Effect of seed mix, sowing time, summer fallow, site location and aspect on the establishment of sown pasture species on uncultivable hill country

ABSTRACT

This study aimed to determine which practices resulted in the greatest establishment (plant density) of a range of grasses, legumes and herbs at four geo-climatically different hill country sites in the North Island (three sites) and South Island (one site) of New Zealand. At each site, on a north and south aspect, a seed mix containing either grasses, legumes and herbs (GLH) or legumes only (LEG), was sown in spring 2011 and autumn 2012. The 2011–2012 summer rainfall was greater than the long-term average at all sites. Establishment of total sown species was greater in the GLH than LEG mix (14% vs 8% of germinable seed sown) although some species (white clover [Trifolium repens]) established more successfully when sown in the LEG mix. Establishment was greater from spring than autumn sowing at the two wetter sites, but greater from autumn than spring sowing at the two drier sites. Greater establishment occurred after a summer fallow or summer crop than after an autumn sowing for most species at most sites. Establishment of total sown species ranged from 5%–39%, depending on the site and treatment. Seedling establishment, as a percentage of germinable seed sown, was greatest for perennial ryegrass (Lolium perenne), white clover, red clover (T. pratense) and plantain (Plantago lanceolata) (up to 36%, 37%, 33% and 35%, respectively, averaged over sites), while cocksfoot (Dactylis glomerata), phalaris (Phalaris aquatica), grazing brome (Bromus stamineus), lotus (Lotus uliginosus), birdsfoot trefoil (Lotus corniculatus) and chicory (Cichorium intybus) were below 15%.

KEYWORDS:

• Hill country
• pasture establishment
• pasture improvement
• pasture renewal
• seed mixes

Introduction

Steep, non-arable hill country below an altitude of 1000 m comprises 40% of New Zealand’s land area and makes a significant contribution to the national economy (Blaschke et al. 1992). In the North Island, pastoral hill country covers 3.5 million ha, which is equivalent to 28% of farmland in New Zealand (Mackay et al. 1993). Over decades, substantial pastoral development of these areas has occurred, including forest clearing, introduction of pasture species and enhancement of soil fertility (Suckling 1966; Levy 1970; White 1973).

More recent changes in land use, such as the expansion of dairy enterprises, mean that hill country is becoming increasingly important for finishing lambs and beef cattle. This has increased the need for high quality feed, with the aim of increasing live weight gains of lambs and rising 2-year-old cattle before weaning (Bray et al. 2013). Once improvements such as subdivision and enhanced soil fertility have been achieved, a key approach to improving the feed supply is through establishment of novel germplasm (Bray et al. 2013). A large range of herbaceous (grass, legume, herb) and shrub species were evaluated on hill and high country between the 1960s and 1980s, with a focus on legume establishment to supply biologically fixed nitrogen to existing or introduced non-leguminous species (Suckling 1966; Chapman & Macfarlane 1985; Lambert et al. 1985; Scott et al. 1985). With new germplasm becoming available, and especially commercial forage herbs such as chicory and plantain (Stewart et al. 2014), there is an opportunity to revisit pasture establishment practices to determine which methods are most successful in different regions and climates.

Site characteristics have a large impact on establishment success through effects on availability of moisture and solar radiation. These vary greatly with aspect and slope. North aspects typically have lower soil water content than southern aspects, especially during summer. For example, potential evapotranspiration was greater on north- than south-facing slopes at Woodville, southern Hawke’s Bay, in early autumn, winter, spring and summer (Lambert & Roberts 1976). On north-facing slopes at Whatawhata in Waikato (average annual rainfall 1640 mm, summer-moist hill country), soil water content (0–7.5 cm depth) was below wilting point from mid-summer until early autumn over a 3 year period (Gillingham 1973). Soil moisture may also be replenished more slowly on north aspects in autumn, leading to slower germination and seedling emergence. Moisture deficits may explain why mortality of establishing seedlings is particularly great over the first summer, especially when combined with competition from resident species (Campbell 1968, 1992).

In New Zealand, summer-safe (or summer-wet) country is considered to be that in which a moisture deficit over summer is uncommon. Regions considered to be summer-safe usually receive an annual rainfall of more than 1500 mm (e.g. south Waikato). This is in contrast to many summer-dry regions, such as Canterbury and Hawke’s Bay, which regularly experience soil moisture deficits from late spring through to autumn and have total annual rainfalls of less than 1000 mm (Moot et al. 2009). These categories have been used in the paper to denote summer-dry (<1000 mm average annual rainfall), summer-moist (1000–1500 mm) and summer-wet (>1500 mm) regions.

Cullen (1971) concluded that the amount and density of the resident vegetation at oversowing was the most critical factor influencing establishment of new pastures. By suppressing or killing resident vegetation, herbicide application can improve seed–soil contact and reduce shading of and competition with establishing seedlings. A range of herbicides has been investigated for use in New Zealand hill country, including paraquat and diquat, dicamba, 2,2 DPA, 2, 4-D and glyphosate, applied in autumn and in spring (Tozer & Douglas 2016). The greatest establishment occurs where the most weeds are killed for the longest period (Campbell 1992). The benefit of herbicide application for establishment was demonstrated by MacFarlane (1985) who found that an aerial application of paraquat and/or diquat increased white clover establishment by 200%–300% when compared with no herbicide application. However, excessive use of herbicide can denude an area such that establishing seeds are more vulnerable to microclimate extremes. For example, Chapman and Fletcher (1985) found that seedling mortality, particularly that of white clover, was substantial in plots in which the resident vegetation was suppressed the most, possibly due to factors such as exposure to wind and surface drying. Thus, either insufficient or excessive removal of resident vegetation can be detrimental to early establishment.

To determine which practices resulted in the greatest establishment of two pasture mixtures containing a range of pasture species, trials were established on north and south aspects at four sites in the North and South Islands of New Zealand. At each site there were eight treatments comprising: seed mix (grasses, legumes and herbs vs legumes only), season (spring vs autumn), chemical fallow over summer (with vs without) and cropping over summer (with vs without). Each treatment was replicated twice within each block in preparation for another treatment which was not imposed until after the time period considered in this paper. Sites were selected to provide a variety of soil type, topographic and climatic conditions, and species were selected to represent those commercially available. We focus on seedling establishment 6 weeks after sowing, to identify those species best suited to establish in the range of climates and landscape types represented in this study, and factors associated with their establishment. We hypothesised that while trends in the establishment of sown grasses, legumes and herbs would be specific to individual sites, some trends in establishment would be common across all sites.

Methods

Sites

Trials were established at three North Island sites and one South Island site (Ngaroma, South Waikato; Poukawa, central Hawke’s Bay; Woodville, southern Hawke’s Bay; Cheviot, North Canterbury, Figure 1, Table 1) in spring 2011 (September) and autumn 2012 (March). Long-term annual mean summer rainfall (December–February) was 143–346 (Table 1) based on the National Institute of Water and Atmospheric Research (NIWA) virtual climate station database records (Tait et al. 2006). Sites at Ngaroma and Cheviot were on commercial sheep and beef properties, whereas the Woodville site was at AgResearch’s Ballantrae Hill Country Research Station and the Poukawa site was at the Poukawa Research Station. During summer 2011–2012, rainfall was greater than average at all sites (Table 1). Monthly rainfall data over the experimental period are shown in Figure 2. Average long-term (approx. 30 years) potential soil water deficits in summer ranged from 28 mm (Ngaroma) to 45 mm (Cheviot) based on data obtained from the NIWA virtual climate station database (Tait et al. 2006).

Figure 1. Locations of sites in which a range of pasture establishment methods were compared: Ngaroma (South Waikato, summer-wet), Poukawa (central Hawke’s Bay, summer-dry), Woodville (southern Hawke’s Bay, summer-moist) and Cheviot (North Canterbury, summer-dry).

