Poa pratensis or Lolium perenne for establishment and overseeding of Scandinavian football (soccer) pitches

Abstract

Scandinavian football pitches are sometimes sodded because of slow establishment of Kentucky bluegrass (KB, Poa pratensis) from seed. Sodding is fast but expensive and brings with it thatch and sometimes annual bluegrass (AB, Poa annua). Perennial ryegrass (PR, Lolium perenne) is used for overseeding and often becomes an important component of the turf. Triggered by experiences in SW Norway and predictions for milder winters, our objectives were (1) to compare KB sod and PR seed for establishment of football pitches in three climatic regions; and (2) for both types of turf to document the effect of overseeding PR or KB in spring, summer or fall on turf cover, botanical composition, and wear tolerance. Field trials with sodded KB and seeded PR on main plots and overseeding on subplots were conducted from June 2008 to Nov. 2011 at the Bioforsk units Landvik (58 °N, coast), Apelsvoll (61 °N, inland) and Kvithamar (64 °N, coast). Seeded PR plots could be opened to play one month after sodded KB plots. KB survived well at all sites, but complete winterkill of PR occurred in two out of three years at Kvithamar and one out of three years at Apelsvoll. At Landvik one of two main plots with PR died during the last winter. The fact that winterkill was not caused by snow mold but long-lasting ice covers or low freezing temperatures combined with little or no snow cover, shows that a milder but more unstable winter climate will not allow for pure PR pitches in the future. Pitches still ought to have KB in the base, but PR is recommended for overseeding, preferably in spring, to reduce AB ingress and improve turf quality. Overseeding of PR into KB did not outcompete KB, but overseeding of KB into PR or KB was mostly not successful.

Keywords:

• Botanical composition
• playing quality
• Poa annua
• sod
• tiller density
• turfgrass
• wear tolerance
• winterkill

Introduction

The most widely used turfgrass species for football (soccer) pitches in the temperate climate zone are Kentucky bluegrass (KB, Poa pratensis) and perennial ryegrass (PR, Lolium perenne). Within the Nordic countries, KB has traditionally been preferred to PR because of better winter hardiness. In Norway, pure PR seed blends for establishment of football pitches have only been used in the coastal south-western part of the country (mainly Stavanger area), which has a mild oceanic winter climate (Sømme 1968). From this region, Andersen (2004) reported better wear tolerance and turf quality of pitches that had been seeded with pure PR or PR/KB mixtures than of pitches that had been seeded with pure KB. This is in line with results showing better wear tolerance of PR than of KB in England, the Netherlands and France (Canaway 1981, 1983). By contrast, ranking of the two species for wear tolerance in North American studies produced inconsistent results depending on how wear tolerance was defined and which method was used to measure it (Shearman and Beard 1975a, 1975b, Minner et al. 1993, Minner and Valverde 2005, Stier et al. 2008).

Despite never being seeded, annual bluegrass (AB, Poa annua) is perhaps the most prevalent species on Scandinavian football pitches. Thanks to its rapid establishment from the soil seed bank, this opportunistic turfgrass fills in voids that are created by wear or other disturbances (Grime and Hunt 1975, Vargas and Turgeon 2004). One of the best ways to control AB is to overseed with a more desirable species that is equally fast in establishment, e.g. PR (Canaway 1981, Canaway et al. 1986). In Canada, overseeding of PR is regarded as a strategy to avoid ingress not only of AB but also of broad-leaved weeds on football pitches that are not allowed to use pesticides (Elford et al. 2008).

Because of its slow establishment from seed, football pitches with KB are often established from sod. Although this is a fast method, the sod usually brings with it soil and a considerable amount of thatch that requires extra attention after installation. The soil problem can be sorted out by sod washing (e.g. Turgeon et al. 1978, Casimaty et al. 1993) or by growing the sod on a sandy soil that is compatible with the rootzone mixture of the football pitch, however, this does not solve the thatch problem which may dramatically reduce infiltration rate as the pitch is brought into play (Canaway 1990, 1993). At least in Scandinavia, experience also shows that KB sod is usually contaminated with AB, and it has been suggested that this problem will become even worse if the sod is handled in a washing machine.

According to the Intergovernmental Panel on Climate Change, global temperatures are likely to increase by 2–6 °C by the end of the 21st century (IPPC 2007). The most significant temperature increases are expected at northern latitudes and during the winter period. In Norway, we have seen this trend already: from 1988 to 2007 (the two last decades before the start of this project) there was only one year (1996) with annual mean temperature below the mean value for the 20th century (Anonymous 2012). Higher temperatures might reduce the risk for PR winter damage, but this is not a simple relationship as winter survival depends on a number of factors. Contrasting predictions can be illustrated by Bélanger et al. (2002) who modeled that global warming would result in more winter injury of PR forage crops in eastern Canada because of less hardening in the fall and less snow cover during winter, and by Thorsen and Höglind (2010) who modeled that global warming would expand the use of PR in Norway. Elongation of the playing season by the use of artificial light and/or soil heating systems is another factor that may well lead to more use of PR instead of KB.

