Straw management and optimal N fertilization in seed production of timothy (Phleum Pratense L.) and Meadow Fescue (Festuca Pratensis Huds.)

ABSTRACT

The market for herbage seed straw has diminished in many seed-production areas due to less livestock. Seed growers are therefore looking for alternatives to straw removal, which up to now has been the most common practice. During 2000–2006, different alternative straw chopping methods, both at the back of the combiner and with a tractor-mounted flail-chopper, and field burning strategies were evaluated in seed crops of timothy (Phleum pratense) and meadow fescue (Festuca pratensis) in southeast Norway. The requirement for an extra N input in autumn (30–40 kg ha^-1) when practising straw chopping was also examined. Compared to straw removal, straw chopping at the back of the combiner during seed harvest did not reduce seed yield in the following year as long as stubble height was low (<10 cm in timothy) and the straw spread uniformly in the field. On average, seed yield was 1–4% and 1–9% higher compared to straw removal in timothy and meadow fescue, respectively. If the chopped straw was spread unevenly, or long stubble was left at combining, it is recommended to use a tractor-mounted flail-chopper after harvest. The experiments did not provide any support for an extra input of nitrogen in autumn, either in timothy or meadow fescue, when the straw was chopped rather than removed. Burning of stubble and straw soon after seed harvest was another efficient clean-up method after harvest, which increased seed yield 9–15% and 17–20% compared to straw removal in the two species, respectively. However, as the burning method is risky and causes smoke emissions, it is normally not recommended. It is concluded that for most seed growers, the most effective, least laborious and most environment-friendly alternative to straw removal will be to chop the straw at the back of the combiner during seed harvest.

KEYWORDS:

• Burning
• flail-chopping
• nitrogen
• residue removal
• seed yield
• stubble height
• straw removal

Introduction

With an annual harvest area of about 2000 and 600 ha respectively timothy (Phleum pratense L.) and meadow fescue (Festuca pratensis Huds.) are the two main species in Norwegian herbage seed production. For both species, the majority of seed growers remove the bale and grass straw from the field soon after seed harvest. However, the demand for straw to animal feed or bedding has declined in recent years and many seed growers have difficulties with straw disposal. Usually growers receive no compensation for the straw and often have to pay contractors 1000–2000 NOK ha^-1 for straw removal.

Besides the expenditure and difficulties with handling and placement of the bales, straw removal takes away valuable organic matter which in the long run would have enhanced soil productivity. Although conflicting views occur (Regina et al. 2015), most European long-term studies have shown incorporation of cereal straw to increase the organic carbon content in soil (e.g. Powlson et al. 1987; Curtin & Fraser 2003).

The most efficient alternative to straw removal is either to chop the straw directly during combining or to burn the straw soon after seed harvest. A third alternative is to chop both straw and stubble with a tractor-mounted straw chopper soon after harvest. Except for straw burning after seed harvest of meadow fescue (Nordestgaard 1981) and timothy (Entz et al. 1994) alternative methods of straw management has not been investigated previously in these two species.

Chopping of straw means return of organic material that has to be degraded by microbes. During the degradation process a temporary shortage for nitrogen may occur (Hadas et al. 2004). Up to now, the recommended autumn fertilization in Norwegian seed crops of timothy and meadow fescue has been 0 and 20–30 kg N ha^-1, respectively (Havstad 2015a, 2015b), but these rates may have to be increased if the straw is going to be chopped and returned.

The objective of the project was to evaluate various straw management methods in seed crops of timothy and meadow fescue. An additional aim was to examine whether straw chopping has consequences for optimal fertilization of the two seed crops.

Materials and methods

Experimental series I. Straw removal vs. straw chopping or field burning

During 2000–2003 seven field trials were established in seed crops of timothy (five trials in cv. Grindstad and one trial in cv. Vega) and meadow fescue (one trial in cv. Fure). All trials were located in southeast Norway, along the coastline from Landvik (58.2°N) to Vestfold (59.4°N) (Figure 1). Except for the southernmost trial at NIBIO Landvik Research Station, all trials were laid out in commercial seed crops on seed growers’ farms.

Figure 1. Map of south Norway showing location of trials in meadow fescue (MF) and timothy (T).

The trials were established at seed harvest of first-year seed crops in late July (meadow fescue) or mid-August (timothy). At establishment, the plot was orientated perpendicularly to the harvest direction of the combiner. Plot length varied from 7.3 to 11.0 m (equal to three cutting widths with the combiner). Plots width was always 3.0 m.

In all timothy trials, plots were combined twice, with straw chopping conducted at the second combining when straw had been dried in windrows for 4–7 days after the first combining. In the meadow fescue trial in Vestfold (2002), the straw was chopped at direct combining.

The height of the stubble at combining varied from 7 to 25 cm (Table 1).

Table 1. Detailed information about six trials in timothy and one trial in meadow fescue seed crops during 2000–2003. Experimental series I.

