early August, combined with cutting of regrowth (forage harvest) in September or October or burning of the wilted aftermath in early spring. Except for one crop with more than 30-cm stubble, stubble removal shortly after seed harvest did not improve seed yield in the following year. On average for plots with and without stubble removal in seven seed crops, forage harvest on 5 September or 1 October reduced seed yield by 9 and 12%, respectively. The reductions were due to smaller inflorescences, probably reflecting lower carbohydrates reserves. In most trials, the highest seed yield, on average 9% above that of the uncut and unburned control, was found after burning in spring. The second experimental series investigated flail-chopping in spring as an alternative to burning, and the effect of delaying either of the two treatments. On average for four trials, burning and flail-chopping before growth initiation increased seed yield by, in turn, 20 and 12% compared with the untreated control. A two-to-three-week delay in fail-chopping had no negative impact, but a two-week delay in burning reduced seed yield back to the uncut/unburned control level. Based on these trials, growers are recommended to burn meadow fescue seed crops in early spring rather than removing stubble and regrowth in autumn. If burning in early spring is not possible, flail-chopping is recommended within two weeks after growth initiation.

Keywords:

• Autumn cutting
• flail-chopping
• seed yield
• spring burning
• winter damage

Introduction

Next to timothy (Phleum pratense L.), meadow fescue (Festuca pratensis Huds.) is the most important species in Norwegian herbage-seed production. Each year 600–800 ha of this species are harvested for seed in the south-eastern part of the country. While seed of meadow fescue is produced both along the Norwegian south coast (58 °N) and in continental areas north of Oslo (61 °N) the largest acreage is found on both sides of the Oslo fjord (59 °N) (Figure 1).

Figure 1.  Map of south Norway showing the experimental sites in the two series. The meteorological stations Kise, Ramnes, Rygge, and Kjevik have also been indicated; see Table I.

In perennial seed crops it is a general experience that cutting and removal of stubble, straw residues and eventual regrowth in autumn will increase the reproductive tiller number in the subsequent year. This may either be because of enhanced tillering, due to better light penetration to plant bases in autumn (Chilcote et al., 1980), or because of less competition for light and thus better survival of reproductive tillers in spring (Meijer & Vreeke, 1988).

Internationally, seed-production trials in meadow fescue have shown diverging results with regard to optimal autumn management. Whereas cutting of stubble and regrowth in late September or October tended to be advantageous in Danish (Thøgersen, 1972; Nordestgaard, 1976) and British experiments (Roberts, 1965), no positive effect of autumn cutting was observed by Cedell (1965) under Swedish conditions.

In Finland, Köylijärvi (1986) found that seed yields of meadow fescue could be reduced by autumn cutting if winters were severe. This is partly in agreement with a Norwegian study showing that forage harvest at the end of the growing season (10 October) reduced winter survival and seed yield at coastal locations with unstable snow cover, but was favourable to seed yield at an inland location with more stable snow cover (Havstad, 1998). The highest seed yield at the coastal location was found on plots where the stubble was removed immediately after seed harvest, with no subsequent cutting in autumn. The possibility of delaying stubble and regrowth removal from autumn to the following spring was not investigated.

The objective of the present research was to investigate optimal stubble and regrowth management of meadow fescue seed crops in autumn and spring.

Materials and methods

Experimental series I. Autumn cutting vs. spring burning

During 1998–2001 six field trials were established in seed crops of meadow fescue cv. Fure (three trials) and cv. Salten (three trials) at locations ranging from Landvik (58 ^oN) on the Norwegian south coast to Hedmark (61 ^oN) in the inland north of Oslo (Figure 1). Except for the southernmost trial, which was located at Bioforsk Landvik Research Station, all trials were located in commercial seed crops on seed growers’ farms. The experiments were laid out in first-year seed crops in early to mid August, not more than two weeks after seed harvest.

The seed crop straw was always removed before experiments were laid out, the height of remaining stubble varying from 13 cm (Hedmark 1998) to 33 cm (Vestfold 1998). Plot size was 8.0 m×1.5 m.

The experimental plan had two factors with the eight treatment combinations being completely randomized within each of three replicates (blocks):

Factor 1. Stubble removal within two weeks after seed harvest/straw removal

1.	No stubble removal.
2.	Stubble cut to 5–7 cm with a reciprocal blade mower and raked off the field.

Factor 2. Forage harvest in autumn vs. burning of wilted aftermath in spring

A.	No harvest or burning of aftermath.
B.	Forage harvest (sward cut to 5–7 cm with a reciprocal blade mower) on 5 September.
C.	Forage harvest (sward cut to 5–7 cm with a reciprocal blade mower) on 1 October.
D.	Aftermath (wilted plant material) burned in early spring.