In the short term, a very relevant question asked by many groundsmen is how to improve the quality of deteriorated pitches with a mixed sward of KB and AB. Both experimental evidence (e.g. Canaway et al. 1986, Minner and Valverde 2006, Elford et al. 2008) and practical experience suggest that such pitches can be improved by overseeding PR, however, the optimal seeding time has not been investigated under Scandinavian conditions. Especially in continental and northern parts of Scandinavia, groundsmen may want to take advantage of the wear tolerance of PR without sacrificing the winter hardiness of KB, but then the question is when to overseed in order to create the best balance between the two components. As in the continental USA (Koski 2000, Lyons and Elford 2009), concern has been raised that PR will outcompete KB and result in more winter damage. The prospects for successful overseeding of KB into AB or PR are, on the other hand, debatable according to British and American experiences (Canaway et al. 1986, Koski 2000, Koski and Newberry 2004, Minner and Valverde 2006), but again, this has not been investigated under Scandinavian conditions where PR is often set back during the winter period.

The objectives of this research were:

1.	To investigate the risk and potential of replacing KB sod with directly seeded PR for establishment of football pitches in three regions of Scandinavia with contrasting climates.
2.	To clarify the potential benefits and optimal time for overseeding PR or KB into existing swards of either species in the same climatic regions.

Figure 1.  Map of south Norway showing experimental sites.

Materials and methods

Experimental sites

Experimental football pitches were established on 10–13 June 2008 on well-drained soils at the Bioforsk Research Centers Landvik (58° 20′ N, 10 m a.s.l.), Apelsvoll (60° 42′ N, 250 m a.s.l.), and Kvithamar (63° 49′ N, 35 m a.s.l.) (Figure 1). Landvik and Kvithamar have coastal climates typical for southernmost and central Norway, respectively, and Apelsvoll has a continental climate (Table I; Sømme 1968). At all sites the existing soil was excavated to a depth of approximately 10 cm. The 10 m×20 m subgrade was profiled with a ridge in the middle and one block having a gentle 2% slope to the north or east, and the other block having a 2% slope to the south or west. Approximately 15 cm depth of sand was added, mixed gently with the existing soil to avoid distinct layers and compacted before seeding or sodding. Characteristics of the amended topsoils are given in Table II.

Table I. Thirty-year normal values (1961–90) for seasonal temperatures, length of growing season, thermal heat units, and annual precipitation at the experimental sites Landvik, Apelsvoll, and Kvithamar.

	Temperature, °C			Growing season
	Year	June–Aug.	Dec.–Feb.	Days>5 °C	Day-degree units>5 °C	Annual precip., mm
Landvik	6.9	15.4	−1.1	16 Apr. – 3 Nov.	1441	1237
Apelsvoll	3.6	14.0	−6.6	29 Apr. – 14 Oct.	1091	600
Kvithamar	5.0	13.1	−2.7	20 Apr. – 19 Oct.	1064	897

Experimental plan and maintenance

The experimental pitches were established according to a two-factorial split-plot design with two blocks, one facing north or east and the other facing south or west.

Table II. Soil characteristics of the amended topsoil at establishment in 2008: Loss on ignition, pH, and particle size analysis.

			Soil texture, per cent of weight
	Loss on ignition %	pH (H2O)	Coarse sand 0.6–2 mm	Medium sand 0.2–0.6 mm	Fine sand 0.2–0.06 mm	Silt 0.002–0.06 mm	Clay<0.002 mm
Landvik	2.0	6.7	8.8	44.1	25.2	17.7	4.2
Apelsvoll	4.0	6.3	18.0	29.9	14.6	22.5	15.0
Kvithamar	6.0	6.7	20.4	22.4	14.5	31.1	11.5

The experimental treatments were:

Factor 1: Initial turf type (main plots 7 m×4 m plus borders). A. Directly seeded perennial ryegrass (PR). B. Sodded Kentucky bluegrass (KB).

Factor 2: Species and time for overseeding (subplots 1 m×4 m plus borders). 1. No overseeding (control). 2. KB overseeded in spring (before the start of the playing season). 3. PR overseeded in spring (before the start of the playing season). 4. KB overseeded in mid-summer (at the start of the 3-week football vacation). 5. PR overseeded in mid-summer (at the start of the 3-week football vacation). 6. KB overseeded in fall (at the end of the playing season). 7. PR overseeded in fall (at the end of the playing season).

Varieties used for direct seeding of PR were 50% ‘Bizet 1’ and 50% ‘Concerto’. The protocol stated a seeding rate of 40 gm⁻² at all sites, but the drop-seeder at Landvik gave 25% more, i.e. 50 g m⁻². After seeding, raking, and rolling, the main plots with perennial ryegrass at Landvik and Apelsvoll were covered with a white permeable tarpaulin (‘Agryl’) for 1–2 weeks to increase soil temperature and prevent the seedbed from drying out. No tarpaulin was used at Kvithamar.