Figure 2. Monthly rainfall data for each of four sites—Ngaroma, Poukawa, Woodville and Cheviot—from September 2011 to June 2012. The arrows denote when the seed mixes were sown.

Table 1. Site location, climate categorisation, soil type, topography and soil fertility status at the time of sowing for four sites, averaged over the north and south aspects. All sites were located on sheep breeding and finishing farms.

Characteristic	Ngaroma	Poukawa	Woodville	Cheviot
Latitude	−38.270	−39.775	−40.325	−42.825
Longitude	175.565	76.725	175.875	73.275
Climate category	Summer-wet	Summer-dry	Summer-moist	Summer-dry
Long-term 30 yr summer rainfall (mm)	346	154	269	143
Summer rainfall 2011–2012 (mm)^a	432	207	284	197
Total annual rainfall (mm)	1663	712	1176	776
Soil type	Ngaroma silt loam	Matapiro silt loam	Kumeroa hill soils	Amberley Te papa silt loam
Slope angle (°)	28	25	29	28
Olsen phosphate (µg mL^−1)	45	24	31	29
Potassium (MAF QT)^b	6	21	9	18
Sulphate sulphur (ppm)	28	11	8	6
pH (H20)	5.3	6.0	5.5	5.5

^aSummer: December–February.

^bMAF QT (Quick Test) units (Cornforth 1980; Cornforth & Sinclair 1984).

Table 2. Eight treatments applied at each of four sites in New Zealand.

Treatment	Spring 2011		Autumn 2012
	Herbicide applied	Seed mix sown	Herbicide applied	Seed mix sown
Spring sown
Standard-GLH	✓	GLH
Standard-LEG	✓	LEG
Autumn sown
Standard-GLH			✓	GLH
Standard-LEG			✓	LEG
Fallow-GLH	✓		✓	GLH
Fallow-LEG	✓		✓	LEG
Crop-GLH	✓	Summer brassica	✓	GLH
Crop-LEG	✓	Summer brassica	✓	LEG

GLH, grasses, legumes and herbs; LEG, legumes only.

Seed mixes were sown in spring or autumn.

Herbicide was applied only in spring or autumn before sowing in the standard treatment, and in both spring and the following autumn for the ‘summer fallow’ and ‘summer crop’ treatments.

Sites were selected that had slope angles averaging 28° and moderate to high soil nutrient status with a pH of 5.3–6.0, Olsen P of 24–45 µg mL⁻¹, K of 6–21 (MAF QT [Cornforth 1980; Cornforth & Sinclair 1984]) and SO₄-S of 6–28 ppm depending on the site (Table 1).

Treatments

At each of the four sites, on a north and a south aspect, seed mixes containing: (1) grasses, legumes and herbs (GLH); and (2) legumes only (LEG) were sown in spring 2011 (September) and autumn 2012 (March) (Table 2). On each aspect, treatments were arranged in four randomised complete blocks comprising 10 × 10 m plots (experimental units). There was no replication of aspects at an individual site. The GLH mix comprised: perennial ryegrass (Lolium perenne L.) cv. ‘Grasslands Samson’ infected with AR37 endophyte; cocksfoot (Dactylis glomerata L.) cv. ‘Grasslands Tekapo’; phalaris (Phalaris aquatica L.) cv. ‘Grasslands Maru’; grazing brome (Bromus stamineus Desv.) cv. ‘Grasslands Gala’; white clover (Trifolium repens L.) cv. ‘Grasslands Bounty’; red clover (T. pratense L.) cv. ‘Grasslands Sensation’; subterranean clover (T. subterraneum L.) cv. ‘Denmark’; plantain (Plantago lanceolata L.) cv. ‘Ceres Tonic’; and chicory (Cichorium intybus L.) cv. ‘Puna II’. The LEG mix comprised the same cultivars of white clover, red clover and subterranean clover included in the GLH mix, and additionally birdsfoot trefoil (Lotus corniculatus L.) cv. ‘Grasslands Goldie’ and lotus (Lotus uliginosus Schkuhr syn. L. pedunculatus Cav.) cv. ‘Grasslands Trojan’ (Table 3). Laboratory germination was between 89% and 98%, when seeds were maintained at 16–20 °C on moistened filter paper for 12 days. All legume seed was inoculated with appropriate Rhizobium strains.

Table 3. Sowing rate, seed weight, germination percentage and germinable seed density of species included in the grasses, legumes and herbs seed mix and legume only seed mix sown in a range of treatments at four sites throughout New Zealand.

Seed mix	Species	Sowing rate (kg ha^−1)	Seed weight (g 1000 seeds^−1)	Seed weight sown (g m^−2)	Sown seed density (seeds m^−2)	Germination (%)	Germinable seed density (seeds m^−2)
GLH	Perennial ryegrass	7.0	2.0	0.70	350	96	336
	Cocksfoot	1.5	0.9	0.15	167	92	153
	Phalaris	1.5	2.0	0.15	75	94	71
	Grazing brome	8.0	11.5	0.80	70	89	62
	Chicory	0.5	1.2	0.05	42	92	38
	Plantain	0.5	2.0	0.05	25	90	23
	White clover	2.0	0.7	0.20	286	98	280
	Red clover	3.0	3.4	0.30	88	97	86
	Sub clover	4.0	6.7	0.40	60	98	59
	Total	28.0		2.80	1163		1108
LEG	White clover	1.0	0.7	0.10	143	98	140
	Red clover	4.6	3.4	0.46	135	97	131
	Sub clover	9.3	6.7	0.93	139	98	136
	Birdsfoot trefoil	1.6	1.2	0.16	133	90	120
	Lotus	0.7	0.5	0.07	140	91	127
	Total	17.2		1.72	690		654

GLH, grasses, legumes and herbs; LEG, legumes only.

The pasture mixtures were established using a range of methods involving season, summer fallow and summer cropping. For standard sowing, plots were sprayed by hand with glyphosate herbicide at an average rate of 2.0 kg ai ha⁻¹, 7–10 days before sowing in spring 2011 and autumn 2012 to kill resident swards. The seed mixes were broadcast by hand and livestock (500 sheep ha⁻¹ for 3 h) were used to trample seed into the ground within several days of sowing to increase seed–soil contact using the method suggested by Hampton et al. (1999). Thereafter, livestock were excluded from all sites.

For summer cropping (Table 2), glyphosate was applied to the swards (average of 2.0 kg ai ha⁻¹) in spring 2011, forage rape (Brassica napus L.) sown at a rate of 4 kg ha⁻¹ in mid-late spring (October or November 2011 depending on the site) and the GLH or LEG mix subsequently hand-broadcast in autumn 2012. For summer fallow, glyphosate was applied as described above in spring (October) 2011 and subsequently in autumn (March) 2012.

Measurements and analysis

Seedlings were counted 6–8 weeks after sowing in 10, 0.2 m² quadrats (0.5 × 0.4 m) randomly positioned within each plot. Data were expressed as seedlings m⁻² and the density of each species established was calculated as percentage of germinable seed sown m⁻². Percentage establishment data are tabulated and reported in the text while seedling density data are reported in the text only.

For each site × aspect × season combination, soil temperatures at 0900 h were recorded at 10 cm depth (Tinytag Transit 2 TG-4080 data loggers, Gemini Scientific) from sowing until seedling establishment. Occasional missing values, due to data logger failure, were replaced with estimated values. These were obtained by fitting linear models to the logger data using as covariates spatially interpolated temperature data from the NIWA virtual climate station database (Tait et al. 2006). This allowed us to predict with confidence the temperature on days with missing data.