All plots were fertilized with 50 kg N h⁻¹ just after trial establishment (i.e., in early to mid August), and 60–100 kg N ha⁻¹ in spring (April or early May) in the form of compound NPK fertilizer (25-2-6).

At forage harvest in autumn, plot yield was weighed and a sample (ca. 1 kg) taken for determination of dry-matter content and forage quality. Per cent dry matter (DM) was determined in each sample after drying at 60 ^oC for 48 h. Pooled samples (across replicates) from each treatment combination cut in autumn (1B, 2B, 1C, and 2C) were analysed for forage quality using the NIRS-analyser (Infraalyzer 500, BRAN + LUEBBE, Norderstedt, Germany) at the forage laboratory, Hellerud experimental farm. Crude protein concentration and milk fodder units (FUm) per kg dry matter were used to estimate forage quality. The latter expresses how efficiently the energy in forage can be utilized for milk production (Skjevdal et al., 1992).

Burning of aftermath (wilted regrowth) in early spring was always carried out before any signs of spring growth. Dates of spring burning varied from 24 March at the southern coastal location Landvik to 26 April at the inland location Hedmark.

Registrations in the seed harvest year were as follows:

1.	Per cent plant cover of meadow fescue in spring (visual assessment of the whole plot in May when the sward had grown to a height of about 10 cm).
2.	Per cent lodging (visual assessment of the whole plot) at flowering and seed harvest.
3.	Panicle numbers as counted in mobile frames (area varying from 0.25 to 0.4 m2).
4.	Weight per unthreshed inflorescence (based on 100 random panicles that were cut 1 cm below the inflorescence less than one week before seed harvest).

The trials were harvested directly with Nurserymaster Field Plot combines at 20–30% seed moisture content. Harvest dates varied from late July at Landvik and Vestfold to early/mid August at Hedmark. The five on-farm trials were harvested for seed only in one year (second ley year). In the trial at Bioforsk Landvik Research Station, experimental treatments were repeated in the autumn of the second ley year, and this trial was therefore harvested for seed in two consecutive years.

The plot yields were dried to 10–12% seed moisture content and cleaned on a laboratory air-screen machine (LALS, Westrup, Denmark). Representative seed samples, pooled across replicates in each trial, were taken from each of the eight treatment combinations and analysed for purity and thousand-seed weight in the seed laboratory at Bioforsk Landvik Research Station. The determination of thousand-seed weight was based on counts of 100 seeds multiplied by 10.

Experimental series II. Timing of cutting or burning in spring

Four on-farm trials were laid out in early spring 2002, 2003, or 2004 on both sides of the Oslo fjord (Figure 1). The trials were laid out in second-year crops of cv. Fure (three trials) or cv. Salten (one trial). Until the start of experimentation, all crops had been managed according to recommended practice, including straw removal and application of 30–40 kg N ha⁻¹ shortly after the previous year's seed harvest, but stubble or regrowth had not been cut in the previous autumn. The length of wilted regrowth and stubble, measured from plant base to the tip of the longest leaves at three random sites within each control plot, varied from 10 cm (Østfold 2004) to 35 cm (Vestfold 2004).

The experimental plan comprised eight treatments:

1.	No cutting or burning in spring (control).
2.	Wilted regrowth and stubble cut with a reciprocal blade mower to 5 cm and raked off the field before start of growth in early spring.
3.	Wilted regrowth and stubble flail-chopped to 5 cm using a tractor-driven chopper before start of growth in early spring. Chopped material not removed.
4.	Similar to treatment 3, but 14 days later.
5.	Similar to treatment 3, but 24 days later.
6.	Wilted regrowth and stubble burned before start of growth in early spring.
7.	Similar to treatment 6, but 14 days later.
8.	Similar to treatment 6, but 24 days later.

The experimental design was a randomized complete block with three replicates.

The first cutting or burning treatments (2, 3, and 6) took place before any visible sign of new tiller growth. Dates varied from 30 March to 6 April. On the second (treatments 4 and 7) and third (treatments 5 and 8) dates of cutting/burning, new tillers had emerged and reached an average length (measured from plant base to the tip of the longest leaves) of 4 and 10 cm, respectively.All plots were fertilized with 60–90 kg N ha⁻¹ in early to mid April in the form of compound NPK fertilizer (25-2-6). In two of the four trials, plots were growth-regulated with trinexapac-ethyl (Moddus, 0.15 kg active ingredient (a.i.) ha⁻¹) at Z 31.

