Occurrence of meadow herbs in sown and unsown ploughed strips in cultivated grassland

Abstract

Creation of permanent, semi-natural strips may contribute to increased plant diversity in agricultural areas if they function as habitats for new species. Non-fertilized strips that were ploughed and either unsown or sown with a grass and meadow herb seed mixture were created in a grass crop field, in order to test the establishment of several meadow herb species, to investigate to what extent creation of sown and unsown strips increased plant species numbers and the number of meadow herbs, and to assess the contribution of species dispersal from an adjacent semi-natural field margin into the created strips. The strips were cut each autumn and plant material removed. Plant species composition was studied in permanent quadrats in the strips and in the surrounding grass crop during four years. Four of the 22 sown meadow herb species did not establish in the strips. Sowing increased both the number of meadow herbs and plant species diversity (calculated by the Shannon index) compared with the grass crop. Creating unsown strips did not increase the number of meadow herbs or plant species diversity. The created strips were colonized by woody species originating from the adjacent semi-natural field margin, but the meadow herbs present in the field margin did not colonize the strips. In the short term, establishment of sown strips would improve biodiversity locally. However, such strips will not serve as suitable habitats for the most vulnerable meadow herbs, and thus creation of new strips cannot compensate for loss of existing species-rich, semi-natural habitats.

Keywords:

• Central Norway
• field margins
• permanent strips
• plant colonization
• plant diversity
• wild flowers

Introduction

Agricultural development during recent decades, such as the change from traditionally managed semi-natural meadows to intensively cultivated, species-poor grass crops (Myklestad & Sætersdal, 2004), has resulted in a reduction in biodiversity (Meeus, 1993; Stoate et al., 2001). In Norway, several meadow herbs that have been prevalent in semi-natural habitats are currently declining in abundance (Losvik, 1999; Bele & Norderhaug, 2004). The preservation of meadow herbs in cultivated areas may be aided by creating ‘safe sites’ for seedling establishment, such as non-fertilized, permanent strips. If the strips are vegetated naturally they may become dominated by early-successional (arable weed) species (Hovd, in press; van der Putten et al., 2000), since colonization by later successional meadow herbs is often restricted due to lack of a local species pool such as a seed bank or nearby vegetation from which plant species can disperse (Kleijn, 1997; Bakker & Berendse, 1999). Sowing of meadow herbs may therefore be necessary (Hutchings & Booth, 1996a; 1996b; Losvik & Austad, 2002; Pywell et al., 2002). The seeds should be collected locally to avoid reduced fitness of the plant populations present (Keller et al., 2000). The strips should be cut regularly, since this prevents establishment of trees, and suppresses tall-growing, dominant species while promoting low-growing and less competitive ones (Persson, 1995; Hovd & Skogen, 2005).

Meadow herbs adapted to semi-natural grasslands may not establish well in managed grass fields on fertile soils. The aims of this study were therefore to test the establishment of several meadow herb species when sown in non-fertilized, permanent strips created in a fertile grass crop field, and to investigate to what extent creation of sown and unsown strips increased plant species numbers and the number of meadow herbs. The plant species composition of a semi-natural field margin bordering the grass crop field was also studied, in order to assess the contribution of species dispersal from the field margin to the strips. The sown meadow herb species were part of a commercially available grass and meadow herb seed mixture. The seed mixture comprised species that are still relatively common in the study region (Hovd & Skogen, 2005), but, with a few exceptions, are decreasing in abundance in central Norway (Fremstad & Moen, 2001) and thus the creation of new habitats for these species may contribute to preserving local plant diversity.