Pure KB sod was delivered from the same commercial sod farm in SE Norway to all experimental sites. The sod had been seeded with 25% ‘Eva’, 25% ‘Conni’, 25% ‘Limousine’, and 25% ‘Julia’ and had a significant thatch layer. The sod was seeded in June 2005, grown on a loam soil and cut at 20 mm thickness. Like main plots seeded with PR, main plots sodded KB received frequent irrigation during the first 2–3 weeks after establishment.

Overseeding was accomplished for the first time in fall 2008 and later three times per year according to the experimental plan. Dates are indicated in Table III. The trials at Apelsvoll and Kvithamar were discontinued after assessments in October 2011; hence there was no overseeding in the fall of that year. Overseeding at Kvithamar in fall 2009 was not accomplished. The seed blend of PR used for overseeding was the same as for establishment in 2008, except that seed of ‘Bizet 1’ was not available and therefore replaced by ‘Vesuvius’ in 2011. The KB seed blend used for overseeding contained 50% ‘Limousine’ and 50% ‘Julius’. Seeding rate of both species was 30 g m⁻². For re-establishment of PR after winterkill (see Results section), we also used 30 g m⁻² except at Landvik which received 50 g m⁻² in 2011. The following procedure was used for overseeding, i.e. three times per year: (1) the entire experimental pitch was aerated with 12 mm cores (plugs removed) or solid tines to approximately 5 cm depth; (2) overseeding of subplots was accomplished by hand or with a drop seeder (90-cm wide) and the seeds raked or blown into the holes; (3) the entire experiment was topdressed with 3 mm of medium to coarse sand with no organic amendment. With this procedure carried out three times per year, we avoided confounding of overseeding with aeration or topdressing treatments.

Table III. Dates for overseeding in spring (treatments 2 and 3), summer (treatments 4 and 5) and fall (treatments 6 and 7) during the experimental years. Not OS = Not overseeded.

		2009			2010			2011
	2008	Spring	Summer	Fall	Spring	Summer	Fall	Spring	Summer	Fall
Landvik	9 Oct.	24 Apr.	6 Jul.	8 Oct.	28 Apr.	25 Jun.	13 Oct.	13 Apr.	1 Jul.	14 Oct.
Apels voll	16 Oct.	14 May	13 Jul.	7 Oct.	5 May	29 Jun.	6 Oct.	4 May	1 Jul.	Not OS
Kvithamar	29 Sept.	8 May	8 Jul.	Not OS	28 Apr.	8 Jul.	3 Nov.	10 May	1 Jul.	Not OS

In the establishment year, directly seeded PR and sodded KB received 29 and 24 g N m⁻² respectively, partly as a preplant application of organic fertilizer, and partly as granular mineral complete fertilizer at 1–2 week intervals for seeded PR and 2-week intervals for sodded KB. In the following years, granular fertilizer was applied at 2-week intervals at a total rate of 23 g N m⁻² at Landvik and 18 g N m⁻² at Apelsvoll and Kvithamar. This difference was justified by Landvik having a longer growing season than Apelsvoll and Kvithamar (Table I). Additional fertilizer was usually applied to main plots of PR following re-establishment after winterkill. Irrigation was carried out as needed, mostly after overseeding/topdressing and fertilizer inputs.

The experimental pitches were mowed twice per week with a rotary mower at 30 mm except during the first 2 months after establishment and in early spring and late fall when mowing height was raised to 35 mm. Clippings were always collected. Except for one application of fluroxypyr on 17 July 2008 for control of broad-leaved weeds at Landvik, and one application of MCPA + MCPP on 25 June 2009 for control of broad-leaved weeds at Kvithamar, there was no use of pesticides in any of the trials.

The experimental pitches were subjected to compaction and wear from a tractor-pulled wear machine consisting of two drums with football studs rotating at different speeds and with a ground pressure of 24 kPa. Wear was initially applied twice per week, each time with eight passes over the plots, but in the last project year it was intensified with up to 50 passes per week at Landvik.