Field registrations, seed harvest, and seed cleaning were performed as described for Experimental series I.

Statistical analyses

In both experimental series, analyses of variance were first performed individually for each seed crop. As the individual analyses in Experimental series I suggested different seed-yield responses to forage harvest in Hedmark than at the other experimental sites, a statistical model with experimental site (Hedmark vs. others) as one of the independent variables was contemplated. However, as this model showed no significant interaction between experimental site and stubble or regrowth treatments, we ended up with a collective analysis regarding each crop (site×year) as one random replicate from the population of Norwegian seed crops of meadow fescue. The analyses were conducted according to PROC ANOVA (SAS Institute, 1990). Results from individual crops will be presented only for seed yield which is considered the most important character in seed production. For other characters, only mean values across all crops in each series will be presented. While the term ‘significant’ in this paper always refer to P <0.05, exact P-values in the range 0.05 < P <0.15 will also be given to indicate tendencies.

Meteorological data

Within the Norwegian seed-production district, there is a climatic gradient with longer growing seasons, higher temperatures, and more rainfall in the south than in the north (Table I). Although winter temperatures during the course of these trials were always above the 30-year normal values (Table I), snow cover was generally more stable in Hedmark than in Vestfold/Buskerud, and especially when compared with Landvik on the south coast (Figure 2).

Figure 2.  Daily mean (⋄) and minimum air temperature (−) (°C), and depth of snow cover (cm) at the meteorological stations a) Kjevik during the autumn and winter 2000–2001, b) Ramnes during the autumn and winter 1998–1999, and c) Kise during the autumn and winter 1998–1999.

Table I. Temperature and precipitation at the meteorological stations Kjevik, Ramnes, and Kise relevant to Experimental series I (Ramnes and Rygge relevant to Experimental series II). Normal values refer to the period 1961–1990. (See Figure 1 for geographical locations.)

		Mean temperature (°C)					Precipitation (mm)
		Sep – Nov	Dec – Feb	Mar – May	Jun	Jul	Sep – Nov	Dec – Feb	Mar – May	Jun	Jul
Experimental series I
Kjevik	1999–2000	9.0	1.5	6.9	12.9	15.8	414	446	264	53	133
	2000–2001	9.6	−0.1	4.8	13.6	17.0	919	495	237	60	54
	Normal	7.5	−1.2	5.2	14.0	15.5	469	317	232	75	88
Ramnes	1998–1999	4.5	−2.4	4.6	13.4	16.6	338	185	184	182	79
	2000–2001	8.4	−3.0	4.0	13.7	16.8	883	125	206	48	56
	Normal	5.8	−4.0	4.6	14.5	15.8	374	226	191	70	90
Kise	1998–1999	4.2	−4.7	3.6	12.9	16.0	185	108	177	132	66
	1999–2000	7.7	−3.2	5.0	12.4	14.7	200	82	127	71	68
	Normal	4.6	−6.9	2.5	13.6	15.2	177	102	105	59	66
Experimental series II
Ramnes	2002–2003	–	–	6.9	15.0	16.3	–	–	185	54	209
	2003–2004	–	–	5.2	15.4	17.5	–	–	70	71	105
	2004–2005	–	–	6.3	13.5	14.3	–	–	56	124	94
	Normal	5.8	−4.0	4.6	14.5	15.8	374	226	191	70	90
Rygge	2003–2004	–	–	6.9	14.0	15.2	–	–	103	83	70
	Normal	6.3	−3.6	4.7	14.7	15.9	287	164	154	63	73

Results

Experimental series I: Autumn cutting vs. spring burning

Yield quality of forage harvested in autumn

Removal of stubble shortly after seed harvest significantly reduced forage yields in autumn. On average for two cutting dates (1 September and 5 October), the dry-matter yield was more than twice as high on plots where the stubble had been left intact after seed harvest compared with plots where the stubble had been removed (Table II).

Table II. Main effects of stubble removal after seed harvest and harvest date in autumn on dry-matter yield (kg ha⁻¹), the percentage of crude protein in forage dry matter (DM), and the concentration of milk fodder units in forage DM (FU_m kg⁻¹ DM). Mean of seven seed crops, Experimental series I.