Materials and methods

This study was carried out in a grass crop field at Bioforsk Kvithamar, central Norway (63°30′N, 10°52′E). Mean annual temperature was 5.8°C (−1.3°C in January and 14.8°C in July), and mean annual precipitation was 929.5 mm (values obtained from an automatic weather station at Kvithamar, 1993–2004). The soil was very fertile. It was of marine origin with high contents of plant available K and P, and with a silt content >80%, a clay content of 5–12%, an organic matter content of 4.5–12.5% and pH 6.5 in the uppermost 20 cm. The field had been sown to grass ley and had not been ploughed for at least 10 years, and was dominated by Phleum pratense and Festuca pratensis. During the experiment, the grass crop was fertilized with 175 kg N ha⁻¹, 22.5 kg P ha⁻¹ and 60 kg K ha⁻¹ and harvested twice a year, which is typical management in this region.

Along one edge, the grass crop field bordered a spinney, and the field and the spinney were separated by an approximately 3-m wide, unmanaged semi-natural field margin. The presence of all plant species in the total area of the field margin (approximately 600 m²) was recorded in the first year. Since agricultural fields in Norway are often small and restricted by the topography, the widths of the permanent strips cannot be too large. Strip widths of 2 m were used in this experiment. Four strips were ploughed in the grass crop in May 2000 (Figure 1). The strips were laid out perpendicular to the semi-natural field margin, in order to be able to assess the contribution of species dispersal from the field margin into the created strips, at different distances from the field margin. After ploughing, a flat soil bed was made by hand raking. One half of each strip was not sown, but left to regenerate naturally. The other half was sown in May 2000 with a slightly modified version of a commercially produced grass and meadow herb seed mixture used for establishment of flowery meadows. The seed mixture (amounts sown are given as the proportion of the total seed mixture weight) consisted of 22 meadow herbs: Achillea millefolium (0.6), Achillea ptarmica (1.8), Anthemis tinctoria (1.6), Anthyllis vulneraria, Campanula rotundifolia, Centaurea jacea (2.0), Galium verum (1.0), Geranium sylvaticum, Geum rivale, Knautia arvensis (0.6), Leucanthemum vulgare (2.8), Lotus corniculatus, Primula veris, Prunella vulgaris (0.4), Ranunculus acris ssp. acris (0.4), Silene dioica (3.4), Silene vulgaris (3.0), Solidago virgaurea, Trifolium pratense (1.2), Vicia cracca, Vicia sepium and Viola tricolor and three grasses: Agrostis capillaris (2.0), Festuca rubra ssp. rubra (63.0) and Festuca trachyphylla (15.0). The species lacking amount values were present in the seed mixture, but at unknown proportions. All meadow herb seeds in the mixture had been collected near the experimental site, but the grasses in the seed mixture were cultivated varieties provided by a national seed producer. The majority of the species were common in semi-natural field margins and/or in road verges in the region (Hovd & Skogen, 2005). The modifications of the seed mixture comprised an increased proportion of Agrostis capillaris, a reduced proportion of Festuca trachyphylla since this species is not native to this area (Lid & Lid, 1994), and an increase in meadow herbs from 5% to 20% of total seed weight. The seed mixture was sown by hand at 5 gm⁻² as recommended by the producer, and pressed into the soil using a garden roller. In contrast to the grass crop, the strips were not fertilized or harvested during the growing season and were cut only once, in mid to late September each year, to ensure sufficient time for all meadow herb species to flower and set seed. The cut biomass was left in the field for two days before removal to ensure that plant seeds were dropped in the field and not removed (Schaffers et al., 1998).

Figure 1.  The location of the ploughed sown and unsown strips and the field margin bordering the grass crop field. The field margin was on average 3 m wide.