Assessments

The following data were recorded:

•	Live turf cover of KB, PR and AB at 2-week intervals in the establishment year and at monthly intervals in the following years. The character was determined visually as per cent of plot area. For presentation, seasonal averages were calculated for spring/early summer (April, May, and June), summer (June and July), and fall (September, October, and November).
•	Tiller density and botanical composition in early June, late July/early August and late October/November at Landvik and Kvithamar in 2010 and 2011. At Kvithamar, tillers of KB, PR and AB were counted in one core sample, 48 mm in diameter from each subplot. At Landvik, counting was carried out on four cores, 23 mm in diameter from each plot. Because of no experienced staff that could separate the different species, only total tiller number was recorded at Landvik in June 2011. In 2009, botanical composition at all three sites was determined by visual assessment only; at Apelsvoll, this was the case even in 2010 and 2011.
•	Shear strength (rotational traction) on all subplots at monthly intervals at Landvik in 2008–2011 and at Apelsvoll and Kvithamar in 2009–2011. We used a disc, 15-cm diameter and with six football studs, loaded to a total weight of 46 kg, and recorded traction (Nm) upon slippage on a double-armed torque wrench (Canaway and Bell 1986, Baker and Canaway 1993).
•	Surface hardness of all subplots at monthly intervals at Landvik and once or twice per season at Apelsvoll. Hardness was measured using a Clegg Impact Tester with 2.25 kg hammer weight (Lafayette Instrument Co., Lafayette, IN, USA) and expressed in Gravity units (G) (Baker et al. 2007).
•	Surface infiltration rates at the end of each growing season using double-ring infiltrometers with inner and outer ring diameters of 45 and 129 mm, respectively. The infiltrometers were filled with 80 mm water and infiltration recorded after three minutes. Measurements were conducted on control plots without overseeding at Landvik and Kvithamar in 2009, 2010, and 2011.
•	Turfgrass root development. Root depth was recorded at approximately 2-week intervals during the two first months after seeding and sodding in 2008. Later, turfgrass root mass was determined in one soil core, 48 mm in diameter and 30 cm deep, taken from unseeded control plot at the end of each growing season at all sites. The intact cores were cut into the depths 0–50 mm and below 50 mm before root washing, drying at 60 °C for 48 h and weighing. Complete separation of roots from soil and sand was very difficult for the upper layer which also included rhizomes and dead roots in the thatch.

Statistical analyses and presentation of results

Experimental data were analyzed according to a mixed model with experimental site, year, initial turf type, overseeding and their interactions as fixed effects and block number and interactions with block number as random effects (PROC MIXED; SAS, 2008). Because of various degrees of winter damage at the various sites and in the different years, these analyses showed turf cover to be influenced by significant (p≤0.05) interactions with site and/or year for both experimental factors (Table IV). This was also the case with the analyses of data for seasonal tiller density/botanical composition at Landvik and Kvithamar in 2010 and 2011. As a consequence of these interactions, subsequent analyses for turf cover and tiller density/botanical composition were performed separately for each year and within each site.

Table IV. Significance levels for the effect of site, year, initial turf type (TT), overseeding (OS) and their interactions on response variables as determined by PROC MIXED procedure. Symbols: ***: p≤0.001, *: 0.001<p≤0.01, *: 0.01<p≤0.05, (*): 0.05<p≤0.10, ns: p>0.10.

	Live turf cover			Tillers m^−2							Root dry weight^4
	Apr.–June	Jul.–Aug.	Sept.–Oct.	KB^1	PR^2	AB^3	Total	Shear strength	Surface hardness	Infiltration rate^4	0–5 cm depth	Under 5 cm depth
Site	*	*	*	ns	(*)	ns	(*)	**	ns	ns	*	(*)
Year	*	ns	*	ns	ns	ns	ns	ns	ns	ns	*	ns
Site×year	*	ns	***	ns	ns	ns	ns	***	ns	ns	*	(*)
Initial turf type (TT)	*	ns	ns	*	*	ns	ns	ns	ns	ns	(*)	ns
Site×TT	**	ns	ns	ns	(*)	ns	ns	ns	ns	ns	ns	ns
Year×TT	(*)	(*)	*	ns	ns	ns	ns	ns	ns	ns	ns	ns
Site×year×TT	***	ns	***	ns	ns	ns	ns	ns	ns	ns	(*)	ns
Overseeding (OS)	(*)	ns	ns	**	**	ns	ns	ns	ns	–	–	–
Site×OS	*	ns	ns	*	(*)	(*)	ns	ns	ns	–	–	–
Year×OS	*	ns	ns	*	*	ns	*	ns	ns	–	–	–
Site×year×OS	***	ns	*	*	*	ns	*	ns	ns	–	–	–
TT×OS	(*)	ns	ns	*	*	ns	(*)	ns	ns	–	–	–
Site×TT×OS	(*)	(*)	ns	(*)	ns	ns	ns	ns	ns	–	–	–
Year×TT×OS	**	ns	ns	ns	ns	ns	ns	ns	ns	–	–	–
Site×Year× TT×OS	**	ns	ns	ns	ns	ns	ns	ns	ns	–	–	–
^1Kentucky bluegrass.
^2Perennial ryegrass.
^3Annual bluegrass.
^4Infiltration rate and root dry weight were determined on unseeded control plots only.