		Forage quality
	Dry-matter yield (kg ha^−1)	% Crude protein	Milk fodder units
Main effect of stubble removal
1. No stubble removal	2280	14.8	0.77
2. Stubble cut to 5 cm	1060	18.5	0.84
Significance (P-value)	<0.01	<0.01	<0.01
Main effect of cutting date
B. 5 September	1490	17.9	0.82
C. 1 October	1840	15.4	0.78
Significance (P-value)	0.02	0.05	0.04

Delaying the forage harvest from 5 September to 1 October increased dry-matter yield in all trials. The highest yield at the latest date of cutting, 2684 kg DM ha⁻¹, was recorded in Vestfold in 1998, and the lowest, 929 kg DM ha⁻¹, in Buskerud in the same year. On average for all trials and different stubble treatments, dry-matter yield increased by about 24% when date of cutting was delayed from 5 September to 1 October (Table II).

On average for the two harvest dates, protein and energy concentrations in plant DM were significantly higher on plots where stubble had been removed after seed harvest than on plots without stubble removal. Likewise, forage quality was significantly reduced when harvest was delayed from 5 September to 1 October (Table II).

The interactions between stubble treatment and forage harvest date were not significant for forage yield or quality characteristics.

Seed yield

As a main effect for all seed crops, stubble removal after seed harvest did not have any significant effect on seed yield in the following year. Only in the trial in Vestfold in 1998–1999 was seed yield significantly reduced by not removing stubble (Table III).

Table III. Main effects of stubble removal after seed harvest and cutting/burning in autumn/spring on seed yield (kg ha⁻¹, 12% water, 100% purity) in each seed crop and for the whole material in Experimental series I.

	Seed yield (kg ha^−1)
	Busk. 98–99	Hedm. 98–99	Vestf. 98–99	Hedm. 99–00	Landv. 99–00	Landv. 00–01	Vestf. 00–01	Mean of seven seed crops
Main effect of stubble removal
1. No stubble removal	545	788	968	636	776	734	883	761
2. Stubble cut to 5 cm	489	794	1053	651	744	584	925	749
Significance (P-value)	0.20	>0.20	0.03	>0.20	>0.20	0.17	>0.20	>0.20
Main effect of regrowth treatment
A. No cutting of regrowth	547	771	976	631	821	780	925	779
B. Cutting on 5 September	489	772	991	620	715	479	885	707
C. Cutting on 1 October	440	775	911	622	682	482	886	686
D. Burning in early spring	590	847	1165	702	823	895	921	849
Significance (P-value)	<0.01	>0.20	<0.01	>0.20	<0.01	<0.01	>0.20	<0.01
LSD, 0.05	62	-	97	-	60	74	-	81

On average for stubble treatments, forage harvest on 5 September and 1 October reduced seed yield compared with uncut plots in all trials except at the northernmost location Hedmark (Table III). On average for all crops, the seed-yield reductions caused by forage harvest on 5 September or 1 October were 9% (almost significant) and 12% (clearly significant), respectively. The strongest negative impact, on average 38%, was found in the third-year seed crop at Landvik in 2000–2001 where the same treatments had been repeated for two consecutive years (Table III).

The highest seed yield in most trials was found on plots that had been burned in early spring. As an overall average, the 9% seed-yield increase compared with uncut and unburned plots was almost significant (Table III).

Interactions between stubble- and regrowth treatments were not significant with regard to seed yield, either in individual trials or for the material as a whole. The highest seed yield, regardless of stubble-removal treatments, was found on plots being burned in early spring (Figure 3).

Figure 3.  Effect of stubble treatment after harvest and cutting/burning in autumn/early spring on seed yield (kg ha⁻¹) in seed crops of meadow fescue. Mean of seven crops harvested in Experimental series I.

Plant cover in spring

As a main effect, stubble removal shortly after seed harvest had no effect on plant cover in the following spring (Table IV). By contrast, the main effect of autumn cutting vs. spring burning was significant as the least plant cover in spring was observed on plots which had been cut on 1 October (Table IV). For this character there was also a significant (P =0.02) interaction as forage harvest in autumn was only detrimental if the stubble had been removed shortly after seed harvest (Figure 4).

Figure 4.  Effect of stubble treatment after harvest and cutting/burning in autumn/early spring on the coverage (%) of meadow fescue in spring. Mean of seven crops harvested in Experimental series I.

Table IV. Main effects of stubble removal after seed harvest and cutting/burning in autumn/spring on per cent plant cover in spring, weight per unthreshed inflorescence (mg), and the number of panicles per m². Mean of seven seed crops, Experimental series I.