Data were analysed by split-plot analysis of variance using GenStat, 13th edition (GenStat 2010). Treatments fitted in the analyses were two aspects (north, south), four establishment methods (spring standard, autumn standard, autumn fallow and autumn crop), two seed mixes (GLH, LEG) and all interactions. Season comparisons were performed using the standard establishment methods only in the analysis, and fallow comparisons were performed using only the autumn standard and autumn fallow establishment methods. Each site was analysed separately and a combined analysis with all sites. For individual site analyses, the aspect treatment was applied to the main plots (hillsides), with the establishment method and seed mixture treatments applied to the subplots. When all sites were included in the analysis, a split-split-plot analysis was performed with site as the top level, aspect applied within site, and the establishment methods and seed mixture treatments applied within aspect. There was no replication of aspect within a site. Mean separation was assessed by Fisher’s protected least significant difference (LSD). Only interactions with plausible biological explanations are reported.

Simple correlations were calculated between establishment of perennial ryegrass, white clover, total grasses, total legumes, total grasses and total sown species, and mean temperature of the soil at 10 cm depth.

Results

Analyses of all sites

Aspect

Perennial ryegrass and birdsfoot trefoil percentage establishment (hereafter referred to as establishment throughout the Results section) was greater on the south than north aspect (perennial ryegrass: 31% vs 18%, birdsfoot trefoil: 5% vs 4%, Table 4) equating to seedling densities of 104 vs 59 (perennial ryegrass) and 7 vs 5 seedlings m⁻² (birdsfoot trefoil).

Table 4. Effects of aspect (north, south), seed mix, season (sown in spring or autumn) and summer fallow (with or without) on percentage establishment (% of germinable seed sown) for each of the sown grasses, legumes and herbs.

Establishment (%)	Aspect^a				Seed mix^a				Season^a				Fallow^b
	North	South	SED	F prob	GLH	LEG	SED	F prob	Spring	Autumn	SED	F prob	No	Yes	SED	F prob
Perennial ryegrass	18	31	4.0	0.043	24	–			13	36	9.8	0.032	36	22	7.6	0.075
Cocksfoot	2	3	1.2	ns	2	–			2	3	1.1	ns	3	7	0.7	<0.001
Phalaris	12	4	8.7	ns	8	–			6	9	3.7	ns	9	11	3.5	ns
Grazing Brome	13	5	10.3	ns	9	–			7	11	6.4	ns				–^c
White clover	10	14	1.8	ns	8	17	2.9	0.004	12	13	2.9	ns	13	26	2.8	<0.001
Red clover	16	22	7.5	ns	25	14	5.5	0.041	25	14	5.5	0.067	14	17	2.3	ns
Subterranean clover	5	6	1.9	ns	7	4	1.1	0.005	7	5	1.1	ns	5	7	1.8	ns
Lotus	4	6	1.0	0.087	–	5			6	3	1.5	0.041	3	8	1.4	0.002
Birdsfoot trefoil	4	5	0.3	0.042	–	5			4	6	1.4	0.097	6	10	1.4	0.017
Chicory	9	11	3.4	ns	10	–			13	7	2.7	0.063	7	12	2.4	0.040
Plantain	14	29	8.3	ns	22	–			26	18	8.4	ns	18	35	5.7	0.005
Sown grasses	13	18	2.1	0.068	16	–			9	22	5.7	0.030	22	15	4.2	ns
Sown legumes	8	11	1.3	ns	11	8	2.0	0.073	11	8	2.0	ns	8	14	1.2	<0.001
Sown herbs	11	18	5.1	ns	14	–			17	11	4.6	ns	11	21	3.2	0.007
Total sown	9	13	1.3	0.073	14	8	2.5	0.017	10	12	2.5	ns	12	14	1.8	ns

GLH, grasses, legumes and herbs; LEG, legumes only; SED, standard error of difference.

‘–’ denotes species not sown; ‘ns’ denotes means not significantly different (P  > 0.1).

Means in bold are significantly different (P  < 0.05).

Assessments were made 6–8 weeks after sowing.

Results are based on ANOVA of four sites.

^aData from the ‘standard’ treatment are used for aspect, seed mix and season comparisons because it was the only treatment sown in both autumn and spring.

^bComparison of standard vs fallow treatments sown in autumn.

^cSeedling densities were low and there were insufficient data to be analysed.

Seed mix

White clover establishment was greater from the LEG than GLH mix (17% vs 8%), whereas red clover, subterranean clover and total sown establishment were greater in swards established from the GLH than LEG mix (total sown: 14% vs 8%, Table 4). However, the density of white clover, red clover and subterranean clover seedlings was similar in the GLH and LEG mix, averaging 23, 20 and 5 seedlings m⁻², respectively (P  > 0.05). This equated to a total sown species seedling density of 153 vs 60 seedlings m⁻² for the GLH and LEG seed mixes, respectively (P  < 0.001).

Season

Perennial ryegrass (36% vs 13%) and total sown grass establishment (22% vs 9%) were greater from autumn than spring sowing equating to densities of 120 vs 44 perennial ryegrass seedlings m⁻² and 138 vs 55 total sown grass seedlings m⁻² (P  < 0.05). Establishment of lotus was greater from spring than autumn sowing (6% vs 3%, Table 4), equating to densities of 8 vs 4 lotus seedlings m⁻² (P  < 0.05). Establishment of red clover was similar from spring and autumn sowing on the north aspect but greater from spring than autumn sowing on the south aspect (aspect × season interaction, Table 5).

Table 5. Significant first-order interactions between aspect (north, south) and seed mix, season (sown in spring or autumn) or summer fallow (with or without) on percentage establishment (% of germinable seed sown) of sown grasses, legumes and herbs.

Interaction	Site	Species	North aspect		South aspect		SED	F prob
Aspect × seed mix			GLH	LEG	GLH	LEG
	Woodville	White clover	10	16	19	39	8.8	0.024
Aspect × season^a			Spring	Autumn	Spring	Autumn
	All sites	Red clover	17	12	33	12	15.6	0.048
	Ngaroma	Subterranean clover	<1	<1	7	<1	1.3	<0.001
	Woodville	White clover	30	6	22	15	9.0	0.010
		Subterranean clover	8	<1	2	5	3.8	<0.001
		Birdsfoot trefoil	10	1	3	11	3.3	<0.001
	Cheviot	Grazing brome	1	79	1	1	15.7	<0.001
		Red clover	5	43	23	8	15.2	<0.001
		Sown legumes	3	14	11	5	4.3	<0.001
		Chicory	6	12	10	3	3.2	<0.001
		Sown herbs	5	12	9	5	3.5	0.008
		Total sown	2	15	7	10	3.7	<0.001
Aspect × fallow^b			No fallow	Fallow	No fallow	Fallow
	All sites	Lotus	3	4	4	12	2.7	0.024
	Poukawa	Sown herbs	27	21	42	21	9.6	0.007
		Total sown	22	18	29	14	5.9	0.007
	Woodville	Perennial ryegrass	3	7	8	28	5.6	0.009
		Cocksfoot	1	2	<1	7	2.1	0.014
		Sown grasses	2	5	5	17	3.7	0.017
		White clover	6	20	15	43	8.8	0.039
		Lotus	<1	2	5	12	2.2	0.049
		Plantain	<1	<1	10	47	8.6	<0.001
		Sown herbs	2	10	8	26	4.1	0.013
		Total sown	2	7	8	19	3.4	0.021
	Cheviot	Phalaris	2	3	5	14	3.0	0.012
		Grazing brome	79	35	<1	<1	19.6	0.034
		Lotus	3	4	6	23	4.6	0.002
		Birdsfoot trefoil	7	8	6	17	4.0	0.030
		Sown legumes	14	15	5	20	5.0	<0.001
		Total sown	15	14	10	19	4.3	0.002

GLH, grasses, legumes and herbs; LEG, legumes only; SED, standard error of difference for comparisons within an aspect.

Assessments were made 6–8 weeks after sowing.

Means in bold are significantly different for comparisons within an aspect within each row (P  < 0.05).

SED and F prob values are for the interactions between aspect × seed mix, aspect × season and aspect × fallow.

^aComparison of standard treatments only.

^bAutumn sown treatments only.

