1. Turfs transplanted from native vegetation can be used to restore diverse plant communities on disturbed sites. There is, however, limited understanding of optimal turf size and the tolerance of different plant communities and species to transplanting.
2. The effects of turf size in restoration of alpine plant communities were studied in SW-Iceland. Treatments tested in 2-m2 plots were as follows: planting of sixteen 5 × 5 cm turfs, four 10 × 10 cm turfs, one 20 × 20 cm turf or one 30 × 30 cm turf; a 20 × 20 cm turf shredded and spread over the plot and controls without turfs. The 10-cm thick turfs were extracted from nearby heath and grassland vegetation and planted in mineral soil and road verges at 260–410 m elevation. Species composition, cover and colonization were monitored for three growing seasons.
3. Grassland vegetation tolerated division into small turfs better than heath vegetation, but responses varied by functional groups. Cover of dwarf-shrubs decreased with decreasing turf size; grass cover was highest in plots with 5 × 5 cm turfs and lowest in plots with shredded turfs, while moss cover increased most rapidly in plots with shredded turfs.
4.Synthesis and applications. Optimum turf size for the restoration of native species varied among functional groups of plants and decreased as follows: evergreen dwarf-shrubs > deciduous dwarf-shrubs > sedges > grasses > mosses. Turfs that are at least 20–30 cm in diameter may be needed for the transplantation of dwarf-shrubs, while turfs as small as 5 cm in diameter can be used to establish many grass species. Even smaller units can be used to facilitate moss colonization. Turfs that are salvaged from development projects can be a valuable source of native species for use in restoration schemes. Turf size for transplanting should be selected with regard to donor vegetation, growth form and abundance of the target species.
Many grass species seem to be tolerant to transplanting, and there are several examples of increased cover of grasses following transplantation (see, e.g., Buckner & Marr 1988; Bay & Ebersole 2006; Trueman, Mitchell & Besenyei 2007). The tolerance of forbs, shrubs, sedges and rushes to transplanting varies depending on the species (May, Webber & May 1982; Cole & Spildie 2006), which has been attributed to different root form and the degree to which roots and rhizomes are damaged by the extraction and transfer of turfs (May, Webber & May 1982). The tolerance of species to transplanting and the ability of plant species to spread from the turfs can influence the vegetation composition of receptor sites with time.
The aim of the study was to develop methods to use turf from native vegetation as a source of species for the restoration of disturbed sites. The study was designed to address the following research questions: (i) What is the minimum turf size for transplanting dwarf-shrub heath and grassland into disturbed areas for effective transfer of species? (ii) What is the effect of turf size on survival of different growth forms and plant species? (iii) What is the importance of colonization by seed and vegetative spread for the establishment of native species in the vicinity of transplanted turfs?
The effects of turf size ranging from 5 × 5 to 30 × 30 cm on transfer of species from a dwarf-shrub heath and grassland were tested in experiments at disturbed highland sites in SW-Iceland. This study presents results for species composition and plant colonization for three growing seasons following the transplanting.
Materials and methods
The study was carried out at Hellisheiði, SW-Iceland, at the construction site of the Hellisheiði geothermal power plant (latitude 64°02′N, longitude 21°23′W). The study area is south of the Hengill volcano on both sides of the Hellisskarð pass. The area above Hellisskarð is characterized by palagonite mountains, volcanic craters and extensive lava flows, about 2000 and 5800 years old; but the area below the pass is mostly on a 10 000-year-old lava field overlain by gravel outwash from nearby canyons (Sæmundsson 1995). The vegetation of the Hellisheiði area is characterized by extensive moss heaths on the lava fields, grassland vegetation, dwarf-shrub heath and small wetlands (Guðjónsson, Egilsson & Skarphéðinsson 2005).
Mean annual temperature for 2007–2009 was 2·6 °C and mean annual precipitation 2500 mm, measured at an automatic weather station at Hellisskarð (elevation 370 m) (Icelandic Meteorological Office, unpublished data). February was the coldest month (average −2·5 °C) and July was the warmest month (average 10·2 °C). All 3 years had high precipitation in autumn and winter, but dry spells in June and July with 34–42 days of <1-mm average daily precipitation.