Results

Establishment from seed and sod

Mean temperatures for June plus July 2008 were 0.6, 1.1, and 1.5 °C higher than the 30-year (1961–90) normal values at Landvik, Apelsvoll, and Kvithamar, respectively. Combined with frequent fertilizer inputs and irrigation, this resulted in fast grow-in of KB sod and rapid germination of PR seed at all sites. By 1 August (6–7 weeks after establishment) turf cover exceeded 90% and roots were deeper than 10 cm on all plots (data not shown). On average for September and October, the mean live turf cover on sodded KB plots was 2% units higher than on seeded PR plots at Landvik, while there was no difference between the two turf types at Apelsvoll or Kvithamar (Table V).

Table V. Seasonal mean values for live turf cover (% of plot area; Kentucky bluegrass plus perennial ryegrass plus annual bluegrass) during 2008–2011 on main plots established by sodding Kentucky bluegrass (KB) and seeding perennial ryegrass (PR). Means of subplots with various overseeding treatments. Significance symbols as in Table IV.

		2008	2009			2010			2011			3- or 4-year mean
Site		Sept.–Oct.	Apr.–Jun.	Jul.–Aug.	Sept.–Oct.	Apr.–Jun.	Jul.–Aug.	Sept.–Oct.	Apr.– June	Jul.–Aug.	Sept.–Oct.	Apr.–June	Jul.–Aug.	Sept.–Oct.
Landvik	KB	99	99	100	99	96	100	86	86	99	96	94	99	95
	PR	97	95	99	97	92	100	96	61^1	100	97	83	100	97
	Sign.	(*)	*	ns	ns	ns	ns	(*)	ns	ns	ns	ns	ns	ns
Apelsvoll	KB	99	98	99	99	97	97	99	91	96	98	96	97	99
	PR	99	68	95	98	43	97	100	45	97	100	52	96	99
	Sign.	ns	*	ns	ns	**	ns	ns	*	ns	***	*	ns	ns
Kvithamar	KB	100	100	100	99	92	99	98	85	100	99	93	100	99
	PR	100	41	96	99	89	98	98	55	100	98	6	97	99
	Sign.	ns	***	(*)	ns	(*)	*	ns	*	ns	ns	*	*	ns
Mean of three sites	KB	99	99	100	99	95	99	95	87	98	98	94	99	98
	PR	99	68	97	98	75	98	98	53	99	98	65	98	98
	Sign.	(*)	*	(*)	ns	*	ns	ns	(*)	ns	ns	*	ns	ns
^1Winterkill of PR on main plot facing north only.

Winterkill and the need for re-establishment of main plots with perennial ryegrass

At Landvik, the winter (December–February) 2008/09 was warmer (−0.3 °C), but the winters 2009/10 (−4.3 °C) and 2010/11 (−4.0 °C) substantially colder than the 30-year normal value of −1.1 °C (Table I). Main plots with PR survived the first winter without damage and also recovered nicely after 102 days of stable snow cover during the second winter. The third winter was tougher with the combination of no or little snow cover and temperature below −20 °C in early December, followed by mild spells and a thick ice cover from early January to mid-March. In spring 2011, the PR main plot with 2% inclination to the south was still alive, but the main plot facing north, which had been subjected to a longer and thicker ice cover, was dead and had to be re-established from seed in April 2011.

At Apelsvoll, the mean winter temperatures in 2008/09, 2009/10, and 2010/11 were −5.7, −9.3 and −8.5 °C, respectively (30-year normal value −6.6 °C, Table I). There were stable snow covers during all winters, usually covering the pitch from mid/late November to mid-April. In most cases, the ground was not deeply frozen before snow fall. Snow mold damage caused by Microdochium nivale and Typhula sp. was recorded in all years, but PR recovered without re-establishment in spring 2009 and 2010. By contrast, the combination of early frost and little snow cover in Nov. 2010 took its toll on both main plots with PR which therefore had to be re-established in spring 2011.

The mean winter temperatures at Kvithamar in 2008/09, 2009/10, and 2010/11 were −2.3, −6.6, and −3.9 °C, respectively (30-year normal value −2.7 °C, Table I). Despite that, main plots with PR survived sufficiently to be able to recover without re-establishment only after the winter 2009/10 (Table V). This winter was characterized by moderate frost in the ground and a stable snow cover from 18 Dec. to 5 April. Although there was damage from snow mold, these conditions were better for survival of PR than the winter 2008/2009 which had mild spells with rainfall causing ice encasement from January to April, and the last winter 2010/11 which had a record-low mean temperature for November of −8.6 °C.

Live turf cover

Effects of initial turf type

After being re-established from seed, main plots with PR recovered fast and always developed at least 90% live turf cover by mid to late June. On average for spring and early summer (April–June) observations, live turf cover was, nonetheless, always lower on main plots seeded with PR than on main plots sodded with KB. This main plot difference was significant in one out of three years at southern location Landvik, three out of three years at the continental location Apelsvoll and two out of three years at the northern location Kvithamar (Table V).