Location	Timothy trials						M. fescue
	Vestfold	Aust-Agder	Landvik	Vestfold	Vestfold	Landvik	Vestfold
Year	2000–2001	2000–2001	2000–2001	2001–2002	2002–2003	2002–2003	2002–2003
Cultivar	Grindstad	Grindstad	Grindstad	Vega	Grindstad	Grindstad	Fure
Soil classification	Silty loam	Clay loam	Silty loam	Silty loam	Silty loam	Silty loam	Clay soil
Autumn (trial establishment year)
Date for straw chopping at combining (treatm, 3–6)/trial establishment	13 Aug.	18 Aug.	16 Aug.	15 Aug.	16 Aug.	14 Aug.	29 Jul.
Average stubble height, cm	25	10	15	15	25	15	7
Straw yield (kg DM ha^-1)	4560	6970	6150	4830	2670	5770	5830
Date for flail-chopping/burning	14 Aug.	19 Aug.	18 Aug.	22 Aug.	19 Aug.	15 Aug.	30 Jul.
Autumn fertilizer (soon after seed harvest), kg N ha^-1	0	0	0	0	0	0	30
Spring/summer (seed harvest year)
Date for spring burning	17 Apr.	17 Apr.	21 Apr.	5 Apr.	27 Mar.	24 Mar.	27 Mar.
Dates for early and late spring fertilization	17 Apr. + 19 May	20 Apr. + 16 May	23Apr. + 22 May	20 Apr + 15 May	23 Apr. + 29 May	7 Apr. + 26 May	30 Apr
N rate applied in early and late spring, kg ha^-1	52 + 41	54 + 31	50 + 25	25 + 49	51 + 39	50 + 30	43 + 0
P + K rate applied in spring, kg ha^-1	18 + 44	14 + 34	14 + 36	14 + 35	17 + 43	13 + 32	16 + 41
Date, plant growth regulation (PGR)	28 May	22 May	21 May	13 May	No PGR	22 May	15 May
Growth regulator used^1)	Moddus	Cycocel	Cycocel	Moddus	–	Cycocel	Moddus
Date for seed harvest	7 Aug.	11 Aug.	7 Aug.	6 Aug.	6 Aug.	5 Aug.	29 July

¹⁾ Dose of active ingredient of Moddus (trinexapac-ethyl) and Cycocel (chlormequat chloride) was 0.15 and 2.0 kg ha^-1

The experimental plan included 10 treatments that were randomized within each of three blocks (replicates)

Table

1.	No chopping or burning. Straw removed after harvest (control)
2.	As treatment 1 + stubble flail-chopped to 5–10 cm soon after harvest
3.	Straw chopped and spread at combining
4.	As treatment 3 + stubble and straw burned after seed harvest
5.	As treatment 3 + stubble, straw and autumn regrowth burned in early spring
6.	As treatment 3 + stubble and straw chopped at 5–10 cm height with a tractor-mounted flail- chopper soon after harvest
7.	Stubble and straw not chopped or burned (intact straw in windrows)
8.	As treatment 7 + stubble and intact straw burned after seed harvest
9.	As treatment 7 + stubble and intact straw, stubble and autumn regrowth burned in spring
10.	As treatment 7 + stubble and intact straw chopped at 5–10 cm height with a tractor-mounted flail-chopper soon after harvest

In most trials, the straw fresh weight in each control plot (tr. 1) was weighed before straw removal and a sample (ca. 1 kg) was taken for dry matter determination after two days drying at 60°C. The straw yield varied from 2670 kg DM ha^-1 (Vestfold 2002) to 6970 kg DM ha^-1 (Aust-Agder 2000) (Table 1).

Flail-chopping (tr. 2, 6 and 10) and burning of stubble and straw (tr. 4 and 8) was accomplished 1–7 days after the final seed harvest. In early spring, burning was always carried out before any signs of spring growth. Dates varied from 24 March to 21 April (Table 1).

All meadow fescue plots at Vestfold in 2002–2003 were fertilized with 40 kg N ha^-1 (calcium nitrate) at the start of the trial in early August, whereas no N was applied to any of the timothy trials in autumn. Spring fertilization in the seed harvest year to timothy and meadow fescue seed crops amounted to 43–90 kg N ha^-1 in the form of compound NPK fertilizer (usually 21-4-10 or 25-2-6) (Table 1). The spring fertilization of timothy seed crops was always split into an early and a late application (Table 1).

In most trials recordings before seed harvest included:

1. Plant cover (%) in spring (assessed in May when plants had grown to a height of about 10 cm).

2. Lodging (%) at flowering 0% = no lodging. 100% = full lodging).

3. Inflorescence numbers as counted in mobile frames between full heading and seed harvest (area varying from 0.25 to 0.4 m²).

4. Weight per unthreshed inflorescence (based on 100 random panicles that had been cut 1 cm below the inflorescence less than a week before seed harvest).

The trials were harvested directly with Nursery master field plot combines (harvest plot size 1.5 × 6.0–9.5 m) at 20–30% seed moisture content, usually in late July (meadow fescue) or early August (timothy) (Table 1).

The plot yields were dried to 10–12% seed moisture content and cleaned on a laboratory air-screen machine (LALS, WESTRUP, Denmark). Representative seed sample, pooled across replicates in each trial, was taken from each of the 10 treatments and analysed for purity and thousand seed weight in the seed laboratory at NIBIO Landvik.