Two study sites were established in May 2007. Each site had a donor area from which turfs of an intact sward were extracted and two turf plantation experiments (Table S1 in Supporting Information). The vegetation of the donor area above Hellisskarð (elevation 400 m) was a dwarf-shrub heath, but grassland at the donor area below Hellisskarð (elevation 260 m). One experiment at each site was on a flat area where all vegetation had been removed and the surface was covered with mineral soil (soil experiment); the other experiment was in a road verge that was at least 5 m wide, with 15 ± 5° slope (road verge experiment). The road verges were covered with poorly sorted gravel with fragments up to 5 cm in diameter, mixed with a few per cent of fine earth materials (coarse silt and sand). The grassland soil experiment had a surface layer of silt loam-textured soil that had had been disturbed by the construction activity, overlying deep (>2 m) fluvial Andosol deposits from the surrounding hill slopes. The heath soil experiment had a much thinner silt loam layer (mostly 0·1–0·5 m), mixed with gravel from nearby construction and fragments from the underlying basaltic lava.
Each experiment comprised 48 plots, 1 × 2 m, where six treatments were tested in a randomized block design: (1) sixteen 5 × 5 cm turfs planted with equal spacing; (2) four 10 × 10 cm turfs planted with equal spacing; (3) one 20 × 20 cm turf planted in the plot centre; (4) one 20 × 20 cm turf shredded and distributed evenly over the plot; (5) one 30 × 30 cm turf planted in the plot centre; and (6) control plot, no turf (Fig. S1, Supporting information). Treatments (1)–(4) had turf area of 400 cm2 corresponding to a turf/receptor area ratio of 1 : 50, but treatment (5) had turf area of 900 cm2 corresponding to a turf/receptor area ratio of 1 : 22. All the turfs had a 10-cm thick soil layer and were planted in 10-cm deep holes.
The experiments with grassland turfs were planted on 30 May to 4 June 2007, and the road verge experiment with turfs from dwarf-shrub heath was planted on 13–15 June 2007. Because of dry weather and delays in construction, the heath soil experiment was not planted until 2–4 July 2007. The turfs were planted on the same day as they were extracted. The plots were fertilized in July 2008 with 20 g m−2 of mineral fertilizer (20% N, 10% P2O5 and 10% K2O) to facilitate colonization of native plants (cf. Elmarsdottir, Aradottir & Trlica 2003).
Activities associated with the construction of Hellisheiði power plant caused some disturbances to the experiments. Grass seed (Festuca rubra and Poa pratensis) and organic fertilizers (horse manure and meat meal) were accidentally spread over the heath road verge experiment during revegetation of a nearby road verge in June 2008. Monitoring of this experiment was nevertheless continued, as it might still provide valuable information about factors affecting the success of turf transplants.
Vegetation composition was assessed in eight 0·5 × 0·5 m quadrats on each donor area before the turf extraction. Vegetation composition of the experimental plots was assessed in late August to early September 2007–2009, with the exception of the heath road verge experiment that could not be accurately assessed in 2008 because of the accidental spreading of organic fertilizer over the plots earlier that summer. Cover (species cover cf. Fehmi 2010) of individual species of vascular plants and main genera of mosses and lichens was estimated in two 1-m2 quadrats positioned over both halves of each plot, using the following cover classes: 1 = <1%; 2 = 1–5%, 3 = 6–10%; 4 = 11–15%; 5 = 16–25%; 6 = 26–50%; 7 = 51–75%; and 8 = 76–100%. Species richness was recorded as all vascular plant species in each 2-m2 plot. Furthermore, the species present in individual turfs were recorded in June and August each year. Nomenclature follows Kristinsson (2010) for vascular plants and Jóhannsson (2003) for mosses.