Fallow

Summer fallow had the greatest impact on establishment, increasing the establishment of cocksfoot, white clover, lotus, birdsfoot trefoil, chicory, plantain, sown legumes and sown herbs (Table 4). Densities were 83 vs 46 for sown legume seedlings m⁻² and 13 vs 7 for sown herb seedlings m⁻² with and without a fallow, respectively (P  < 0.01). Summer fallow increased lotus establishment on the south aspect but had no effect on the north aspect (aspect × fallow interaction, Table 5).

Among species, the greatest establishment was achieved by perennial ryegrass (36% of seed sown equating to 120 seedlings m⁻²), plantain (35%, 8 seedlings m⁻²), white clover (26%, 48 seedlings m⁻²) and red clover (25%, 25 seedlings m⁻²). Establishment of lotus, birdsfoot trefoil and subterranean clover was below 10%.

Comparison of the eight treatments

Establishment of perennial ryegrass was almost three-fold greater in the autumn-standard-GLH than spring-standard-GLH (36% vs 13%, Table 6). Establishment of lotus in autumn-fallow-LEG was not significantly different from that of spring-standard-LEG. Cocksfoot establishment was greater in the autumn-fallow-GLH and autumn-crop-GLH (7%) than spring-standard-GLH (2%) and autumn-standard-GLH (3%). Sown grass establishment was greater in the autumn-standard-GLH than spring-standard-GLH (22% vs 9%, equating to 138 vs 55 seedlings m⁻²).

Table 6. Effects of season (sown in spring or autumn), summer fallow (with or without) and seed mix on percentage establishment (% of germinable seed sown) for each of the sown grasses, legumes and herbs.

Season	Spring 2011		Autumn 2012
Herbicide application	Standard		Standard		Fallow		Crop
Seed mix	GLH	LEG	GLH	LEG	GLH	LEG	GLH	LEG	SED	F prob
Perennial ryegrass	13b	–	36a	–	22a,b	–	24a,b	–	8.7	0.023
Cocksfoot	2b	–	3b	–	7a	–	7a	–	1.1	<.001
Phalaris	6	–	9	–	11	–	14	–	3.8	ns
Grazing brome	7	–	11	–	7	–	5	–	5.5	ns
White clover	10b,c	14b,c	6c	19b	16b	37a	14b,c	33a	4.8	0.023
Red clover	33	16	18	11	21	13	23	13	7.5	ns
Subterranean clover	8	5	6	3	10	5	14	4	2.8	ns
Lotus	–	6a,b	–	3b	–	8a	–	7a	2.0	0.046
Birdsfoot trefoil	–	4c	–	6b,c	–	10a	–	8a,b	1.7	<.001
Chicory	13	–	7	–	12	–	13	–	2.7	0.058
Plantain	26	–	18	–	35	–	34	–	8.1	0.062
Sown grasses	9b	–	22a	–	15a,b	–	17a,b	–	5.0	0.023
Sown legumes	14	8	8	7	16	13	16	11	2.0	ns
Sown herbs	17a,b	–	11b	–	21a	–	20a	–	4.4	0.050
Total sown	11	8	16	7	16	13	17	11	3.3	ns

GLH, grasses, legumes and herbs; LEG, legumes only; SED, standard error of difference for comparisons within each row.

‘–’ denotes species not present; ‘ns’ denotes not significantly different (P > 0.1).

Means in bold with different subscripts are significantly different for comparisons within a row (P < 0.05).

Results are based on ANOVA of four sites and are averaged over aspects.

Assessments were made 6–8 weeks after sowing.

Establishment of white clover, lotus and birdsfoot trefoil was greatest in autumn-fallow-LEG although it was not significantly different from autumn-crop-legume (Table 6). Establishment was lowest in the autumn-standard-GLH for white clover, autumn-standard-LEG for lotus and spring-standard-LEG for birdsfoot trefoil.

Establishment of sown herbs was greatest in autumn-fallow-GLH and autumn-crop-GLH and lowest in autumn-standard-GLH (21, 20% vs 11%).

Analyses of individual sites

Seed mix

Establishment of white clover was greater in the LEG than GLH mix at all sites (Table 7), with densities ranging from 11 to 38 white clover seedlings m⁻². However, at Woodville, establishment of white clover was greater in the LEG than GLH mix on the south aspect but similar in both seed mixes on the north aspect (aspect × seed mix interaction, Table 5).

Table 7. Effects of seed mix, season (sown in spring or autumn) and summer fallow (with or without) on percentage establishment (% of germinable seed sown) of sown grasses, legumes and herbs at four sites.

Site	Establishment (%)	Seed mix^a				Season^a				Fallow^b
		GLH	LEG	SED	F prob	Spring	Autumn	SED	F prob	No	Yes	SED	F prob
Ngaroma	Perennial ryegrass	16	–			16	15	2.8	ns	15	19	3.4	ns
	Cocksfoot	3	–			5	1	0.7	<0.001	1	8	1.4	<.001
	Phalaris	4	–			5	2	1.9	ns	2	12	2.8	0.002
	Grazing brome	3	–			4	1	1.2	0.008	1	3	1.6	ns
	White clover	4	10	1.6	<0.001	10	4	1.6	<0.001	4	19	2.5	<0.001
	Red clover	10	8	1.6	ns	15	3	1.6	<0.001	3	18	2.0	<0.001
	Subterranean clover	2	2	0.5	ns	5	0	0.5	<0.001				-^c
	Lotus	–	5			7	3	1.9	ns	3	5	1.8	ns
	Birdsfoot trefoil	–	6			5	8	3.4	ns	8	8	3.4	ns
	Chicory	11	–			18	4	1.8	<0.001	4	16	2.6	<0.001
	Plantain	18	–			31	5	5.2	<0.001	5	41	6.6	<0.001
	Sown grasses	10	–			11	9	1.8	ns	9	14	2 . 2	0.028
	Sown legumes	5	6	0.8	ns	8	2	0 . 8	<0.001	3	11	1 . 1	<0.001
	Sown herbs	13	–			23	4	2 . 6	<0.001	4	25	3 . 6	<0.001
	Total sown	8	6	0 . 9	0.009	9	5	0 . 9	<0.001	5	11	1 . 1	<0.001
Poukawa	Perennial ryegrass	52	–			9	94	10.9	<0.001	94	29	9.3	<0.001
	Cocksfoot	5	–			2	8	0.7	<0.001	8	11	1.5	ns
	Phalaris	24	–			18	30	11.0	ns	30	19	10.0	ns
	Grazing brome	5	–			10	1	1.6	<0.001				-^c
	White clover	8	22	4.6	0.005	4	26	4.6	<0.001	26	29	5.0	ns
	Red clover	14	9	3.1	0.086	5	18	3.1	<0.001	18	11	2.9	0.019
	Subterranean clover	15	8	3.2	0.020	11	12	3.2	ns	12	13	5.2	ns
	Lotus	–	3			3	3	1.0	ns	3	7	1.9	0.077
	Birdsfoot trefoil	–	2			2	3	0.8	0.093	3	11	2.5	0.004
	Chicory	10	–			7	13	4.7	ns	13	8	4.2	ns
	Plantain	32	–			13	50	6.5	<0.001	50	42	9.5	ns
	Sown grasses	33	–			9	57	6 . 0	<0.001	57	21	5 . 2	<0.001
	Sown legumes	10	8	1.7	0.155	5	13	1 . 7	<0.001	13	14	2.0	ns
	Sown herbs	18	–			9	27	4 . 5	<0.001	27	21	4.8	ns
	Total sown	23	8	2 . 2	<0.001	6	25	2 . 2	<0.001	25	16	2 . 1	<0.001
Woodville	Perennial ryegrass	15	–			25	5	3.6	<0.001	5	17	2.8	<0.001
	Cocksfoot	1	–			1	1	0.3	0.045	1	4	1.1	0.001
	Phalaris				-^c				-^c				-^c
	Grazing brome	8	–			11	4	2.2	0.006	4	6	2.5	ns
	White clover	13	23	3.2	0.006	26	10	3.2	<0.001	10	32	3.1	<0.001
	Red clover	51	23	4.7	<0.001	64	10	4.7	<0.001	10	13	2.3	ns
	Subterranean clover	6	2	1.3	0.004	5	2	1.3	0.050	2	3	1.2	ns
	Lotus	–	9			15	3	2.5	<0.001	3	7	1.1	<0.001
	Birdsfoot trefoil	–	6			6	6	1.6	ns	6	6	0.9	ns
	Chicory	12	–			18	5	3.0	<0.001	5	13	2.1	<0.001
	Plantain	30	–			53	6	4.6	<0.001	6	26	4.3	<0.001
	Sown grasses	9	–			15	4	2 . 0	<0.001	4	11	1 . 9	<0.001
	Sown legumes	20	11	1 . 9	<0.001	25	6	1 . 9	<0.001	6	15	1 . 6	<0.001
	Sown herbs	18	–			31	5	3 . 1	<0.001	5	18	2 . 1	<0.001
	Total sown	14	11	1 . 3	0.026	19	5	1 . 3	<0.001	5	13	1 . 2	<0.001
Cheviot	Perennial ryegrass	15	–			2	29	5.1	<0.001	29	22	4.4	ns
	Cocksfoot	1	–			0	1	0.5	ns	1	6	1.2	<0.001
	Phalaris	2	–			1	3	1.7	ns	3	8	1.5	0.003
	Grazing brome	20	–			1	40	7.9	<0.001	40	18	9.8	0.036
	White clover	7	13	2.1	0.005	9	10	2.1	ns	10	27	3.1	<0.001
	Red clover	25	14	5.4	0.048	14	26	5.4	0.037	26	27	4.7	ns
	Subterranean clover	6	4	1.4	ns	5	5	1.4	ns	5	13	3.1	0.008
	Lotus	–	3			1	5	1.5	0.053	5	14	2.3	<0.001
	Birdsfoot trefoil	–	4			2	6	1.4	0.002	6	13	2.0	0.004
	Chicory	8	–			8	8	1.6	ns	8	12	3.0	ns
	Plantain	8	–			5	10	2.7	ns	10	31	7.7	0.010
	Sown grasses	11	–			1	20	3 . 0	<0.001	20	16	2.7	ns
	Sown legumes	10	7	1 . 5	0.025	7	10	1.5	ns	10	17	1 . 8	<0.001
	Sown herbs	8	–			7	8	1.8	ns	8	19	4 . 3	0.021
	Total sown	10	7	1 . 3	0.009	5	12	1 . 3	<0.001	12	17	1 . 5	0.003