Analyses of variance (PROC ANOVA, SAS Institute 2011) for seed yield and seed yield components were performed both individually for each trial and collectively for all trials in timothy. In the overall analyses, each trial was regarded as one replicate. Results from individual trials will only be presented for seed yield, which is considered the most important character in seed production.

Significant differences, which have been indicated by different letters, were separated by least significant difference (LSD) 0.05. While the term ‘significant’ in this paper always refer to P < .05, exact P-values in the range .05 < P < .20 have also been given in the tables to indicate tendencies.

Experimental series II. Straw chopping and N application in autumn in relation to other management factors

Four trials in timothy (three trials in cv. Grindstad and one trial in cv. Vega) and six trials in meadow fescue (all in cv. Fure) were carried out during 2003–2006 at locations ranging from Landvik (58.2°N) on the Norwegian south coast to Hønefoss, Buskerud (60.2°N), NW of Oslo (Figure 1). Similar to series I, all trials, except at NIBIO Landvik research station, were laid out in commercial seed crops on seed growers’ farms.

In timothy, the experimental plan was a split-split-plot with two replicates, fertilizer distribution of 75 kg N ha^-1 in spring (Factor 1) on main plots, stubble and straw management treatments at/or shortly after seed harvest on subplots (Factor 2) and increasing N rates in autumn (Factor 3) on sub-subplots.

The treatments were:

Factor 1:

X. Split application with 50 kg N ha^-1 at the start of growth + 25 kg N ha^-1 at the start of tiller elongation (both applications were given in the form of NPK 25-2-6).

Y. Split application with 25 kg N ha^-1 at the start of growth + 50 kg N ha^-1 at the start of tiller elongation (i.e. opposite distribution compared with treatment X).

Factor 2. Stubble and straw management at harvest

A. No chopping. Straw removed after harvest (control).

B. Straw chopped and spread at combining

C. Stubble and intact straw in windrows chopped at 5–10 cm height with a tractor-mounted flail-chopper soon after harvest.

D. As treatment B + stubble and straw chopped at 5–10 cm height with a tractor-mounted flail-chopper soon after harvest.

Factor 3. Autumn N rates applied as calcium nitrate soon after stubble and straw management

1. 0 kg N ha^-1

2. 30 kg N ha^-1

The experimental plan for meadow fescue was different from timothy in that fertilizer distribution was replaced by burning vs. no burning of the main plots before start of growth in early spring (Factor 1) and that three N rates (0, 20 and 40 kg N ha^-1) were compared on sub-subplots (Factor 3).

Experimental details from the trials in series II are given in Table 2. In one trial in timothy (Buskerud 2004–2005) and one trial in meadow fescue (Vestfold 2003–2004) straw chopping at combining was performed on straw that had been dried in windrows for about one week (straw was chopped at recombining of the windrows). In all other trials straw was chopped at the single, direct combining. All trials were established at the seed harvest of first-year crops except the timothy trial at Østfold in 2004–2005 which was established at the seed harvest of a second-year crop. Plot length varied according to the cutting width of combiner as described for Experimental series I.

Table 2. Detailed information about four trials in timothy and six trials in meadow fescue seed crops during 2003–2006. Experimental series II.

Location	Timothy trials				Meadow fescue trials
	Østfold	Østfold	Buskerud	Landvik	Aust-Agder	Vestfold	Østfold	Vestfold	Landvik	Østfold
Year	2003–2004	2004–2005	2004–2005	2005–2006	2003–2004	2003–2004	2003–2004	2004–2005	2004–2005	2005–2006
Cultivar	Grindstad	Grindstad	Vega	Grindstad	Fure	Fure	Fure	Fure	Fure	Fure
Soil classification	Clay loam	Clay loam	Clay soil	Medium loam	Silt soil	Silt soil	Clay soil	Silty loam	Silty loam	Clay loam
Autumn (trial establishment year)
Date for straw chopping at combining (Factor 2)/trial establishm.	19 Aug.	13 Aug.	20 Aug.	15 Aug.	31 Jul.	30 Jul.	4 Aug.	26 Jul.	21 Jul.	12 Aug.
Average stubble height, cm	15	5	20	10	16	16	15	20	15	5
Straw yield (kg DM ha^-1)	–	4000	5800	5710	4260	3610	3640	4250	5360	3750
% N in straw dry matter	0.7	0.8	–	–	–	0.9	1.4	–	–	–
Mineral N in soil (0–20 cm) kg ha^-1, after seed harvest	–	7	27	16	–	10	35	16	8	36
Spring/summer (seed harvest year)
Date for spring burning (m. fescue)	–	–	–	–	29 Mar.	30 Mar.	27 Mar.	5 Apr.	1 Apr.	No burn^1)
Dates for early and late spring fertilization	15 Apr. + 18 May	13 Apr. + 13 May	21 Apr + 16 May	21 Apr. + 23 May	10 Apr.	4 Apr.	17 Apr.	22 Apr.	15 Apr.	28 Apr.
N rate applied in spring, kg ha^-1	75	75	10	75	88	60	95	84	80	80
P and K rate applied in spring, kg ha^-1	14 + 36	12 + 30	19 + 48	12 + 30	16 + 40	11 + 29	18 + 45	13 + 34	13 + 32	13 + 32
Date, PGR	No PGR	No PGR	14 May	22 May	19 May	20 May	22 May	No PGR	19 May	18 May
Growth regulator used^2)	–	–	Moddus	Cycocel	Moddus	Moddus	Cycocel	–	Moddus	Moddus
Date for seed harvest	9 Aug.	18 Aug.	18 Aug.	4 Aug.	24 Jul.	27 Jul.	27 Jul.	29 Jul.	27 Jul.	24 Jul.