Plant colonization in the vicinity of the turfs was assessed on two permanent 5-cm-wide belt transects lengthwise and crosswise over each plot. The transects crossed the centre of the turfs in plots with a single turf; crossed randomly selected turfs in other plots with whole turfs; and were randomly located in control plots and plots with shredded turfs. Location and species of all vascular plant seedlings on transects were recorded. Location of the turf edges was recorded and used to calculate turf diameter. This was used to estimate turf area, assuming a rectangular shape and equal size of all turfs in a plot. Transects were surveyed for colonizers in August 2007–2009, with the following exceptions: the heath soil experiment was not surveyed in 2007 as it was recently established at that time; and transects in the heath road verge experiment could not be accurately surveyed in 2008 and 2009 because of high density of seeded grasses. Vegetative tillers growing out from the turfs were recorded on transects in 2007 and 2009.
Statistical analysis was performed separately on each experiment. Plots were experimental units. The cover scores were transformed to percentages by using the central value of each cover class and averaged over both quadrats in each plot. anova for repeated measures was used to test the effects of treatments and year on total vegetation cover, plant species richness and the sum of cover of all species within each growth form. The data were square-root or ln transformed where needed to meet assumptions of equal variances and normal distribution of residuals. Factorial anova (generalized linear models) was used to test the effect of experimental treatments on cover, species richness and total density of seedlings within each year, and Tukey’s HSD (P ≤ 0·05) was used to test for differences among 2009 treatment means. Statistical analysis was carried out with sas 9.2 for Windows (SAS Institute Inc., Cary, NC, USA).
Species composition of donor sites
Thirteen vascular plant species were recorded at the heath donor site and eight species at the grassland donor site. As the species composition of the sward donor sites was measured very early in the 2007 growing season, cover and species richness of vascular plants were probably underestimated compared to measurements later in the growing season.
Dwarf-shrubs dominated the vascular plant flora of the heath donor site (Table 1). Empetrum nigrum was by far the most abundant evergreen dwarf-shrub species, Vaccinium uliginosum and Salix herbacea were the main deciduous dwarf-shrubs, and Festuca richardsonii and Festuca vivipara were the most abundant grasses. The mosses Racomitrium lanuginosum and Racomitrium ericoides had 69% cover, and lichens, mostly Cladonia arbuscula (Wallr.) Flot., had over 10% cover.
Table 1. Mean (±standard error) vegetation cover of the donor sites for turf transplantation experiments at Hellisheiði (N = 8), measured before the turf extraction in late May to early June 2007
6 ± 1·3
52 ± 4·8
Sedges and rushes
2 ± 0·5
8 ± 2·1
1 ± 0·5
27 ± 3·7
10 ± 2·8
Non-flowering vascular plants
69 ± 6·2
89 ± 1·0
11 ± 2·7
3 ± 1·6
Festuca richardsonii and F. vivipara had about 50% cover at the grassland donor site, Carex bigelowii had 7% cover, and forb cover was negligible. Mosses had nearly 90% cover, Hylocomium splendens was the most abundant moss species, but Rhytidiadelphus squarrosus, R. ericoides and R. lanuginosum were also common.
Plant species richness of experimental plots
Average plant species richness in control plots and plots with shredded turfs was lower than in plots with whole turfs after the first growing season (Fig. 1). By the end of the third growing season, plant species richness in control plots was still significantly lower than in most treatments with whole turfs, while treatments with shredded turfs were usually not significantly different from treatments with whole turfs and comparable turf/receptor area ratio.
Vascular plant species richness in the soil experiment with heath turfs was always greatest in treatments with the 30 × 30 and 20 × 20 cm turfs (Fig. 1a). In the grassland soil experiment, the 20 × 20 cm plots had significantly higher plant species richness in 2009 than the 5 × 5 cm plots, but there were no significant differences in species richness among other treatments with whole turfs (Fig. 1c). Furthermore, there were no significant differences in species richness of treatments with whole turfs in either of the road verge experiments after three growing seasons (Fig. 1b,d).
Vegetation cover and composition of experimental plots
Total vegetation cover of the experimental plots after the first growing season reflected the area of the planted turfs, which corresponded to 2% cover in plots with 5 × 5, 10 × 10 and 20 × 20 cm turfs and 4·5% cover in plots with 30 × 30 cm turfs (Fig. S2, Supporting information). There was a significant increase in vegetation cover from 2007 to 2009, especially in the grassland soil experiment and the heath road verge experiment that was accidentally seeded and fertilized, but the interaction between treatment and time was only significant in the heath soil experiment (Table S2, Supporting information).