GLH, grasses, legumes and herbs; LEG, legumes only; SED, standard error of difference for comparisons within seed mix, season or fallow.

‘–’ denotes species not present; ‘ns’ denotes means not significantly different (P  > 0.1).

Means in bold are significantly different (P  < 0.05).

Results are averaged over aspects. Assessments were made 6–8 weeks after sowing.

Ngaroma: 1663 average annual rainfall, 346 average summer rainfall, silt loam soils; Poukawa: 712 mm, 154 mm, silt loam soils; Woodville: 1176 mm, 269 mm, hill soils; Cheviot: 776 mm, 143 mm, silt loam soils.

^aData from the ‘standard’ treatment are used for aspect, seed mix and season comparisons because it was the only treatment sown in both autumn and spring.

^bComparison of standard vs fallow treatments sown in autumn.

^cSeedling densities were low and there were insufficient data to be analysed.

Establishment was greater in the GLH than LEG mix for red clover at Woodville and Cheviot, and for subterranean clover at Poukawa and Woodville. Densities ranged from 9 to 44 red clover seedlings m⁻² and 1 to 10 subterranean clover seedlings m⁻². Establishment of total sown legumes was either similar in the two seed mixes or greater in the GLH than LEG mix (Woodville: 20% vs 11%; Cheviot: 10% vs 7%). Establishment of total sown species was consistently greater across all sites in the GLH than LEG mix (Table 7). Seedling densities ranged from 21 to 85 sown legume seedlings m⁻² and 44 to 257 total sown species seedlings m⁻², with lowest legume and total sown species densities occurring at Ngaroma and the greatest at Woodville (sown legumes) and Poukawa (total sown species).

Season

Season had a large influence on establishment, which varied between the wetter and drier sites (Table 7). Establishment of grasses, legumes and herbs at the two wetter sites was greater from spring than autumn sowing but there were interactions between season and aspect at Ngaroma and Woodville. At Ngaroma, subterranean clover establishment was greater from spring than autumn sowing on the south aspect but negligible from both spring and autumn sowing on the north aspect (Table 5). At Woodville, establishment of white clover, subterranean clover and birdsfoot trefoil was greater from spring than autumn sowing on the north aspect but similar from spring and autumn sowing on the south aspect. In the case of birdsfoot trefoil, establishment was lower from spring than autumn sowing on the south aspect.

For both summer-dry sites (Poukawa, Cheviot), establishment of grasses, legumes and herbs was greater from autumn than spring sowing but there were also some consistent trends with interactions between season and aspect at Cheviot. The establishment of grazing brome, red clover, sown legumes, chicory, sown herbs and total sown species followed a general trend of greater establishment from autumn than spring sowing on the north aspect but either similar or greater establishment from spring than autumn on the south (Table 5). Another exception was grazing brome, which had greater establishment in spring than autumn at Poukawa (Table 7).

Seedling densities of most grass species were below 10 seedlings m⁻², irrespective of site, except for perennial ryegrass, which ranged from 2 seedlings m⁻² at Cheviot from spring sowing to as great as 94 seedlings m⁻² from autumn sowing at Poukawa (data not shown). Seedling densities of the two Lotus species and subterranean clover were also below 10 seedlings m⁻² irrespective of site, except for lotus at Woodville, which reached 15 seedlings m⁻² from spring sowing. Densities ranged from 6 to 51 white clover and 3 to 63 red clover seedlings m⁻² (with least and greatest densities at Ngaroma and Woodville, respectively). Of the sown herbs, densities ranged from 1 to 7 chicory and 1 to 12 plantain seedlings m⁻². Densities of total sown species ranged from 44 seedlings m⁻² at Ngaroma to 261 seedlings m⁻² at Poukawa (data not shown).

Fallow

A summer fallow increased establishment of most species at Ngaroma, Woodville and Cheviot (Table 7). However, it increased establishment of a range of grasses, legumes and herbs on the south aspect at Woodville and Cheviot but had no effect on the north aspect (aspect × fallow interaction, Table 5). The only exception was the establishment of grazing brome at Cheviot, which was reduced by a fallow on the north aspect but similar with and without a fallow on the south aspect. Total sown species seedling densities with and without a fallow, respectively, were 111 vs 44 seedlings m⁻² at Ngaroma, 126 vs 49 seedlings m⁻² at Woodville and 159 vs 120 seedlings m⁻² at Cheviot.

At Poukawa, summer fallow reduced establishment of perennial ryegrass. Ryegrass seedling densities were 99 vs 317 seedlings m⁻², with and without a fallow, respectively. The low ryegrass establishment was reflected in the lower sown grass and total sown species establishment with a fallow, equating to densities of 129 vs 352 sown grass, and 155 vs 261 total sown seedlings m⁻², with and without a fallow, respectively. A fallow reduced sown herbs and total sown species establishment on the south aspect by about 50% but had no effect on the north aspect (aspect × fallow interaction, Table 5).