Permanent quadrats (0.5 m×0.5 m) were analysed for plant content in June each year from 2000 to 2003. Quadrats were sampled both in the grass crop 1 m from the strip edges, and in the strip centres at regular intervals along the strip lengths. The distance between sample quadrats was 3.5 m. In total, 80 quadrats were sampled each year (sown strips: n = 24, unsown strips: n = 24, grass crop: n = 32). In each quadrat, percent biomass of all vascular plant species was visually estimated. This method is useful to estimate herbage composition or herbage mass (Frame, 1993; Martin et al., 2005). The Shannon index (H) of diversity was calculated for each quadrat using the formula H = − Σ p_i ln p_i (Kent & Coker, 1992) where p_i is the estimated proportion biomass of a species in the quadrat. Shannon index values were compared between quadrats in sown strips, unsown strips and the grass crop by Generalized Linear Models (GLM) analyses with Tukey's post hoc tests, using mean values for each strip and analysing each year separately (n = 4). The total number of plant species and the number of meadow herbs per quadrat were compared between sown strips, unsown strips and the grass crop by GLM with repeated measures (data from all four years) using mean values for each strip (n = 12). To test whether the distance (log-transformed) from the semi-natural field margin affected species numbers in the field, a GLM with repeated measures was carried out, using mean values of total number of species per quadrat for each strip (n = 12). The GLM analyses were performed by SYSTAT11 (SPSS Inc., 2004). Botanical nomenclature follows Lid and Lid (1994).

Results

Most of the 22 sown meadow herb species germinated and established in the strips, but four of them (Centaurea jacea, Anthyllis vulneraria, Primula veris and Solidago virgaurea) were not recorded in the sown strips during the experiment, and Knautia arvensis was recorded only in the third year. The proportion of Achillea millefolium, A. ptarmica, Festuca spp. and Leucanthemum vulgare increased over time, while the proportion of Anthemis tinctoria and Trifolium pratense increased the first two years and then decreased. In the fourth year, some of the sown species were recorded in unsown strips or in the grass crop too: Anthemis tinctoria, Festuca spp., Galium verum, Leucanthemum vulgare, Prunella vulgaris, Trifolium pratense, and Vicia cracca were recorded in unsown strips, while Leucanthemum vulgare and Silene vulgaris appeared in the grass crop (Table I).

Table I. Plant species recorded in quadrats sampled in the field (in ploughed and sown or unsown 2-m wide strips and in the grass crop) in the first (2000) and the fourth year (2003) after strip ploughing. The abundance of each species is given as% of total biomass per quadrat grouped as follows: 0: no observation, 1; <1%, 2: 1–10%, 3: >10%. Nomenclature follows Lid and Lid (1994).