The functional groups responded differently to the experimental treatments. Grasses had <1% cover in plots with heath turfs to begin with, but their cover increased with time, especially in the heath road verge experiment where the seeded Festuca and Poa dominated all treatments in the autumn of 2009 (Fig. 2). The increase in grass cover in experiments with grassland turfs was greatest in plots with 5 × 5 cm turfs, although this treatment did not have a significantly higher grass cover by in 2009 than most other treatments with whole grassland turfs (Fig. 2c,d). F. richardsonii was the most common grass species, but F. vivipara, Agrostic capillaris and Agrostis vinealis were also abundant. Plots with shredded turfs did not have significantly greater grass cover than the controls in any of the experiments in 2009 (Fig. 2).
Sedges and rushes had <1% cover in plots with heath turfs throughout the study (Fig. S3, Supporting information). Carex bigelowii, the most common sedge, was only found in treatments with 20 × 20 cm or larger turfs in the heath soil experiment but in all treatments with whole turfs in the heath road verge experiment. In plots with grassland turfs, the cover of C. bigelowii increased with time in the treatments with 20 × 20 cm and 30 × 30 cm turfs, but remained negligible over time in treatments with shredded turfs where, conversely, cover of the rush Luzula multiflora increased with time.
Forbs had <1% cover in all the experiments throughout the study, with the exception of the grassland soil experiment where Cerastium fontanum had 5% cover in plots with shredded turfs and 1–3·5% cover in other treatments in 2009 (Fig. S4, Supporting information). By 2009, Galium normanii had disappeared from all 5 × 5 cm turfs in both the grassland experiments and the heath soil experiment, and Thalictrum alpinum was only found in 20 × 20 and 30 × 30 cm turfs in the heath soil experiment. Both G. normanii and T. alpinum were found in turfs of all sizes in the heath road verge experiment. The effect of treatment on forb cover was significant only in the heath soil experiment (Table S2).
Cover of evergreen dwarf-shrubs, mostly E. nigrum but also Loiseleuria procumbens, decreased with time in all treatments in both heath experiments (Fig. 3). Cover of deciduous dwarf-shrubs, mostly V. uliginosum but also Salix herbacea and Vaccinium myrtillus, increased with time in treatments with the largest turfs (Fig. 3). At the end of the third growing season, the Vaccinium species had disappeared from all plots with 5 × 5 cm turfs, and L. procumbens was only found in 20 × 20 and 30 × 30 cm turfs in the heath soil experiment, but these species were found in all treatments with whole turfs in the heath road verge experiment. No dwarf-shrubs were found in the grassland turfs.
Non-flowering vascular plants were uncommon at the donor sites. Selaginella selaginoides, a species with negligible cover at the dwarf-shrub heath donor site, was found in some heath turfs in 2007, but had disappeared from all treatments in 2009. Equisetum arvense started to colonize the grassland soil experiment during the first growing season and had 0·5–5% cover in experimental plots by 2009. Equisetum arvense was also found in low abundance in the other experiments.
Moss cover increased most rapidly in treatments with shredded turfs in all experiments (Fig. 4). Racomitrium ericoides was always the most common moss species. Sanionia uncinata, H. splendens and R. squarrosus were also common in some treatments with grassland turfs and R. lanuginosum in the treatments with heathland turfs.
The total area of grassland turfs increased with time in all treatments in the grassland soil experiment (Fig. S5, Supporting information). The increase was greatest in the treatment with 5 × 5 cm turfs, that is, 750%, but total turf area of the 10 × 10, 20 × 20 and 30 × 30 cm treatments increased by 650%, 430% and 240%, respectively. Total turf area increased by 12–88% in the grassland road verge experiment and 5–55% in the heath soil experiment during the same period. The increase was greatest in the 10 × 10 cm treatment and smallest in the 30 × 30 cm treatment in both experiments.