On average for sites and years, there was no significant effect of initial turf type on turf cover in summer or fall. At Apelsvoll, PR had significantly better turf cover than KB in the fall of 2011, and the same tendency (0.05<p≤0.10) was seen at Landvik in the fall of 2010. This might reflect a slightly better wear tolerance in PR than in KB.

Effects of overseeding

The main effect of overseeding on live turf cover in spring was not significant (Table IV). The highly significant interaction site×year×overseeding (Table IV) was mainly due to reduced live cover in spring 2011 after overseeding PR into KB at Kvithamar (Figure 2). A similar trend was seen in the same year at Landvik (data not shown). By contrast, overseeding of PR and, to a lesser extent, KB into PR in the fall of 2010 improved live turf cover on PR plots in the spring of 2011 at Kvithamar (Figure 2), but that was due to overseeding in fall 2010 being conducted as late as 4 Nov. (Table III), i.e. shortly before soil freeze-up. In other words, these late overseeding treatments acted as early spring seedings that initiated development of turf cover before the entire main plots with PR were re-established on 10 May. This explanation is supported by the fact that the same effects of overseeding KB or PR in the fall on live turf cover on PR plots in spring were not seen at Kvithamar in 2009 or 2010, and not at Landvik or Apelsvoll in any year.

Figure 2.  Effect of 2–3 years overseeding of Kentucky bluegrass (KB) or perennial ryegrass (PR) at various times of year on live turf cover in spring 2011 at Kvithamar. Vertical bars indicate 1 standard error (S.E., n = 2).

Botanical composition and tiller density

Landvik

Visual assessments of turf cover at Landvik in 2009 showed an average of 4% AB on main plots sodded with KB, but no AB on main plots seeded with PR. At the end of the season there was significantly less AB on KB plots that had been overseeded with PR in the summer (1% AB) than on unseeded control plots (7% AB). On average for seasonal observations, PR made up 4, 8, and 2% of turf cover on KB plots after overseeding in spring, summer, and fall, respectively. Visual assessment showed no effect of overseeding KB on AB occurrence in KB plots (data not shown).

Results from tiller countings at Landvik in 2010 and 2011 are shown in Figure 3.

Figure 3.  Effect of overseeding treatments on tiller populations of Kentucky bluegrass, perennial ryegrass and annual bluegrass on (a) main plots sodded with Kentucky bluegrass and (b) main plots seeded with perennial ryegrass at Landvik in 2010 and 2011. Vertical bars indicate 1 standard error for total tiller number (S.E., n = 2).

Overseeding of KB or PR into KB had no significant effect on total tiller density in 2010. On average for seasonal countings, overseeding of PR resulted in replacement of about one-third of the KB tillers with PR tillers; this proportion increased to an average of 59% in 2011 (Figure 3a). In 2011, more KB tillers were counted after overseeding with KB, but the difference from unseeded control was significant only after overseeding in fall. In the last year of the project, there was also a trend for infestation of AB to be higher on unseeded control plots than on overseeded plots, and higher after overseeding in spring than after overseeding in fall. Unusually high temperatures in October and November (monthly means mean 8.9 and 6.5 °C compared with 30-year normal values of 7.9 and 3.2 °C, respectively) caused a flush of new seedlings of KB or PR after overseeding in October 2011 (treatments 6 and 7).

Overseeding of KB or PR had little effect on the total tiller density on plots initially seeded with PR (Figure 3b). The exception was summer seeding in 2010 which caused a flush of KB seedlings, however, most of these seedlings were out-competed during the summer and could not be found as mature plants in 2011.

Apelsvoll

Botanical composition at Apelsvoll was determined by visual assessment only. On average for monthly registrations during the growing season, overseeding of PR resulted in an average of 2, 6 and 5% PR in KB plots in 2009, 2010, and 2011, respectively. There was no significant difference depending on seeding date. Vice versa, overseeding of KB resulted in less than 1% KB in PR plots in 2009 and 2011 and a maximum of 2% in 2010; again, differences between seeding dates were not significant. AB infestation was lower at Apelsvoll than at Landvik; on average 2% on plots sodded with KB and less than 1% on plots seeded with PR (data not shown in tables or figures).

Kvithamar

On average for seasonal countings at Kvithamar, overseeding of PR resulted in a significant increase in the total tiller number on plots sodded with KB in 2010, and an almost significant (p = 0.07) increase in 2011 (Figure 4a). The optimal time for overseeding PR into KB varied among years: in 2010 the greatest change in botanical composition was recorded after seeding in spring or summer; while in 2011 there was a good response to seeding in fall but a poor response to seeding in summer. Compared with the unseeded control treatment, overseeding of KB into KB plots in spring or summer reduced the total tiller population in 2010 but had an insignificant positive effect in 2011.

The effects of overseeding PR or KB into PR plots were mostly small (Figure 4b). Overseeding with PR in spring 2010 enhanced recovery and resulted in an overall higher tiller density than on other subplots in 2010, but the effect was not statistically significant. A certain response to overseeding of KB into PR was recorded only with the fall treatments.