	Year 1			Year 4
	Sown	Unsown	Grass crop	Sown	Unsown	Grass crop
n	24	24	32	24	24	32
Total species number	43	33	32	40	37	28
No. of meadow herbs	16	9	9	23	14	11
Meadow herbs
Achillea millefolium*	1	0	0	2	0	0
Achillea ptarmica*	1	0	0	2	0	0
Anemone nemorosa	0	0	0	1	0	1
Anthemis tinctoria*	2	0	0	1	1	1
Campanula rotundifolia*	0	0	0	1	0	0
Cerastium fontanum ssp. vulgare	1	1	0	1	1	0
Epilobium montanum	1	1	1	1	1	1
Galium verum*	1	0	0	1	1	0
Geranium sylvaticum*	0	0	0	1	0	0
Geum rivale*	0	0	1	1	0	0
Leucanthemum vulgare*	2	1	0	3	1	1
Lotus corniculatus*	1	0	0	1	0	0
Prunella vulgaris*	0	0	0	1	1	0
Ranunculus acris ssp. acris*	1	1	1	1	1	1
Sagina procumbens	1	1	1	0	1	0
Silene dioica*	0	0	0	1	0	0
Silene vulgaris*	2	0	0	2	0	1
Stellaria graminea	0	0	0	1	1	1
Trifolium hybridum	1	2	2	0	1	1
Trifolium pratense*	2	0	2	2	1	1
Trifolium repens	2	2	2	1	1	1
Veronica serpyllifolia ssp. serpyllifolia	0	1	1	1	1	1
Vicia cracca*	2	0	1	1	1	0
Vicia sepium*	0	0	0	1	0	0
Viola tricolor*	1	1	0	1	0	0
Grasses
Agrostis capillaris*	1	2	0	2	2	0
Alopecurus geniculatus	0	0	1	0	0	1
Dactylis glomerata	2	2	2	2	2	3
Elymus repens	2	2	1	2	2	2
Festuca pratensis	3	3	3	3	3	2
Festuca spp.*	1	0	0	3	1	0
Phleum pratense ssp. pratense	3	3	3	3	3	3
Poa annua	1	2	1	0	1	0
Poa pratensis ssp. pratensis	0	0	1	1	1	1
Poa trivialis	1	0	2	1	1	1
Annuals
Capsella bursa-pastoris	1	1	1	0	0	0
Chenopodium album	1	1	0	0	0	0
Fallopia convolvulus	0	1	0	0	0	0
Fumaria officinalis	1	2	1	0	0	0
Galeopsis bifida	1	0	1	1	0	1
Lamium purpureum	2	2	0	0	0	0
Matricaria perforata	0	0	0	0	0	1
Persicaria maculosa	1	0	0	0	0	0
Polygonum aviculare	1	1	1	0	0	0
Senecio vulgaris	1	2	0	0	0	0
Sonchus asper	1	0	1	0	0	0
Sonchus oleraceus	0	1	0	0	0	0
Spergula arvensis	1	2	0	0	0	0
Stellaria media	1	1	0	0	1	1
Perennial weeds/generalists
Anthriscus sylvestris	0	0	1	0	1	0
Chamomilla suaveolens	2	2	0	0	1	1
Cirsium arvense	1	2	1	1	2	1
Cirsium helenoides	0	0	0	1	1	0
Myosotis arvensis	1	1	1	0	0	0
Plantago major	1	1	1	0	1	0
Ranunculus repens	1	1	1	1	1	1
Rumex acetosa	0	0	1	1	1	1
Rumex sp.	1	1	1	1	0	1
Taraxacum officinale (coll.)	3	3	2	3	3	3
Tussilago farfara	1	1	0	0	1	0
Woody species
Alnus incana	0	0	0	1	1	0
Betula pubescens	0	0	1	0	0	0
Picea abies	0	0	1	0	0	0
Prunus padus ssp. padus	0	0	0	0	1	0
Salix caprea	0	0	0	0	1	0
Sorbus aucuparia ssp. aucuparia	0	0	0	1	1	1
*sown species.

In the first year, both sown and unsown strips contained several annual species, but after four years most of the annuals had disappeared and the number of perennial species had increased (Table I). In the fourth year, 23 meadow herbs were recorded in sown strips, of which 17 were sown species (Table I). The plant species composition of sown strips varied markedly from the unsown strips and the grass crop. In unsown strips, grasses and perennial weeds were the most dominant species (Table I). Fourteen meadow herb species were recorded in the unsown strips in the fourth year, but most of them were also present in the grass crop (e.g. Epilobium montanum, Stellaria graminea, Trifolium repens). Generally, after four years the unsown strips and the grass crop had many species in common (Table I). In all years, the total number of plant species and the number of meadow herbs per quadrat was significantly higher in sown strips than in unsown strips and in the grass crop (Table II, GLM: p<0.01). The Shannon index of diversity was highest in sown strips, lower in unsown strips and lowest in the grass crop (Table II, GLM: p<0.05).

Table II. Total number of plant species, number of meadow herbs (mean±SE) and mean Shannon species diversity index values of 0.25-m² quadrats sampled in ploughed and sown or unsown 2-m wide strips and in the surrounding grass crop during four years. Different letters indicate significant differences (plant numbers p<0.01; Shannon index p<0.05). The strips were ploughed in spring in year 1.