Colonization near the turfs
The heath soil experiment had on the average 7 seedlings m−2 in the second growing season and 6 seedlings m−2 in the third growing season (Table S3, Supporting information). Average density of seedlings and E. arvense in the grassland soil experiment was 4 m−2 in August 2007, but increased to 30 m−2 in 2008 and 144 m−2 in 2009. Only two seedlings were found on transects in the grassland road verge experiment in 2007, but seedling density was 17 plants m−2 in 2008 and 31 plant m−2 in 2009. Grass seedlings, especially Agrostis and Festuca spp., were the most abundant species colonizing outside the turfs in both the grassland experiments in 2008. Density of C. fontanum seedlings increased greatly between 2008 and 2009 in most treatments and was greatest in plots with shredded turfs (Fig. 5).
Vegetative sprouts of grasses (Festuca and Poa spp.) were found extending from the turfs or in fragments of shredded turfs. Their average density was usually <5 sprouts m−2.
Effects of turf size
Larger turfs generally had higher plant species richness than small turfs, but these differences among treatments with whole turfs were only significant in the soil experiments (Fig. 1). Rare species are more likely to disappear in community translocation than are common species (Buckner & Marr 1988), and non-grass species with negligible cover (<0·5%) at the donor sites disappeared from all experimental plots with the smallest turfs. Non-grass species with 2–6% cover at the donor site also disappeared from the smallest turfs in the heath soil experiment, but not in the heath road verge experiment that was accidentally seeded with grasses and fertilized in the second growing season. This indicates that not only turf size, but also management, differences in soil materials or other conditions of the receptor sites may have affected survival of transplants (cf. Buckner & Marr 1988; Good et al. 1999; Bay & Ebersole 2006).
The only sedge found at the donor sites, Carex bigelowii, forms tillers on long rhizomes with connections between many tiller generations, and severing these connections causes tiller mortality (Jónsdóttir & Callaghan 1988). C. bigelowii disappeared completely from the smallest turfs in the heath soil experiment by 2009; its cover decreased with time in the smallest grassland turfs in both the grassland experiments and was negligible in plots with shredded turfs. The smallest turfs had one or few interconnected tillers of C. bigelowii, which may explain its poor performance in those treatments. These results agree with earlier studies showing poor transplant survival of rhizomatous species (May, Webber & May 1982; Bay & Ebersole 2006; Cole & Spildie 2006).
Turf size had a significant effect on forb cover in the heath soil experiment (Fig. S4) where G. normanii and T. alpinum were found only in treatments with 20 × 20 and 30 × 30 cm turfs by 2009. Both species are rhizomatous (Grime, Hodgson & Hunt 2007), which can explain their susceptibility to division into small turfs. On the other hand, cover of G. normanii increased with time in most treatments of the heath road verge experiment that received an accidental fertilization in 2008. It was also found in plots with shredded turfs in 2009, possibly due to colonization from seed bank of the shredded turfs.
Evergreen dwarf-shrubs were susceptible to division into small turfs. The minimum turf size for the transfer of evergreen dwarf-shrubs may be even greater than the 30 × 30 cm turfs tested here, as their cover decreased with time in all treatments with whole turfs (Fig. 3). Empetrum nigrum, the most abundant dwarf-shrub species at the heath donor site, has a low root-shoot ratio (Karim & Mallik 2008) and spreads by creeping lateral shoots that can form adventitious roots where they touch the ground (Bell & Tallis 1973). If a sward with E. nigrum is divided into small turfs, the branches present in each turf may have limited or no adventitious roots and a low root-shoot ratio. This can limit their survival at the receptor area. Thickness of the turfs (10 cm) was, however, probably not limiting for E. nigrum establishment as most of the adventitious root mass is found within the topmost 10 cm of soils (Bell & Tallis 1973; Karim & Mallik 2008).
Cover of deciduous dwarf-shrubs decreased with time in treatments with 5 × 5 and 10 × 10 cm turfs but increased in plots with 30 × 30 cm turfs (Fig. 3). The minimum size of turfs for the transplantation of deciduous dwarf-shrubs could, therefore, be somewhat smaller than for evergreen shrubs, or 20–30 cm. The most abundant dwarf-shrub species, Vaccinium uliginosum and S. herbacea, spread by rhizomes (Jacquemart 1996; Beerling 1998) and adventitious roots are readily formed on S. herbacea branches (Hagen 2002) which may increase its tolerance to division into turfs.