Figure 4.  Effect of overseeding treatments on tiller populations of Kentucky bluegrass, perennial ryegrass, and annual bluegrass on (a) main plots sodded with Kentucky bluegrass and (b) main plots seeded with perennial ryegrass at Kvithamar in 2010 and 2011. Vertical bars indicate 1 standard error for total tiller number (S.E., n = 2).

Turfgrass shear strength (rotational traction) and hardness

Initial turf type or overseeding had no significant effect on turfgrass shear strength at any site (Table IV). On average for three years of observations at three sites, the shear strength of sodded KB and seeded PR were both 47 Nm. The corresponding mean values for surface hardness were 68 gravities for sodded KB and 73 gravities for seeded PR; however, this difference was also not significant (p = 0.13).

Infiltration

On average for unseeded control plots at Landvik and Kvithamar, turfgrass infiltration rate decreased from 257 mm h⁻¹ in fall 2009 to 218 mm h⁻¹ in fall 2010 and 98 mm h⁻¹ in fall 2011. The effect of year was not significant (p = 0.12). The average infiltration rate was higher in sodded KB (266 mm h⁻¹) than in seeded PR (117 mm h⁻¹), but again, the difference was not significant.

Root dry weight

The dry weight of living and dead roots in the top 5-cm layer (including thatch) was almost significantly (p = 0.06) higher on sodded KB plots than on seeded PR plots. Differences below 5 cm were not significant (Table VI).

Table VI. Dry weight of living and dead roots above and below 5-cm soil depth on plots that were initially established with Kentucky bluegrass sod (KB) or perennial ryegrass seed and that were not overseeded.

	Root and thatch dry weight, g m^−2
	0–5 cm	Below 5 cm
KB	2030	112
PR	1053	143
P%	(*)	ns

Discussion

KB vs. PR for establishment of Scandinavian football pitches

After two decades with higher than normal temperatures, the two last winters of the project period were the coldest since 1987 (Anonymous 2012). Main plots with PR suffered total winter damage during 2008/09 and 2010/11 at Kvithamar and during 2010/11 at Apelsvoll. During 2010/11, PR's limits for winter survival were exceeded even on the Norwegian south coast, as indicated by the total winterkill of one out of two main plots at Landvik. Although the experimental pitches were not sprayed with fungicides before winter and snow mold contributed to damage at Apelsvoll, it was mostly the physical conditions, either the long-lasting ice cover at Kvithamar in 2008/09 and Landvik in 2010/11, or the combination of low freezing temperatures and little or no protective snow cover in November/December 2010, that killed the turf. After this physical damage, the turf was all dead, but after snow mold there were always some surviving crowns or buds which enabled the turf to fill in the gaps thanks to the high tillering potential of PR (Ebdon et al. 2002). These observations support Canadian models suggesting that more unstable winters with less snow cover are likely to increase rather than decrease the risk for winter kill of PR (Bélanger et al. 2002). Even though the winters 2009/10 and 2010/11 may not be typical for the future winter climate, the results lead us to the conclusion that monocultures of PR are too risky for Scandinavian football pitches. Up to present, winter hardiness has not been a priority in turfgrass breeding programs with PR (e.g. Brede 2000), but this will hopefully change as there is no negative correlation between winter tolerance and turf quality components (Hulke et al. 2008).

Sodded vs. seeded turf on various rootzones

Two common objections against KB sod for football pitches are that the sod brings with it AB and that the thatch reduces infiltration as the pitch is brought into play. The first of these objections was confirmed by this project, the second was not. Unlike in British comparisons of seed vs. sod for football pitches and golf courses (Canaway 1990, 1993), infiltration rates in fact tended to be higher in sodded KB than in seeded PR. Even though the sod was three years old and relatively thatchy, it seems that our management program with spiking or coring followed by topdressing three times per year, the total amount of sand corresponding to 110 tons of sand per year on a 0.8 ha football pitch, resulted in sufficient dilution of the organic matter to avoid reductions in turfgrass infiltration rates. The systematic topdressing also leveled out the difference in surface hardness between sodded KB and seeded PR, Clegg impact values always being within the range 45–90 gravities as recommended by Baker et al. (2007).