¹⁾ Burning not performed due to wet conditions in spring. ²⁾ Dose of active ingredient of Moddus (trinexap-ethyl) and Cycocel (chlormequat chloride) was 0.15–0.18 and 2.0 kg ha^-1, respectively

In most trials in both species, a soil sample (0–20 cm depth) was taken from each of the four main plots for determination of soil mineral N at seed harvest. 2003 straw samples (ca 1 kg, across replicates) were also analysed for (Kjeldahl) N content (Table 2).

Field registrations, seed harvest, seed cleaning and seed analyses were performed as described for experimental series I. The data were analysed according to a split-split-plot model with results presented as main effects and interactions among factors (Tables 6 and 7). Two trials were reduced to split-plots with four blocks as all main plots were treated in the same way in spring. This was the case for the timothy trial in Buskerud in 2004–2005 where all plots received 50 kg N ha^-1 in early spring plus 50 kg N ha^-1 at early tiller elongation, and for the meadow fescue trial at Østfold in 2005–2006 where spring burning of main plots was impossible due to high precipitation. Since the main objective was to study the effect of stubble and straw treatments, this factor will always be mentioned first in the tables.

Results

Experimental series I

Plant cover in spring. In both timothy and meadow fescue, the poorest plant cover after start of growth in spring was found on plots where the straw had been left intact in windrows (tr. 7), The maximum plant cover was found on plots where either chopped straw (tr. 4) or intact straw (tr. 8) had been burned in the previous autumn (Table 4 and 5).

Lodging at flowering. On average for the timothy trials, the most lodging at flowering was found on plots where the stubble had been flail-chopped after straw removal (tr. 2). The least lodging was found on plots where the straw had been left intact in windrows (tr. 7) (Table 4). At Vestfold in 2003, plots of meadow fescue where stubble and straw had been burned in autumn (tr. 4 and 8) had significantly more lodging at flowering than the other plots (Table 5).

Seed yield. On plots without straw, reducing the stubble height to 5–10 cm by flail-chopping (tr. 2 vs 1) had a positive effect on timothy seed yield in all trials except at Aust-Agder in 2000–2001. On average for all trials, a 12% significant increase in seed yield by stubble height reduction was found (Table 3). The lowest seed yield in all timothy trials was produced on plots where intact straw was left in windrows behind the combiner without any further chopping/burning treatments (tr. 7). On average for all trials, plots where straw, stubble and autumn regrowth were burned in spring also produced lower seed yield than control plots without straw (tr. 5 and 9 vs. 1). The highest seed yield was harvested on plots where stubble and chopped straw were burned soon after harvest (tr. 4) (Table 3).

Table 3. The effect of various stubble and straw treatments on seed yield (kg ha^-1) of timothy seed crops in trials in southeast Norway. Experimental series I.

Treatments	Seed yield (kg ha^-1, 12% water, 100% purity)
	Vestfold 2000–2001	Aust-Agder 2000–2001	Landvik 2000–2001	Vestfold 2001–2002	Landvik 2002–2003	Vestfold 2002–2003	Mean	Rel
No of trials	1	1	1	1	1	1	6	6
1. Straw removed after harvest	985dc	1133a	601abc	505de	715	691bcd	772bc	100
2. Tr.1 + stubble chopped (5–10 cm)	1116abc	1101a	646abc	592bc	829	887ab	862a	112
3. Straw chopped during combining	1027bcd	1028ab	621abc	542cde	816	627cd	777bc	101
4. Tr. 3 + Straw/stubble burned in autumn	1120abc	1000abc	664abc	687a	890	952a	886a	115
5. Tr. 3 + Straw/stubble burned in spring	920d	939abc	545bcd	536cde	790	755abcd	748c	97
6. Tr. 3 + Straw/stubble flail-chopped in aut.	1231a	1010ab	697ab	557cd	705	819abc	837ab	108
7. Unchopped straw in windrows	705e	792c	408d	301f	668	555d	572d	74
8. Tr. 7 + Straw/stubble burned in autumn	1151abc	1043a	772a	643ab	724	730bcd	844ab	109
9. Tr. 7 + Straw/stubble burned in spring	863de	757c	494cd	488e	577	597d	630d	82
10. Tr. 7 + Straw/stubble flail-chopped in aut.	1189ab	1129a	698ab	493e	824	863ab	866a	112
Significance, P	<.01	.05	.02	<.01	>.20	.01	<.01
LSD 0.05	190	246	183	63	–	211	84

In the meadow fescue trial in Vestfold in 2003, differences in seed yield between treatments were not significant (Table 5).