Moss cover increased with time in all treatments, but most rapidly in plots with shredded turfs (Fig. 4). The shredded turfs contained whole moss branches and branch fragments that can be effective propagules for the distribution of moss species (McDonough 2006; Malson & Rydin 2007). This applies to R. ericoides, R. lanuginosum, H. splendens and R. squarrosus (Magnúsdóttir & Aradóttir 2011) that were common colonizers in the experiments at Hellisheiði. The heath road verge experiment that was accidentally fertilized in 2008 had by far the highest moss cover (Fig. 4), which is consistent with earlier studies showing stimulation of moss cover by revegetation actions that include fertilization (Elmarsdottir, Aradottir & Trlica 2003; Gretarsdottir et al. 2004). The results of the heath road verge experiment suggest that colonization of mosses, especially R. ericoides, can be facilitated by combining the spreading of moss fragments and fertilization.
Colonization by seed or vegetative spread
Colonization by seed was limited in the heath soil experiment, but increased with time in both grassland experiments. The whole turfs apparently did not serve as seed sources for colonization, as observed by Klimešet al. (2010), because seedling density was never significantly higher in plots with whole turfs than control plots (Fig. 5, Table S3).
Equisetum arvenses, a competitive ruderal that can form extensive rhizomatous patches (Grime, Hodgson & Hunt 2007), was a common colonizer in the grassland soil experiment (Table S3). It is, however, unlikely that the E. arvensis originated from the turfs as it had much higher cover in the control plots than other treatments.
Density of vegetative sprouts that extended from the turfs was low. On the other hand, expansion of turfs in the grassland soil experiment (Fig. S5), especially in the treatments with largest edges to area ratio of turfs (5 × 5 and 10 × 10 cm), seemed to result from vigorous tillering of grasses in the turfs, comparable to expansion of 15 × 15 cm tundra plugs in Alaska observed by Kidd, Streever & Jorgenson (2006).
Whole vs. shredded turfs
The spreading of excavated soil and vegetation (sod dumping) can yield a transfer rate of species that is comparable to the use of whole turfs (Good et al. 1999; Kiehl et al. 2010). Most studies of sod dumping have, however, used soil depth of up to 50 cm and donor-to-receptor site area ratio of 1 : 1 (reviewed by Kiehl et al. 2010), that is, much higher concentration of sod than the 1 : 50 ratio and 10-cm soil thickness used in this experiment.
Shredded turfs were most effective for the transfer of mosses as discussed earlier. Few grass tillers seemed to survive the shredding as grasses had very low cover during the first growing season in plots with shredded turfs (Fig. 2), but established there from seed in subsequent growing seasons (Table S3). The rhizomatous C. bigelowii, T. alpinum and dwarf-shrub species did not colonize plots with shredded turfs. On the other hand, species with low abundance at the donor sites, L. multiflora and C. fontanum, established readily from seed in the plots with shredded turfs (Fig. 5), indicating colonization from seed bank of the turfs.
The overall effects of using whole turfs in comparison with shredded turfs varied greatly between species and growth forms. Consequently, the use of shredded turfs may lead to a vegetation composition that differs more from the donor sites than would be expected when whole turfs are used. Furthermore, C. fontanum, the most abundant species allegedly colonizing from seed bank of the shredded turfs, is a ruderal that can be a prolific colonizer of disturbed sites (Grime, Hodgson & Hunt 2007). The value of introducing such ruderal species into restoration areas may be limited.
Conditions at receptor sites
Conditions at receptor sites can affect the success of turf transfer, especially if they are very different from the donor sites (Bullock 1998). The coarse gravel substrate of the road verge experiments differed substantially from the donor sites as they lacked fine soil fragments and probably had very low nutrient and water-holding capacity. The survival of individual species and colonization in the grassland road verge experiment was limited, but the additional fertilization of the heath road verge experiment had a positive effect on the survival of individual species, as well as on the cover of mosses and grass species (Figs 2 and 4). Fertilization can also stimulate tillering of C. bigelowii (Carlsson & Callaghan 1990), which may explain why this species was found in all treatments of the heath road verge experiment, but only in turfs ≥20-cm diameter in the heath soil experiment. Likewise, dwarf-shrubs persisted in plots with smaller turfs in the heath road verge experiment than in the heath soil experiment (Fig. 3), which might also be associated with the greater fertilization of this treatment. Other studies in Iceland have shown that revegetation by seeding with grasses and fertilization can facilitate the colonization of many dwarf-shrub species (Gretarsdottir et al. 2004). Furthermore, fertilization can stimulate growth of E. nigrum and Vaccinium spp. (Parsons et al. 1994), although other studies have shown negative effects of fertilization on the proportion of dwarf-shrubs in undisturbed plant communities (e.g. Gough, Wookey & Shaver 2002; Kelley & Epstein 2009).