Treatment	Per cent plant cover in spring	Weight per unthreshed inflorescence (mg)	Panicles per m^2
Main effect of stubble removal
1. No stubble removal	76	233	788
2. Stubble cut to 5 cm	76	221	791
Significance (P-value)	>0.20	<0.01	>0.20
Main effect of regrowth treatment
A. No cutting of regrowth	77	238	747
B. Cutting on 5 September	76	216	775
C. Cutting on 1 October	72	215	733
D. Burning in early spring	79	238	902
Significance (P-value)	<0.01	0.04	0.04
LSD, 0.05	3	21	119

Lodging

Lodging at flowering or at seed harvest was never significantly affected by treatments (data not shown). On average for the whole material, lodging increased from 68% at flowering to 89% at seed harvest. Interactions were also not significant for these characters

Panicles per m²

As a main effect for all seed crops, stubble removal shortly after seed harvest did not significantly influence panicle density in the following year (Table IV). The significantly highest panicle number, on average 21% higher than on uncut and unburned control plots, was found on plots that had been burned in early spring (Table IV). The interaction between stubble and regrowth treatments was not significant.

Inflorescence weight

On average for all seed crops, cutting of stubble after seed harvest significantly reduced inflorescence weight by 5% in the following year (Table IV).

Forage harvest in autumn also reduced inflorescence weight compared with uncut plots. Burning in spring had no effect on this character compared with the control treatment, and the interaction between stubble and regrowth treatments was not significant.

Thousand-seed weight

The thousand-seed weight was not significantly affected either by any of the experimental factors (data not shown). On average for all treatments, thousand-seed weight was 2424 mg.

Experimental series II. Timing of cutting and burning in spring

Seed yield

Although differences in the individual trials were not always significant, cutting or chopping, regardless of date (treatments 2–5), and burning at the earliest date in spring (treatment 6) usually increased seed yield compared with the uncut and unburned control. The highest seed yield, on average 20% above the control, was found on plots where stubble and wilted regrowth was burned at the earliest date in spring (treatment 6). A two-week delay in burning reduced seed yield back to the control level (Table V).

Table V. Effects of various spring treatments on the number of reproductive tillers per m², weight per untreshed inflorescence (mg), and seed yield (kg ha⁻¹, 12% water, 100% purity) in Experimental series II. The number of trials is given in parenthesis.

			Seed yield (kg ha^−1)
Treatment	Panicles per m2 (4)	Weight per untreshed inflorescence (mg) (4)	Vestfold 2002 (1)	Vestfold 2003 (1)	Vestfold 2004 (1)	Østfold 2004 (1)	2002–2004 (4)
1. No cutting or burning in spring (control)	690	187	532	136	595	324	397
2. Wilted regrowth and stubble cut and removed in early spring	751	176	652	204	666	294	454
3. Wilted regrowth and stubble flail–chopped in early spring	702	176	644	167	639	333	446
4. Similar to treatment 3, but 14 days later	676	164	624	204	661	346	459
5. Similar to treatment 3, but 24 days later	629	165	625	186	549	337	424
6. Wilted regrowth and stubble burned in early spring	726	186	557	204	771	376	477
7. Similar to treatment 6, but 14 days later	732	168	576	113	523	366	394
8. Similar to treatment 6, but 24 days later	692	166	496	126	622	286	383
Significance (P-value)	>0.20	0.13	0.05	<0.01	<0.01	>0.20	0.08
LSD, 0.05	–	–	104	45	81	–	–

Seed-yield components

Early cutting, flail-chopping, or burning had no significant effect on panicle number. The weight per inflorescence tended to be lower on most plots cut, chopped, or burned in spring except those burned on the earliest date (Table V). Spring treatments had no effect on thousand-seed weight, which, on average for all trials and treatments, was 2215 mg (data not shown).

Lodging at flowering and seed harvest

Lodging scores were not significantly different at any of the registration dates (data not shown).

Discussion

Improved light penetration to tiller bases, and thereby stimulated tiller development (size and number) in autumn, was probably the main reason why removal of stubble shortly after seed harvest had a positive effect on seed yield in the trial at Vestfold in 1998–1999 (Experimental series I), where a long stubble (>30 cm) had been left after harvest (Chilcote et al., 1980). In most of the other trials stubble height after harvest was less than 15 cm, and thus the effect of stubble removal was either absent or insignificant (Table III). Also, in straw-residue management experiments, both in meadow fescue and timothy, Havstad (2007) found the strongest positive impact of stubble removal (flail-chopping or field burning) in those crops which had been left with the longest stubble at harvest. By contrast, red fescue seed production benefitted from removal of stubble that was only 12–15 cm high (Aamlid et al., 2005).