Comparison of the eight treatments

At Ngaroma, establishment of most species was greatest when sown in autumn after either a fallow or crop and lowest when sown in autumn-standard-GLH (Table 8). This resulted in the greatest establishment in autumn-crop-GLH for sown grasses (18%) and sown herbs (26%). However, neither of these figures was significantly different from establishment in autumn-fallow-GLH. Lowest establishment for sown grasses (9%) and sown herbs (4%) occurred in autumn-standard-GLH (Table 8). Perennial ryegrass had the greatest establishment of the sown grasses (up to 27% equating to 89 seedlings m⁻²), red clover of the sown legumes (28%, 24 seedlings m⁻²) and plantain of the sown herbs (up to 41%, 9 seedlings m⁻²). Up to 16% of total sown species established, equating to 178 seedlings m⁻².

Table 8. Effect of season (sown in spring or autumn), summer fallow (with or without) and seed mix on percentage establishment (% of germinable seed sown) of sown grasses, legumes and herbs at four sites.

	Season	Spring 2011		Autumn 2012
	Herbicide application	Standard		Standard		Fallow		Crop
Site	Seed mix	GLH	LEG	GLH	LEG	GLH	LEG	GLH	LEG	SED	F prob
Ngaroma	Perennial ryegrass	16b	–	15b	–	19b	–	27a	–	3.8	0.028
	Cocksfoot	5b	–	1c	–	8a	–	8a	–	1.5	<.001
	Phalaris	5b	–	2b	–	12a	–	11a	–	2.9	<.001
	Grazing brome	4a,b	–	1b	–	3b	–	7a	–	1.8	0.012
	White clover	7	13	1	6	10	27	9	17	6.4	ns
	Red clover	16b,c	15b,c	4d,e	2e	24a,b	13c,d	28a	14c	4.9	0.015
	Subterranean clover	0	0	0	0	0	0	0	0	–	–
	Lotus	–	7	–	3	–	5	–	2	2.3	0.095
	Birdsfoot trefoil	–	5	–	8	–	8	–	8	3.5	ns
	Chicory	18a	–	4b	–	16a	–	17a	–	3.0	<.001
	Plantain	31a	–	4b	–	41a	–	41a	–	8.0	<.001
	Sown grasses	11b,c	–	9c	–	14a,b	–	18a	–	2.5	0.002
	Sown legumes	8	8	2	3	12	9	12	7	2.7	ns
	Sown herbs	22a	–	4b	–	25a	–	26a	–	4.2	<.001
	Total sown	11	8	6	3	14	9	16	7	2.6	0.063
Poukawa	Perennial ryegrass	9b	–	94a	–	29b	–	28b	–	10.1	<.001
	Cocksfoot	2b	–	8a	–	11a	–	11a	–	1.7	<.001
	Phalaris	18	–	30	–	19	–	36	–	11.4	ns
	Grazing brome	0	0	0	0	0	0	0	0	–	–
	White clover	3c	5c	13c	39a,b	15c	42a	17b,c	40a	11.1	0.023
	Red clover	6	4	23	13	15	7	14	9	6.5	ns
	Subterranean clover	15	8	16	8	14	11	39	11	10.8	ns
	Lotus	–	3c	–	3c	–	7b	–	12a	2.0	<.001
	Birdsfoot trefoil	–	2b	–	3b	–	11a	–	9a	2.8	<.001
	Chicory	7	–	13	–	8	–	7	–	4.9	ns
	Plantain	13b	–	50a	–	42a	–	54a	–	10.1	<.001
	Sown grasses	9c	–	57a	–	21b	–	22b	–	5.6	<.001
	Sown legumes	5	4	15	12	15	14	19	14	4.3	ns
	Sown herbs	9b	–	27a	–	21a	–	24a	–	5.6	0.003
	Total sown	7b,c	4c	39a	12b,c	18b	14b	21b	14b	4.7	<.001
Woodville	Perennial ryegrass	25a	–	5c	–	17a,b	–	12b,c	–	4.3	<.001
	Cocksfoot	1b	–	1b	–	4a	–	2a,b	–	1.1	<.001
	Phalaris	0	–	0	–	0	–	0	–	–	–
	Grazing brome	11a	–	4b	–	6a,b	–	3b	–	3.2	0.028
	White clover	22	29	5	16	24	39	13	38	8.5	ns
	Red clover	92a	36b	10c	11c	17c	8c	17c	11c	9.3	<.001
	Subterranean clover	7	3	5	0	4	2	3	0	2.7	ns
	Lotus	–	15a	–	3c	–	7b,c	–	8b	2.4	<.001
	Birdsfoot trefoil	–	6	–	6	–	6	–	7	1.6	ns
	Chicory	18a	–	5c	–	13a,b	–	10b,c	–	3.1	<.001
	Plantain	53a	–	6c	–	26b	–	14b,c	–	6.4	<.001
	Sown grasses	15a	–	3c	–	11a,b	–	8b,c	–	2.7	<.001
	Sown legumes	34a	15b,c	6c	6c	20b	11b,c	13b,c	11b,c	4.7	<.001
	Sown herbs	31a	–	5c	–	18b	–	12b	–	3.5	<.001
	Total sown	23a	15b	5c	6c	15b	11b,c	10b,c	11b,c	3.4	0.004
Cheviot	Perennial ryegrass	2b	–	29a	–	22a	–	30a	–	4.9	<.001
	Cocksfoot	<1b	–	1b	–	6a	–	7a	–	1.3	<.001
	Phalaris	1b	–	3b	–	8a	–	9a	–	2.4	<.001
	Grazing brome	1b	–	40a	–	18b	–	10b	–	10.6	<.001
	White clover	8b	10b	5b	15b	14b	40a	18b	38a	7.1	<.001
	Red clover	17	11	34	18	29	25	34	17	11.0	ns
	Subterranean clover	7b	4b	5b	5b	22a	5b	14a,b	6b	6.6	0.017
	Lotus	–	1b	–	5b	–	14a	–	5b	2.6	<.001
	Birdsfoot trefoil	–	2c	–	6b,c	–	13a	–	10a,b	2.2	<.001
	Chicory	8b	–	8b	–	12a,b	–	18a	–	3.3	0.016
	Plantain	5b	–	10b	–	31a	–	24a,b	–	7.8	<.001
	Sown grasses	1b	–	20a	–	16a	–	20a	–	3.1	<.001
	Sown legumes	10	5	11	8	18	17	20	13	4.2	ns
	Sown herbs	7b	–	8b	–	19a	–	20a	–	4.6	0.002
	Total sown	5d	5d	16a,b	8c,d	17a,b	17a,b	20a	13b,c	3.5	0.018

GLH, grasses, legumes and herbs; LEG, legumes only; SED, standard error of difference for comparisons within each row.

‘–’ denotes species not present; ‘ns’ denotes not significantly different (P  > 0.1).

Means in bold with different subscripts are significantly different for comparisons within a row (P  < 0.05).

Results are averaged over aspects.

Assessments were made 6–8 weeks after sowing.

Ngaroma: 1663 average annual rainfall, 346 average summer rainfall, silt loam soils; Poukawa: 712 mm, 154 mm, silt loam soils; Woodville: 1176 mm, 269 mm, hill soils; Cheviot: 776 mm, 143 mm, silt loam soils.

At Poukawa, establishment of most species was greatest from autumn sowing. Establishment was greatest in autumn-standard-GLH for perennial ryegrass and similar across all autumn sown treatments for the other grasses and for plantain. The greatest establishment of white clover, lotus and birdsfoot trefoil occurred in the autumn-LEG treatment, after either a fallow or crop. This resulted in an establishment percentage of 57% vs 9% for sown grasses, 27% vs 9% for sown herbs and 39% vs 7% for total sown species in autumn-standard-GLH and spring-standard-GLH, respectively (Table 8). Perennial ryegrass had the greatest establishment of the sown grasses (up to 94% equating to 317 seedlings m⁻²), white clover of the sown legumes (42%, 59 seedlings m⁻²) and plantain of the sown herbs (up to 54%, 12 seedlings m⁻²). Up to 39% of total sown species established, equating to 432 seedlings m⁻².