Seed-yield components. On average for the timothy trials, neither inflorescence density nor weight or length per inflorescence was significantly affected by the experimental treatments (Table 4). Numerically, the lowest inflorescence number in timothy was found on plots where intact straw had been left in windrows with no further treatment (tr. 7) or with burning in early spring (tr. 9). In contrast, the highest inflorescence density was recorded after removal of the straw and chopping of the stubble (tr. 2). The highest thousand seed weight was produced on plots where windrows with intact straw had been left in the field during the winter (tr. 7 and 9), that is, the same plots which also had the lowest inflorescence density.

Table 4. The effect of various stubble and straw treatments on the percentage of plant cover in spring, percentage of lodging at flowering inflorescence number per m², weight (g) and length (mm) per unthreshed inflorescence and thousand seed weight (at 12% water content) in timothy seed crops. Experimental series I.

Treatments	% plant cover in spring	% lodging at flowering	Inflor-escences per m²	Weight per unthreshed inflorescence, mg	Length per inflor-escence, mm	Thousand seed weight, mg
No. of trials	5	5	5	6	6	6
1. Straw removed after harvest	80bcd	16b	797	391	54	635bc
2. Tr.1 + stubble chopped (5–10 cm)	83abc	23a	878	390	54	597e
3. Straw chopped during combining	71de	14b	752	359	52	630bcd
4. Tr. 3 + Straw/stubble burned in autumn	90ab	12bc	789	399	54	610cde
5. Tr. 3 + Straw/stubble burned in spring	76cd	17ab	844	382	53	640bc
6. Tr. 3 + Straw/stubble flail-chopped in autumn	79bcd	17ab	782	374	54	602de
7. Unchopped straw in windrows (untreated)	41f	6c	699	398	55	678a
8. Tr. 7 + Straw/stubble burned in autumn	92a	15b	771	404	56	611cde
9. Tr. 7 + Straw/stubble burned in spring	61e	13bc	686	369	51	657ab
10. Tr. 7 + Straw/stubble flail-chopped in autumn	77cd	17ab	728	390	56	585e
Significance, P	<.01	.01	>.20	.19	>.20	<.01
LSD 0.05	8	7	–	–	–	31

In meadow fescue, there was a tendency for the lowest number (P = .13), but the heaviest inflorescences (P = .08), to be found on plots where straw had been left in windrows behind the combiner (tr. 7). The thousand seed weight was not affected by treatments, the average value at 12% water content being 2607 mg (not shown in the table).

Experimental series II

Plant cover in spring. As a main effect in both species, there was a tendency for less plant cover in spring on plots where the straw had been flail-chopped than in the other treatments (Tables 6 and 7).

Various N strategies in autumn (both species) or in spring (timothy) did not affect plant cover significantly (Tables 6 and 7). In meadow fescue, a significantly better plant coverage in spring was found after spring burning than on unburned plots (Table 7).

Lodging at flowering. Per cent lodging at flowering was usually less than 30 and not significantly affected by either stubble and straw managements or N strategies in autumn (both species) or in spring (timothy). In meadow fescue, significantly more lodging at flowering was found on burned (22%) than on unburned (11%) plots.

In both species, the two- or three-factor interactions were not significant for either plant cover in spring or lodging at flowering.

Seed yield. As the main effect, significant timothy seed yield differences between stubble and straw management treatments were only found at Buskerud in 2004–2005 (Table 6). In this trial, the highest seed yield was harvested on plots where stubble and either intact (tr. 3) or combiner-chopped (tr. 4) straw had been flail-chopped after harvest. On average for all trials, treatments involving straw chopping at combining and/or flail-chopping after harvest, produced higher seed yields than the control treatment with straw removal (tr. 2-4 vs. 1) (Table 6). However, differences were not significant.

A significant increase in timothy seed yield due to autumn N-fertilization was only found in Østfold in 2003–2004 (Table 6). On average for all trials, this increase amounted to 6%, which was not statistically significant.

Except for the timothy trial at Østfold in 2004–2005, where plots receiving 25 + 50 kg N ha^-1 in early and late spring produced significantly higher seed yield than the opposite combination (50 + 25 kg N ha^-1), only small and insignificant differences were found between the two distribution patterns in spring (Table 6).

Interactions between factors for timothy seed yield were either insignificant or gave no meaningful information.

Of the six meadow fescue trials, significant seed yield differences between stubble and straw managements were found at Vestfold in 2002–2003 and at Landvik in 2004–2005 (Table 7). In these two trials, the highest seed yield was harvested on plots where stubble and combiner-chopped straw had been flail-chopped after harvest (tr. 4). On average for all trials, none of the stubble and straw chopping treatments had any negative impact on seed yield compared with straw removal (tr. 2-4 vs. 1).

Increasing N rate in autumn had no significant impact on meadow fescue seed yield in any of the six trials (Table 7).