	Total number of plant species			Number of meadow herbs			Shannon index
	Sown	Unsown	Grass crop	Sown	Unsown	Grass crop	Sown	Unsown	Grass crop
Year 1	18.1±0.5a	11.9±0.5b	7.5±0.4b	7.1±0.3a	1.0±0.2b	1.2±0.2b	1.84a	1.61a	1.02b
Year 2	17.3±0.4a	9.8±0.4b	7.4±0.3b	8.2±0.3a	1.7±0.3b	0.8±0.2b	1.82a	1.42b	1.17b
Year 3	17.8±0.4a	11.0±0.4b	7.8±0.4b	9.2±0.4a	2.5±0.3b	1.3±0.1b	1.91a	1.39b	1.18b
Year 4	17.6±0.4a	11.0±0.4b	9.0±0.4b	9.7±0.4a	2.5±0.3b	2.2±0.2b	1.96a	1.31b	1.31b

A total of 58 plant species was recorded in the 600 m² semi-natural field margin that bordered the grass crop field. Most of them were woody species, tall-growing grasses and perennial weeds, but 16 meadow herbs were also present. In the fourth year, four woody species present in the field margin (Sorbus aucuparia, Alnus incana, Prunus padus, Salix caprea) were also recorded in the strips and/or in the grass crop as small seedlings established from seeds. The distance from the field margin did not affect total species numbers in the studied quadrats significantly, but there was a tendency that the woody seedlings were more common in quadrats close to the field margin than in more distant quadrats. The meadow herbs Epilobium montanum, Ranunculus acris, Trifolium hybridum, T. pratense, and Vicia cracca were present both in the field margin and in the grass crop at the start of the experiment, and all of them were recorded both in sown and unsown strips during the experiment (Table I). The field margin contained 11 meadow herb species that were not recorded in the grass crop at the start of the experiment (Achillea millefolium, Achillea ptarmica, Campanula rotundifolia, Fragaria vesca, Hypericum maculatum, Lathyrus pratensis, Oxalis acetosella, Potentilla erecta, Sedum acre, Vicia sepium), and none of these colonized the created strips during the four years. A. millefolium, A. ptarmica, C. rotundifolia and V. sepium were recorded in the sown strips, but it is most likely that those individuals established from the sown seeds rather than originating from the field margin. The grasses Dactylis glomerata and Elymus repens (and the cultivated Festuca pratensis and Phleum pratense), and the generalists Ranunculus repens and Taraxacum officinale were frequently encountered both in sown strips, unsown strips and in the grass crop throughout the experiment. The perennial weed Cirsium arvense was also frequently recorded in all three groups but was most dominant in unsown strips (Table I).

Discussion

Creation of unsown (i.e., naturally regenerated) strips did not increase plant species numbers or the number of meadow herbs compared with the grass crop. As also found in other experiments (Hovd, in press; Denys & Tscharntke, 2002), the unsown strips were dominated by grasses and weeds. This could be due to the fact that although the strips were not fertilized, they were created on fertile soil with relatively high soil nutrient levels, which is known to promote the growth of weeds and competitive grasses (Bakker & Berendse, 1999). Sowing a grass and meadow herb seed mixture significantly increased both total plant species numbers and the number of meadow herbs. Within sown communities, successful species tend to be habitat generalists and species characteristic of fertile grasslands, while stress-tolerators, habitat specialists and species of infertile habitats often perform poorly (Pywell et al., 2002; Bokenstrand et al., 2004). In the present experiment, the majority of the sown meadow herbs had relatively broad habitat tolerances and were adapted to intermediate soil nutrient levels, and could therefore be expected to germinate on the fertile agricultural land where this experiment was set up. Succession in sown sites, with increasing species richness and abundance of perennials, is found to continue after five years (Critchley & Fowbert, 2000). Thus, four years is a short time to establish grassland vegetation, and therefore all meadow herbs could perhaps not be expected to establish during the course of this experiment. Of the 22 sown meadow herbs, Anthyllis vulneraria, Centaurea jacea, Primula veris and Solidago virgaurea did not establish in the strips. It is not known why these species were absent in the sown strips, but one possibility is low germination capacity. In addition, A. vulneraria may take several years to germinate due to its hard-shelled seeds, and P. veris is dependent on a cold period (stratification) prior to germination. Since the seeds were sown in spring, germination and seedling establishment of P. veris the following year could have been restrained by competition from the established vegetation. The absence of C. jacea in the strips was unexpected, since this species often tends to become dominant when sown as part of this species mixture (personal communication from seed producer). Successful establishment of C. jacea species from seeds has also been reported from studies in Germany and Sweden (Hofmann & Isselstein, 2002; Bokenstrand et al., 2004). Campanula rotundifolia did not establish in the sown strips until the fourth year, perhaps because it is a late successional species and does not thrive under the arable conditions created by the strip ploughing in the first year (Grime et al., 1988). Bare soil may be quite inhospitable to seedlings of species adapted to regenerate in grasslands, and cover by established plants may protect seedlings from abiotic extremes. Results from a field experiment in the Netherlands indicated that establishment of C. rotundifolia was more successful in plots with one-year-old or two-year-old vegetation than in bare plots (Kleijn, 1997).