At Woodville, establishment of most species was greatest when sown in spring-standard-GLH and lowest in the autumn-standard-GLH. This resulted in establishment percentages of 15% vs 3% for sown grasses, 34% vs 6% for sown legumes, 31% vs 5% for sown herbs and 23% vs 5% for total sown species in spring-standard-GLH and autumn-standard-GLH, respectively (Table 8). Perennial ryegrass had the greatest establishment of the sown grasses (up to 25%, equating to 82 seedlings m⁻²), red clover of the sown legumes (92%, 79 seedlings m⁻²) and plantain of the sown herbs (up to 53%, 12 seedlings m⁻²). Up to 23% of total sown species established, equating to 256 seedlings m⁻² (Table 8).

At Cheviot, establishment of most species was greatest from autumn sowing. Establishment of total sown grasses did not vary significantly between the autumn standard, fallow and cropping treatments (Table 8). Establishment of sown herbs in the autumn-fallow-GLH and autumn-crop-GLH, 19% and 20%, respectively, was double that in the autumn-standard-GLH (8%) and spring-standard-GLH (7%). Perennial ryegrass had the greatest establishment of the sown grasses (up to 30% equating to 100 seedlings m⁻²), white clover of the sown legumes (up to 40%, 56 seedlings m⁻²) and plantain of the sown herbs (up to 31%, 7 seedlings m⁻²). Up to 20% of total sown species established, equating to 222 seedlings m⁻².

Soil temperature

There was a trend towards greater soil temperatures at 10 cm depth in spring than in autumn when averaged over the time of sowing until the time of establishment assessments for each site × season × aspect combination (Figure 3). There were no significant correlations between establishment of perennial ryegrass, white clover, total grasses, total legumes, total grasses or total sown species, and mean temperature of the soil at 10 cm depth (results not presented).

Figure 3. Mean soil temperature at 10 cm depth at 0900 h for each of four sites: Ngaroma, Poukawa, Woodville and Cheviot. The means are calculated from all data from the time of sowing until the time of counting for each site for each aspect × season combination.

Discussion

Seed mix

The consistent trend across all sites of greater establishment and seedling densities of sown species from the GLH than LEG mix, with few interactions, reflected the greater establishment of sown grasses and sown herbs than of sown legumes, and the amount of seed sown, which was approximately 1.6-fold greater in the GLH than LEG mix (Table 3). However, other factors contributed to successful establishment of individual species; red clover and subterranean clover establishment was 1.8-fold greater from the GLH than LEG mix, despite being sown at lower rates in the GLH mix. Reasons for this greater establishment in the GLH mix are speculative; possibly there was less intraspecific competition with the lower legume sowing rates in the GLH mix, resulting in less legume density-dependent mortality and greater seedling establishment percentages. Regardless of percentage establishment, seedling densities were similar from both seed mixes, averaging 20 red clover seedlings m⁻² and 5 subterranean clover seedlings m⁻².

In contrast with trends for red clover and subterranean clover, percentage establishment of white clover was more than two-fold greater from the LEG than GLH mix, despite being sown in the LEG mix at half of the rate in the GLH mix. Greater establishment of white clover in the LEG than GLH mix may be explained by less shading in the LEG mix, because of lower seedling densities and the absence of taller seedlings of rapidly germinating sown grasses (e.g. perennial ryegrass) and herbs (e.g. plantain) present in the GLH mix. During seedling assessments, white clover seedlings were shorter than those of red and subterranean clover, which would make them more vulnerable to shading. These observations are consistent with those of Grime et al. (2007) who noted in floristic surveys of the British Isles that adult red clover foliage was typically below a height of 300 mm, whereas foliage of white clover was below 100 mm, and that white clover was intolerant of shade. While species such as white clover may benefit from being sown in a legume only mix, the low total legume and total sown species establishment in the LEG mix suggest that pasture establishment may be more successful from sowing a mix of grasses, legumes and herbs. Another option to aid legume establishment would be to sow legumes first, followed by grasses 6 or 12 months later (e.g. White et al. 1972).

Overall, establishment percentages were low: only up to a third of germinable seed of perennial ryegrass and plantain produced seedlings, and a quarter of red and white clover seed. These four species dominated the swards at most sites; percentage establishment of cocksfoot, grazing brome, subterranean clover and lotus species was low at all sites. Total sown species establishment also remained low, ranging from maximums of 16% at Ngaroma to 39% at Poukawa depending on the treatment combination. In some treatments at the summer-dry sites, Poukawa and Cheviot, sown seedling establishment was as low as 4% or 5%. These percentages are consistent with other studies. For example, at three hill country sites (Manawatu, southern Hawke’s Bay and northern Wairarapa), establishment of perennial ryegrass ranged from 11% to 17% at 8 weeks after sowing. Losses were attributed to molluscs and other pests and it was suggested that methods that increased seed–soil contact (trampling) and pesticides may reduce seedling losses (Falloon & Charlton 1984). In another study, 65%–76% of seed sown in hill country failed to produce seedlings, with losses being attributed to environmental stresses and pest damage (Chapman et al. 1985). Barker and Zhang (1988) have also attributed perennial ryegrass seedling losses to pest damage and environmental stresses. Clovers can be difficult to establish because of their epigeal germination, in which ‘the forces of the extending hypocotyl and radical combine to push the germinating seed away from the point of entry of the radical, leaving the seedling exposed to desiccation and attacks by pests and pathogens’ (Charlton 1977). In this study, mob stocking of plots after broadcasting would have facilitated seed–soil contact. Other factors such as competition from seedlings that emerged from the resident seedbank and microclimatic extremes were more likely to have contributed to the low seedling densities.

The success of the seed mix sown also depended on other factors, such as aspect (Allan & Chapman 1987). For example, establishment of white clover at Woodville was greater in the LEG than GLH mix (39% vs 19%) on the south aspect but similar on the north (averaging 13%). Reasons for this may include differences in the microclimate, with greater moisture retention on the south aspect resulting in a greater establishment percentage (39%). Aspect will also influence the soil temperature regimes. Figure 3 demonstrates that there was a trend towards cooler soil temperatures at 10 cm depth on the south than north aspects in spring and in autumn during the seedling establishment phase at Poukawa, Woodville and Cheviot. A number of the interactions observed in this study are likely because of differences between the aspects in soil temperature.

In any species mix, some species will be more compatible than others. For example, phalaris and cocksfoot do not compete well when young (Peeters 2004) and may be outcompeted by more vigorous species such as perennial ryegrass. In these harsh hill country environments, the species with the greatest percentage establishments were most likely the species that compete well during the seedling stage (e.g. perennial ryegrass, red clover and plantain). Ongoing assessments will determine if these species remain dominant or if a new equilibrium is established in which plants that can tolerate stress become more abundant (for example, cocksfoot rather than perennial ryegrass in summer-dry environments, due to its deeper roots and greater drought tolerance [Evans 1978]).

Aspect

Most sown species had greater establishment on the south than north aspects, although significant differences were noted only for perennial ryegrass and birdsfoot trefoil, which were 1.7- and 1.2-fold greater on the south than north aspect, respectively, when averaged over all sites. The south aspect usually has greater soil moisture than the north aspect (Lambert & Roberts 1976) and is the most likely reason for greater establishment on the south aspect.

There was an exception at Cheviot, with greater establishment of grazing brome (40% vs 1%) and red clover (24% vs 16%) on the north than south aspect. The reason for the finding for red clover is unknown because red clover may be best adapted to moist, cool situations on shady aspects at altitudes below 1000 m (Rhodes & Clare 1983) but there is a likely explanation for the high establishment of grazing brome. Many brome species tolerate hot, dry climates (Sithamparanathan 1979) and Grime et al. (2007) found greater establishment of goose grass (Bromus hordeaceous) on the south (sunnier) than north (shadier) aspects in the United Kingdom. These findings are consistent with our seedling establishment data and also the observation that goose grass was present on the north face in the surrounding pastures at Cheviot. Since many Bromus species look similar in the juvenile stage, it is also possible that unsown brome species, such as goose grass, were included in the counts and contributed to the greater establishment of grazing brome recorded on the north aspect, particularly in autumn.