Signs of extensive frost heaving, that probably hampered seedling survival and plant establishment, were observed in the heath soil experiment and to a lesser degree in the grassland soil experiment. Furthermore, the shallow surface layer of the heath soil experiment dried out during periods of low precipitation each summer. A combination of frost heaving and drought may partly explain the poor results from turf transplantation in the heath soil experiment. On the other hand, survival of individual species in the turfs was greatest in the heath road verge and grassland soil experiments, and turf expansion was greatest in the grassland soil experiment (not measured in the heath road verge experiment).
The effects of turf extraction on the donor sites were not measured, but observation of the extraction plots in 2009 revealed some colonization of mosses and a few vascular plant species. The vegetation cover was, however, still low and the effects of the turf extraction will probably be long lasting.
Conclusions and recommendations
In this study, turfs from grasslands and dwarf-shrub heath served as a source for colonization of many native species into disturbed areas, but optimal turf size varied between functional groups of plants, decreasing in size as follows: evergreen dwarf-shrubs > deciduous dwarf-shrubs > sedges > grasses > mosses.
Selection of turf size for transplanting should be based on factors such as the objectives of the transplantation, the donor vegetation and the growth form, size and abundance of the target species. If community translocation and preservation of rare species are the objective, it may be necessary to use larger turfs than were used in this study (>30-cm diameter) to increase the chances of transplanting viable individuals of those rare species. On the other hand, if introduction of native species is the main objective, optimal turf size depends on the target species. Successful transplantation of dwarf-shrubs may require turfs that are at least 20–30 cm in diameter, while turfs as small as 5 cm in diameter can be used to establish many grasses. Such small turfs are easy to transplant with standard tree planting tools. Distribution of shredded turfs containing moss branches or branch fragments can be used to accelerate the formation of moss cover.
Turf expansion and establishment varied depending on the growing conditions of the receptor sites. Accidental fertilization or other factors contributing to favourable conditions at some of the receptor sites seem to have facilitated survival and establishment of species with low abundance at the donor site. This suggests that fertilization and other interventions to enhance growing conditions at the receptor sites could be used to improve establishment and vegetation expansion after transplantation of turfs in infertile sites. Controlled experiments are needed to clarify the effects of environmental factors at the receptor sites.
Transplanting of native turfs is a promising method of establishing a diverse range of native species on disturbed areas, but disturbance of natural vegetation for the sole purpose of turf extraction may not be justified. Sward from roadbeds, mine sites and other areas that are being stripped of vegetation during construction is, on the other hand, a valuable source of native species for use in the restoration of disturbed areas. A strategy to salvage whole turfs of sward should, therefore, be included in plans for construction projects that cause disturbance to natural vegetation.
This project was planned and implemented in cooperation with Orkuveita Reykjavíkur (OR) at the construction site of the Hellisheiði Geothermal Plant. The project was supported by the OR Environmental- and Energy Research Fund. OR provided logistic support and labour for implementation and management of the experimental sites, under the supervision of Herdís Friðriksdóttir. I also thank Erla Sturludóttir, Karólína Einarsdóttir, Brita Berglund, Sunna Áskelsdóttir and Ásta Kristín Guðmundsdóttir for their work on the field surveys, and Járngerður Grétarsdóttir, Jón Guðmundsson and Ingibjörg Svala Jónsdóttir for valuable discussions during the planning of the project. Finally, I thank Ólafur Arnalds, two anonymous referees and the editors for comments that improved the manuscript.