In all trials, the highest meadow fescue seed yield was harvested on plots being burned in early spring. Compared to non-burned plots, the averaged seed yield increase was 22% (Table 7).

None of the two- or three-factor interactions were significant. However, autumn N (Factor 2) and spring burning (Factor 3) showed a noteworthy tendency (P = .12) to interact as N in autumn increased meadow fescue seed yield on plots that were burned in spring, but not on unburned plots (Figure 2).

Figure 2. Effect of autumn N application rates and spring burning strategies on seed yield (kg ha^-1) in seed crops of meadow fescue. Mean of five crops harvested in Experimental series II. Bars represent standard error (n = 20). P = .13.

Seed-yield components. On average for all trials, neither the main effect, nor the two- or three-factor interactions were significant for inflorescence density, weight per inflorescence or thousand seed weight in any of the two species (data for inflorescence weight, inflorescence length (in timothy) or thousand seed weight are not shown). In meadow fescue, there was strong tendency (P = .06) for inflorescence density to be higher on plots that had received 20 or 40 kg N ha^-1 than on plots that had not received fertilizer in autumn (Table 7).

Discussion

The light intensity influences plant growth through photosynthetic activity and developmental responses. In cocksfoot (Dactylis glomerata), Auda et al. (1966) reported that tillering, dry matter production and carbohydrate percentage of dry weight dropped when light intensity was reduced to 75% of normal sunlight. Also Mitchell (1953) and Patel and Cooper (1961) demonstrated similar effects of light intensity on tiller development in various grasses. In addition to stimulate tillering, and thereby increase the potential number of reproductive tillers, better light penetration to the tiller bases may lead to less competition for light and thus better survival among reproductive tillers (Meijer & Vreeke 1988). In the present trials (Experimental series I), a severe negative effect of shading on seed yield, both in timothy and meadow fescue, was found on plots where straw was not removed from the field (Tables 3 and 5). For adequate amounts sunlight to be received, Hart et al. (2012) suggested that straw should be removed as soon as possible after harvest, preferably within a week.

Table 5. The effect of various stubble and straw treatments on plant cover in spring (%), lodging at flowering (%), panicle number per m², weight per inflorescence (mg) and seed yield (kg ha^-1) in meadow fescue seed crops at Vestfold (2002–2003). Experimental series I.

Treatments	% plant cover in spring	% lodging at flowering	Inflor-escences per m²	Weight per inflor-escence, mg	Seed yield (12% water, 100% purity)
					kg ha^-1	Rel.
1. Straw removed after harvest	83c	25b	816	259	812	100
2. Tr.1 + stubble chopped (5–10 cm)	90b	23b	856	243	901	111
3. Straw chopped during combining	77d	27b	689	258	882	109
4. Tr. 3 + Straw/stubble burned in autumn	97a	73a	948	251	972	120
5. Tr. 3 + Straw/stubble burned in spring	87bc	23b	800	225	929	114
6. Tr. 3 + Straw/stubble flail-chopped in autumn	75d	32b	802	260	913	112
7. Intact straw in windrows (untreated)	33g	18b	525	279	705	87
8. Tr. 7 + Straw/stubble burned in autumn	97a	65a	952	238	951	117
9. Tr. 7 + Straw/stubble burned in spring	53e	18b	722	222	917	113
10. Tr. 7 + Straw/stubble flail-chopped in autumn	47f	10b	742	242	840	103
Significance, P	<.01	<.01	.13	.08	>.20
LSD 0.05	6	30	–	–	–

Although straw removal allows for more light penetration to the tiller bases, the results from Experimental series I (Tables 3 and 5) and II (Tables 6 and 7) showed that other stubble and straw management methods are also acceptable with regard to seed yield (i.e. no significant differences found between straw removal and the best alternative stubble and straw treatments).

Table 6. Main effects of various stubble and straw treatments, N-application rates in autumn and spring on inflorescences/m² and seed yield (kg ha^-1) in seed crops of timothy during 2003–2005. Experimental series II.

Treatments	Plant cover in spring (%)	Inflor-escences/m^2	Seed yield (kg ha^-1, 12% water, 100% purity)
			Østfold	Østfold	Buskerud	Landvik	Mean	Rel
			2003–2004	2004–2005	2004–2005	2005–2006
No of trials		4	1	1	1	1	4	4
Straw/stubble management
1. Straw removed after harvest	82	616	607	490	776b	1099	743	100
2. Straw chopped during combining	81	587	589	598	768b	1135	772	104
3. Straw chopped after combining	75	661	589	546	872a	1129	784	106
4. Straw chopped twice^1)	80	608	645	550	829ab	1113	784	106
Significance, P	.08	>.20	>.20	.14	<.01	>.20	>.20
LSD, 0.05	–	–	–	–	66	–	–
Autumn N fertilization
A. 0 kg N ha^-1	79	604	531	519	817	1137	749	100
B. 30 kg N ha^-1	80	632	683	574	806	1106	793	106
Significance, P	>20	>0.20	<0.01	0.09	>0.20	0.12	>0.20
Split N-application in early + late spring
X. 25 + 50 kg N ha^-1	80	682	605	600a	–	1104	770	100
Y. 50 + 25 kg N ha^-1	78	522	610	492b	–	1136	745	97
Significance, P	>20	>.20	>.20	<.01	–	.11	>.20

¹⁾ Straw chopped both at the back of the combiner during seed harvest and after combining by a tractor-mounted flail-chopper.