Season

Establishment was greater from autumn than spring sowing for most species at Poukawa and Cheviot, with as great as 25% establishment of total sown species from autumn sowing and less than 5% at Cheviot from a spring sowing. Lack of soil moisture over summer was the most likely reason for the greater establishment from autumn than spring sowing at these sites. In summer-dry regions, rainfall is often inadequate for seedling growth and survival over summer, especially as seedlings are more vulnerable to desiccation than adult plants which have better established and deeper root systems. Species sown in autumn would be unlikely to be drought stressed from late autumn until the following summer, by which time they would be larger, with more developed root systems and better able to tolerate drought.

However, establishment of most legume species was greater from spring sowing at Ngaroma and Woodville with up to 30% of white clover establishing at Woodville from spring sowing on the north aspect. Greater establishment from spring than autumn sowing at Ngaroma and Woodville is consistent with Sithamparanathan et al. (1986) who found that establishment of perennial ryegrass, prairie grass (B. catharticus Vahl) and phalaris was greater from spring than autumn oversowings in summer-moist hill country at Whatawhata, North Island.

There are several reasons for our results. First, summer rainfall was more than adequate for growth and survival. Second, increasing temperatures as spring progressed may have enhanced seedling growth. Third, competition from unsown species emerging from the seedbank may be less in spring than autumn because most unsown species, particularly annual grasses, produce seed in spring and germinate in the subsequent autumn rather than remaining dormant until spring (Cousens & Mortimer 1995).

The year in which the seeds were sown had greater rainfall than normal at all sites. Had the rainfall been closer to the long-term average, drying of the soil profile may have been more severe at some of the sites, with greater differences in establishment between spring and autumn sowing or north and south aspects for example. A subset of treatments is being repeated to increase our confidence in the knowledge of the impact of sowing season on seedling establishment.

There were a number of interactions between season of sowing and aspect. While the magnitude and direction of the interactions often varied between sites, on the wetter sites there was a trend towards greater legume establishment from spring than autumn sowing on one aspect, but similar establishment from spring and autumn sowing on the other. For example, most of the legume species established more successfully from spring than autumn sowing on the north aspect at Woodville. Consequently, it is inferred that broadcasting grasses, legumes and herbs on summer-safe country in spring is more likely to result in greater total sown species establishment, but where there is increased risk of summer drought, autumn sowing is recommended.

Fallow and cropping

There were significant benefits of summer fallow for most species at each of the sites and also consistent effects across all sites. A summer fallow was particularly beneficial for seedling establishment on a south aspect but had less of an impact on the north aspect. Benefits most likely occurred due to a reduction in competition with resident species. Application of herbicide in the spring would suppress or kill existing vegetation, reduce the amount of seed entering the seedbank and subsequent emergence of unsown seedlings in autumn. A follow-up application of herbicide in autumn at the end of the fallow period would reduce competition further by eliminating remaining resident vegetation. The reduction in resident vegetation during the fallow period may reduce water use, leaving more moisture available for the establishment of the sown pasture species.

Fallow resulted in a two- to six-fold increase in establishment of sown grasses, sown legumes, sown herbs and total sown species at Ngaroma and Woodville. At these sites, the herbicide killed the dense mat of Yorkshire fog (Holcus lanatus L.), sweet vernal (Anthoxanthum odoratum L.) and browntop (Agrostis capillaris L.)-dominant vegetation prior to broadcasting seed. Further, visual observations showed that the dense thatch of dead material, present after the initial application of herbicide in spring, disintegrated during the fallow period. This would enhance seed–soil contact and establishment of sown species in comparison with the standard autumn treatment, which received application of herbicide only in the autumn.

At Cheviot, there were significant increases in establishment for most species in response to fallow, except for grazing brome establishment that was greater without fallow. This may be due to resident seedbank dynamics. Without spring application of herbicide, unsown brome species would set seed and replenish the seedbank, and germinate in autumn. Since brome species are difficult to distinguish when seedlings, it is possible that both sown and unsown brome species were included in the seedling counts, thus overestimating the establishment percentages of sown brome. A similar situation likely occurred at Poukawa with perennial ryegrass, because its establishment was greater in the treatments without fallow. Difficulties with eliminating resident vegetation have also been documented in other studies. For example, application of diclofop and amitrole herbicides did not eliminate resident perennial ryegrass (Charlton & Henderson 1985) and glyphosate-treated pastures were prone to flatweeds, Californian thistle and browntop reinvasion within 2 months of sowing, leading to suppression of the newly sown white clover (Charlton & Henderson 1985). However, the current results highlight an opportunity to manipulate the resident seedbank to improve botanical composition. If the seedbank contains desirable species, such as adventive clovers, management strategies to encourage replenishment of the seedbank and the emergence of desirable species in the subsequent season may be worthwhile (Maxwell et al. 2014).

There were interactions between fallow and aspect at each of the sites. One consistent trend across most of the sites was that establishment was greater with than without a fallow on the south aspect but similar with than without a fallow on the north aspect. The most likely reason for this trend across sites is reducing competition and the thatch of dead material that is often more prevalent on the shadier south aspects, where species such as browntop and Yorkshire fog are dominant as discussed above. The grass species for which benefits were observed—cocksfoot and phalaris—do not compete well as seedlings (Grime et al. 2007) and legumes are especially vulnerable to shading and competition when young because of their epigeal germination as discussed previously (Charlton 1977).

In contrast with other sites, at Poukawa, establishment often declined in response to a fallow. During assessments, many of the fallowed plots at Poukawa were particularly bare and denuded of dead vegetation (litter). While the reasons for this are unknown, it is postulated that the lower establishment after fallow is due to this excessive litter removal. These observations are consistent with the previously mentioned study of Chapman and Fletcher (1985) who observed a negative impact of herbicide on white clover establishment, most likely because of excessive vegetation removal and exposure of white clover seedlings to desiccation, and also with Campbell and Swain (1973), who suggested that ‘protective cover should be provided for all surface-sowings where moisture stress is likely to occur’.

Similar trends to fallow were observed in response to cropping with a summer brassica (forage rape). For example, when compared with standard autumn sowing, fallow or cropping increased the establishment of sown grasses at Ngaroma, and sown herbs at Cheviot. In a similar manner to a fallow, the cropping phase involved a spring and autumn herbicide application that would have increased sown seed–soil contact by breaking down the resident thatch. A cropping phase would also reduce competition between resident vegetation/emerging unsown seedlings and the sown species. Given these results, either fallow or cropping are likely to benefit establishment of pasture species regardless of the location. These practices will particularly benefit species that are slow to establish and vulnerable to competition, most likely through increasing seed–soil contact and reducing competition with resident vegetation.

This study has provided evidence for our hypothesis, that while trends in the establishment of sown grasses, legumes and herbs are specific to individual sites, some trends in establishment remain common across all sites. For example, there were consistent trends for greater establishment from the GLH than LEG mix across all sites (although legumes such as white clover established more successfully from being sown in the LEG mix), greater establishment of perennial ryegrass from sowing on the south than north aspect, and benefits from fallow and cropping for most species at all sites except Poukawa. However, the site-specific impact of fallow at Poukawa most likely demonstrated the importance of the quantity of litter rather than the process of fallow per se. The effect of season was more site-specific, with greater establishment from autumn than spring sowing at the summer-dry sites but spring than autumn sowing at the two wetter sites.

Acknowledgements

Advice on farmer needs was provided by Beef + Lamb New Zealand. Thanks to the managers and staff on whose properties the trials were established at Ngaroma, Poukawa, Woodville and Cheviot; Pastoral 21.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

Funding was received from the Ministry of Business, Innovation and Employment, DairyNZ, Fonterra, Beef + Lamb New Zealand and DCANZ (Dairy Companies Association of New Zealand).