Forage harvest in autumn, especially in October, had a strong negative influence on subsequent seed yield in many trials. The strongest negative impact occurred at the southernmost location, Landvik, which had the most unstable snow cover (Figure 2c). In such winter climates, removal of the insulating cover of stubble and regrowth before winter may lead to greater temperature fluctuations at ground level (i.e., around the grass apices) and thus increased damage by frost (Havstad, 1998). This also explains the general reduction in plant cover in spring on plots cut on 1 Oct. (Figure 4). A reduction in winter survival after late cutting in autumn has been described in forage production of various grasses, e.g., meadow fescue and timothy (Larsen, 1994) and perennial ryegrass (Nadeau et al., 1997).

Besides frost damage to tiller apices, cutting before growth cessation in autumn most likely also reduced the plants’ reserves of carbohydrates in stem bases. Although the interaction between stubble and regrowth treatments was not significant with regard to seed yield, the reduction of carbohydrate reserves may well have been most severe where removal of stubble, including some leaf area, was combined with forage harvest on 5 September, i.e., about one month later. In a perennial-ryegrass-dominant sward, Lee et al. (2009) found up to 48% less water-soluble carbohydrates in stem bases after frequent defoliation treatments in autumn. In coastal areas of south Norway, tillering in meadow fescue usually continues well into October (Havstad & Aamlid, 2007), and in the present trials such late, but for seed production mostly wasteful, tillering was probably stimulated.

The fact that the seed-yield reductions after these treatments were due to lower weight per inflorescence (Table IV) was probably also a direct reflection of lower carbohydrate reserves. The findings are in general agreement with those of Niemeläinen (1990), who reported that depletion of carbohydrates in autumn had stronger negative implications for the plants’ reproductive ability than for their capability for vegetative growth.

Regardless of experimental site, the highest seed yield was found on plots that had been burned in early spring. The main advantages of delaying regrowth removal from autumn to early spring is that the insulating cover of leaves around the apices is preserved during winter and that plants maintain a better carbohydrate status. In addition, when the wilted leaves are removed by burning in early spring, the competition for light among tillers is strongly reduced. Thus, more of the tillers that were primary induced (Heide, 1994) in the previous autumn will be able to survive and become reproductive (Table III). A possible fourth advantage of spring burning is that the black ashes will promote energy absorption and thus lead to higher soil temperature and faster green-up.

In Experimental series II, burning at the earliest date in spring (treatment 6) was superior to other stubble/regrowth treatments in all trials except at Vestfold in 2002 (Table V). Although plots being burned in early spring produced 5% higher seed yield than did untreated control plots in this trial, the positive effect on seed yield tended to be less than when the wilted aftermath was cut or chopped on the same date. Although no visible growth of new tillers was observed on the relatively late date of burning (6 April), some apices may well have been under development and thus injured by the flames. This explanation is substantiated by the results from another trial in meadow fescue in Vestfold in the same year (2002), in which spring burning on 22 March (i.e., 14 days earlier than in Experimental series II) increased seed yield by 25% compared with untreated control plots (Havstad et al., 2003). This suggests that there is a critical time in spring after which burning should be avoided. In most years the two last weeks of March seem to be a safe period for spring burning in meadow fescue seed crops in coastal areas in SE Norway.

Besides rendering the developing apices more vulnerable to heat damage the practical efficiency of field burning will normally decrease as the growth of green leaves progresses in spring. A less efficient burning, hampered by plant growth in spring, was especially seen in late April (treatment 8) in Østfold in 2004, when the length of new tillers was measured to be about 20 cm.

Therefore, although early spring burning is usually the optimal method to maximize seed yield (Table V), mechanical management of the wilted stubble and regrowth seems to be an acceptable alternative for seed growers unwilling to burn, or as an emergency method in years when burning has not been possible within the critical time frame due to unfavourable weather conditions. The most efficient mechanical method seems to be chopping with a tractor-mounted flail-chopper. As long as the chopped material is evenly spread, removal from the field is not necessary after chopping.

Acknowledgements

Thanks are extended to Trygve S. Aamlid for valuable comments on the manuscript and to the staff at Bioforsk Landvik Research Station and the farmers' experimental and advisory groups in Vestfold, Buskerud, Østfold, and Hedmark for skilful implementation of these experiments.