The experimental pitches were all located on well-drained soils, but the construction was very simple in that 10 cm of the topsoil was replaced with medium to coarse sand before mixing the two layers gently together. The rootzones did not meet UGSA or any other standard, but the rootzone at Landvik (2% organic matter, 4% clay, 18% silt) was not too far from the Norwegian recommendation for general-purpose football pitches (Norwegian Ministry of Culture 2002). Provided that subsurface drainage is not limiting, the content of fines (silt, clay and organic matter) not higher than at Landvik, and the construction work carried out under good weather conditions, both our results and those of Baker and Canaway (1990) suggest that this is an acceptable and cost-effective method for construction of general-purpose football pitches. The establishment of turf from seed on such rootzones is obviously faster and simpler than on pure sand rootzones, and the use of unwashed sod is also less problematic as there will be no distinct layers in the profile. For all experimental sites, it is noteworthy that the wear tolerance, as indicated by traction levels throughout the project period, were very similar for sodded KB and seeded PR, and always well above the ‘preferred minimum level’ of 30 Nm indicated by Bell and Holmes (1988). Our results are contradictory to the relatively old results from England, the Netherlands and France showing better wear tolerance in PR than in KB (Canaway 1981, 1983), but in agreement with more recent results from North America showing similar wear tolerance in the two species (Minner and Valverde 2005, Stier et al. 2008).

Although the sod used in these trials was of high quality, our results confirm the general experience that AB is more of a problem on pitches dominated by KB than on pitches dominated by PR (e.g. Canaway et al. 1986). AB was most conspicuous on KB plots when flowering during the first part of the growing season, in particular on control plots without overseeding and plots that been overseeded with KB in spring. Except when followed by overseeding of PR that was able to compete with AB in establishment rate at low soil temperature (Larsen and Bibby 2005), our data suggest that the spring aeration treatment favored AB and thus had a negative effect on the botanical composition on KB plots (Grime and Hunt 1975, Vargas and Turgeon 2004).

Overseeding with PR

One concern when overseeding PR into KB is whether PR will become too dominant and make the pitch more susceptible to winter damage (Koski 2000, Lyons and Elford 2009). In our trials we did see a reduction in turf cover on KB plots at Kvithamar and, to a lesser extent, at Landvik in spring 2011 after overseeding PR in the previous years, but the maximal reduction of approximately 20% units was not dramatic. Although PR contributed about one half of the total tiller population at Landvik and Kvithamar in 2010, KB tillers were still present in sufficient numbers to ensure a smooth transition into the next growing season. In other words, the positive effect of PR in preventing AB ingress was clearly greater than the negative effect of PR as a competitor against KB. While there is no doubt that PR is more competitive than KB at the seedling stage, the relationship may well be opposite for mature plants of the same species. After comprehensive studies on shoot and root competition between KB, PR and AB, Brede and Duich (1986) concluded that one of the advantages of KB was its ability to maintain consistent growth while sustaining interspecific interaction.

Because of its competiveness at the seedling stage, PR was able to establish itself in the sodded KB sward regardless of seeding time. On average for 2010 and 2011 at Landvik and Kvithamar, PR contributed approximately the same proportion of the sward after being overseeded in spring, summer, or fall. Given the harsh winter conditions during the last two years of the project, an important and perhaps surprising finding was that seedlings and mature plants of PR were equally susceptible to winter damage. The only stage at which PR survived the winter 2010/11 at Kvithamar was as ungerminated seeds, but this appears to be an uncertain method given the predictions for a more unstable winter climate. Trials in Wisconsin showed spring seeding to produce better turf with less weeds than seeding in late fall (Stier et al. 2008), and trials in Iowa showed seeding of PR at the beginning of the playing season to produce more consistent turf cover than splitting and cleating in the same amount of total seed during game events (Valverde and Minner 2008). While it may be a good idea to apply a certain amount of seed in conjunction with every aeration and topdressing treatment, the safest time for overseeding PR into KB pitches in Scandinavia is probably in spring, as this will maximize the benefit of the seeding effort throughout the playing season.

Overseeding with KB

The effect of overseeding KB into existing KB swards was inconsistent with lower tiller densities being recorded at Kvithamar in 2010 and significantly higher tiller densities at Landvik in 2011. Overseeding KB into PR in summer 2010 and fall 2011 resulted in flushes of KB tillers at Landvik, but most of these tillers were seedlings that never made it into adult plants. In other words, the problem was not the germination phase, but rather that the weak and spindly KB seedlings were outcompeted by surrounding turf. Similar observations were reported by Brede (1985) and Koski and Newberry (2004). Given the high seeding rate used in these trials (30 g m⁻² corresponding to 8–10 KB seeds cm⁻²), the results of overseeding KB were altogether disappointing, as also shown by the fact that repeated overseeding over three years had no effect on winter survival on plots initially seeded with PR. The best prospects for KB to establish itself in PR are probably with overseeding in the late summer or fall before a winter that severely weakens the PR. Overseeding of KB in early spring ought to be avoided, partly because KB is very slow in germination at low soil temperatures (Larsen and Bibby 2005) and partly because spiking or verticutting at this time of the year may stimulate AB encroachment (Vargas and Turgeon 2004).

Acknowledgements

This research was funded by grants from The Norwegian Ministry of Culture and The Norwegian Football Association. We thank Anne A. Steensohn and Ove Hetland for help with botanical analyses and Agnar Kvalbein for constructive criticism of this manuscript.