A rapid formation of ground cover by the sown species may have reduced dominance of Cirsium arvense, since the proportion of this species was higher in the unsown than in the sown strips. Sown grass and/or meadow herbs may reduce the abundance of several weeds, e.g., C. arvense and Elymus repens, compared with naturally regenerated sites (Kleijn, 1997; Denys & Tscharntke, 2002; Bokenstrand et al., 2004; Asteraki et al., 2004, Critchley et al., 2006). Plant species mixtures of relatively high diversity (15 species) are in general more successful in weed suppression than low-diversity mixtures, but the effect of low-diversity mixtures depends on the species composition (van der Putten et al., 2000).

The distance from the quadrats to the semi-natural field margin that bordered the grass crop field did not affect the number of meadow herbs or total species numbers in the quadrats significantly. Nevertheless, there was a tendency in both unsown strips and in the grass crop that total species numbers were higher in quadrats close to the field margin than in more distant quadrats. The higher species numbers in quadrats close to the field margin were probably caused by the dispersal of woody species and perennial weeds from the field margin. The field margin did not contribute to increase the number of meadow herbs in the strips. Five meadow herb species that were present in the field margin were also recorded in the strips, but these species were also present in the grass crop at the start of the experiment and were likely to have been spread from the grass crop. In general, the meadow herbs present in the field margin did not colonize the strips during the four years of this study, not even the closest quadrats that were situated only 3.5 m from the field margin. This supports the necessity of introducing seeds to maintain or increase the meadow herb flora in some areas, but also that although sowing of meadow herb strips may aid in preserving some meadow herb species in an area, such strips will not serve as suitable habitats for the most vulnerable or demanding herbs (Smart et al., 2002).

During the four years of this experiment, individuals of a few of the sown species appeared in the grass crop or in the unsown strips and had probably dispersed from the sown strips. Over time, it is expected that several of the sown meadow herb species will be dispersed to suitable ‘gaps’ in the agricultural landscape. Since the strips were studied for four years only, it is not known how their plant species composition may change over time. Other, longer-term experiments indicate that sown meadow herb strips may maintain most of their plant species content for at least up to 10 years (Bokenstrand et al., 2004; Critchley et al., 2006), and in the short term, establishment of sown strips would improve biodiversity locally. However, the establishment of grassland vegetation on previous intensively used agricultural land may take several decades (Bekker et al., 1997), and the long-term coexistence of the grassland plant species in the strips may be hampered by the high soil nutrient levels (Janssens et al., 1998). Although the sown strips contained several meadow herbs four years after establishment, they were still species-poor compared to some of the road verges in the study area (Hovd & Skogen, 2005), and thus creation of such strips cannot compensate for loss of species-rich semi-natural habitats.

Acknowledgements

Thanks are expressed to Anne Kjersti Bakken, Ann Norderhaug and Arnfinn Skogen for valuable comments and discussions, Anne Langerud for maintaining the field experiment and for help with data preparation, Vibecke Melhuus for assisting with field recordings, Line Rosef for critical reading of the manuscript and Wendy Fjellstad for critical reading and improving the English text. This project was supported by the Norwegian Research Council.