Table 7. Main effects of various stubble and straw treatments, N-application rates in autumn and field burning in spring on inflorescences/m² and seed yield (kg ha^-1) in seed crops of meadow fescue during 2003–2005. Experimental series II.

Treatments	Plant cover in spring (%)	Inflor-escences m^2	Seed yield (kg ha^-1, 12% water, 100% purity)
			Aust-Agder	Vestfold	Østfold	Vestfold	Landvik	Østfold	Mean	Rel
			2003–2004	2003–2004	2003–2004	2004–2005	2004–2005	2005–2006
No of trials		6	1	1	1	1	1	1	6	6
Straw/stubble management
1. Straw removed after harvest	75	815	1020	648b	325	865	814b	872	757	100
2. Straw chopped during combining	72	843	920	671b	340	901	830b	913	762	101
3. Straw chopped after combining	69	891	1034	688ab	347	867	813b	894	774	102
4. Straw chopped twice^1)	70	888	904	731a	331	941	880a	966	792	105
Significance, P	.09	>.20	>.20	.02	>.20	.18	.03	.08	>.20
LSD, 0.05	–	–	–	51	–	–	49	–	–
Autumn N fertilization
A. 0 kg N ha^-1	72	826	975	679	331	913	807	909	769	100
B. 20 kg N ha^-1	72	881	933	708	334	906	855	892	771	100
C. 40 kg N ha^-1	72	872	1001	667	343	861	841	932	774	101
Significance, P	>.20	.06	>.20	.17	>.20	>.20	.08	>.20	>.20
Burning of straw/stubble in spring
X. No burning	66b	792	901	593	287b	807	767	–	671b	100
Y. Spring burning	79a	863	1038	776	385a	980	901	–	816a	122
Significance, P	5	>.20	>.20	.02	.04	.10	.12	–	.03

¹⁾ Straw chopped both at the back of the combiner during seed harvest and after combining by a tractor-mounted flail-chopper.

For most seed growers, the most efficient and least laborious method will be to chop the straw at the back of the combiner during seed harvest. On average for all trials in both experimental series, seed yield was 1–4% (Tables 3 and 6) and 1–9% (Tables 5 and 7) higher when using this method compared to straw removal in timothy and meadow fescue, respectively. Straw chopping as an alternative method to straw removal was also demonstrated in Danish seed crops of perennial ryegrass (Lolium perenne) and tall fescue (Festuca arundinacea), the latter seeded at wide row spacing (Boelt 2006). In experiments conducted over several years in Oregon (USA), straw chopping never negatively affected seed yield in cocksfoot and in some cultivars of perennial ryegrass and tall fescue, but severely reduced seed yield in red fescue (Festuca rubra), colonial bentgrass (Agrostis capillaris) and smooth-stalked meadow grass (Poa pratensis) (Chastain et al. 1996, 1997).

Stubble height at combining was not an experimental factor in these trials. However, in Experimental series I all timothy fields with long stubble (15 cm or higher, i.e. all trials except in Aust-Agder in 2000–2001 where stubble height was 10 cm) gave higher seed yield, if the stubble was removed (flail-chopped or burned) after harvest (Table 3). This indicates that stubble height should be kept low (preferable less than 10 cm) at combining. If the stubble is longer, the height should be reduced by using a tractor-mounted flail-chopper, in order to improve light penetration. This is in concurrence with Wallenhammar et al. (2011) who found a 12% increase in seed yield of timothy when the stubble height was reduced from 25 to 40 cm to about 5 cm soon after harvest. In meadow fescue, treatments with stubble height reduction was only carried out in one trial (Table 5), which is too sparse for any conclusion to be drawn. However, due to better light conditions at tiller bases, a low stubble height at combining can probably be recommended even in this species.

In both timothy and meadow fescue, straw chopping at the back of the combiner was carried out either on straw from standing crops (direct combining) or on dried straw in windrows. Although the latter was easier to chop, chopping straw functioned well even at direct combining provided that the knives were sharp. It is, however, very important that the chopped straw is distributed evenly in the field, so that newly developed tillers are able to penetrate easily. If otherwise, an additional cutting and spreading of the straw, for instance using a tractor-mounted flail-chopper or hay rake, is recommended after harvest.

Flail-chopping of intact straw in windrows after combining was, in both species, on level or better than straw removal with regard to seed yield (Tables 3 and 5). However, in some cases, intact straw tended to be blown through the flail-chopper, especially at higher tractor speed. Flail-chopping at reduced speed will alleviate this problem, but this is hardly compatible with the requirements to efficiency in practical seed production. We therefore suggest that it is too uncertain to leave intact straw in windrows for later chopping and spreading with flail-choppers only.

As earlier demonstrated in meadow fescue (Nordestgaard 1981), timothy (Entz et al. 1994), red fescue, perennial ryegrass and cocksfoot (Chilcote et al. 1981), field burning soon after harvest was a fast and efficient method to remove stubble and straw from the field without any negative impact on the following year's seed yield (Tables 3 and 5). However, due to smoke emissions, especially near high traffic roads and in populated areas, open field burning can hardly be recommended as the best preferable stubble and straw management method. Such a practice is also restricted by regulation in some areas of Norway and can be risky, especially during windy conditions, as the fire may come out of control.

In timothy, the burning of stubble, straw and autumn regrowth in early spring was an uncertain method that normally reduced seed yield (Table 3). Especially the high temperatures resulting from burning windrows of intact straw (tr. 9) caused stand injury and lower seed yields. Although only one trial was conducted, meadow fescue was more tolerant to spring burning than timothy in Experimental Series I (Table 5), and in Experimental series II, the seed yield of meadow fescue was, as a main effect, 22% higher after spring burning (Table 7). This is in agreement with other studies which have shown spring burning to be an efficient method to remove wilted stubble and autumn regrowth and thus increase seed yield in meadow fescue (Havstad 2011).

When the straw is chopped and spread on the field, nutrients are recycled. Based on the average N concentration in straw and the straw yield (Table 2), the supply of organic nitrogen by chopping straw was about 38 kg N ha^-1 in meadow fescue and 42 kg N ha^-1 in timothy. In a Danish study, where chopped straw of three temperate grasses was placed on the top of the soil, about one-third of the initial herbage nitrogen content was released during autumn. However, it was uncertain whether the released nitrogen was taken directly up by the seed crop or temporarily incorporated in soil micro-organism (Clausen 2002).

In Experimental series II, nitrogen application in autumn normally did not increase seed yield in either timothy (Table 6) or meadow fescue (Table 7). One exception was the timothy trial in Østfold in 2003–2004 which was located on an infertile sandy soil and suffered from drought in spring 2004. The mean temperature for April and May was 1.7°C above the 30-year normal and the precipitation was only 48% of the normal in the same period. Thus, the beneficial effect of autumn nitrogen (30 kg N ha^-1) in this trial was probably due to the poor availability of nitrogen in spring.

In meadow fescue, the optimal N level in autumn was usually dependent on whether the crop was burned in spring or not (Figure 2). On unburned plots, nitrogen application in autumn usually had a negative effect on seed yield. This was probably related to a higher production of tillers (green biomass) in autumn on fertilized than on non-fertilized plots. When the wilted biomass (autumn regrowth) on fertilized plots was not removed (burned) in spring, the generative development was probably hampered due to shading.

All in all, Experimental series II (Tables 6 and 7) lent no support for the hypothesis that chopping and return of straw should increase the N demand in autumn beyond the present recommendation of 0 kg N ha^-1 in timothy or 20–30 kg N ha^-1 in meadow fescue. This is in agreement with the conclusions by Christensen (1985) and Fog (1988) that nitrogen is usually not a limiting factor for decomposition of straw in agricultural soils. Also in incubation studies with straw from cereals and grasses Havstad et al. (2010) found no stimulation of microbial activity by adding mineral nitrogen to the soil.

In timothy, split N application strategies in spring did not interact with autumn N strategies in any of the trials with regard to seed yield. However, as a main factor, the highest seed yield in two of the three trials was harvested on plots receiving a main application of 50 kg N ha^-1 in early spring and a supplemental input of 25 kg N ha^-1 at BBCH 31 (Table 6). This is in agreement with the Norwegian recommendation for practical seed production of timothy (Havstad & Aamlid 2006). The reason why the opposite combination (25 + 50 kg N ha^-1) produced significantly more seed in Østfold in 2004–2005 is not clear. However, as this was the only trial that was established in the second harvest year, one explanation might be that there was a sufficient number of strong tillers to become reproductive without further stimulation by a large N input (50 kg ha^-1) in early spring.

In conclusion, these experiments showed that several methods for stubble and straw management may be acceptable alternatives to the standard straw removal method in seed production of both timothy and meadow fescue. For most seed growers, the most effective, least laborious and most environment-friendly alternative will be to chop the straw at the back of the combiner during seed harvest. Using this method, it is important to leave a short stubble ( < 10 cm) and distribute the straw evenly on the field during harvest. Otherwise, it is recommended to use a tractor-mounted flail-chopper after harvest to reduce stubble height and unevenly distributed straw. No extra N input was required when the straw was chopped rather than removed. Also burning of straw and stubble soon after seed harvest was an efficient clean-up alternative to straw removal. However, as the burning method is risky and causes smoke emissions, it is normally not recommended.

Acknowledgements

Thanks are extended to Trygve S. Aamlid for valuable comment on the manuscript and to the staff at NIBIO Landvik and the Farmers Experimental and Advisory groups in Vestfold (NLR Viken), Agder (NLR Agder), Østfold (NLR SørØst) and Buskerud (NLR Østafjells) for skillful implementation of these experiments.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

This research was funded by a grant from the Norwegian Seed Growers’ Union (Norsk frøavlerlag) and [grant number 155865/110] from the Norwegian Research Council.