Effects of slash harvest on bryophytes and vascular plants in southern boreal forest clear-cuts



    Corresponding author
    1. Landscape Ecology Group, Department of Ecology and Environmental Science, Umeå University, SE-901 87, Umeå, Sweden
      M. Åström, Landscape Ecology Group, Department of Ecology and Environmental Science, Umeå University, SE-901 87, Umeå, Sweden (fax + 46 90 7867860; e-mail marcus.astrom@emg.umu.se).
    Search for more papers by this author

    1. Landscape Ecology Group, Department of Ecology and Environmental Science, Umeå University, SE-901 87, Umeå, Sweden
    Search for more papers by this author

    1. Landscape Ecology Group, Department of Ecology and Environmental Science, Umeå University, SE-901 87, Umeå, Sweden
    Search for more papers by this author

    1. Landscape Ecology Group, Department of Ecology and Environmental Science, Umeå University, SE-901 87, Umeå, Sweden
    Search for more papers by this author

M. Åström, Landscape Ecology Group, Department of Ecology and Environmental Science, Umeå University, SE-901 87, Umeå, Sweden (fax + 46 90 7867860; e-mail marcus.astrom@emg.umu.se).


  • 1Slash harvesting from forests to provide bioenergy reduces the amount of woody debris in the managed forest landscape and changes the physical and chemical environment in clear-cuts. We examined previously unstudied effects of commercial (i.e. non-experimental) slash harvest on species composition and richness of liverworts, mosses and vascular plants. The results call for modification of commercial slash harvest practices.
  • 2Differences between conventionally harvested (i.e. slash left) and slash-harvested stands were investigated 5–10 years after clear-cutting through analysis of 28 paired stands, with one 0·1-ha plot divided into five 0·02-ha subplots in each stand.
  • 3The species composition of mosses and liverworts in 0·1-ha plots was significantly affected by slash harvest, whereas the composition of vascular plant species was not.
  • 4The species richness of liverworts was significantly reduced by slash harvest in plots of both sizes, whereas moss richness was reduced only in 0·02-ha plots. The loss of liverwort species was largest, with approximately one-third of the species disappearing. The species richness of vascular plants was not significantly affected by slash harvest in either plot size.
  • 5Slash harvest reduced species richness of forest bryophytes and of bryophytes typically growing on organic substrates in open habitats. Species richness of non-forest bryophytes on inorganic substrates remained unchanged.
  • 6Synthesis and applications. Our results show that slash harvest reduces shelter and woody substrates, which changes species composition and reduces species richness of liverworts and mosses in clear-cuts. Increased mechanical disturbance that removes remnant vegetation and exposes mineral soil may also play a role. In order to conserve bryophytes, we advocate mitigation of adverse ecological effects through enhanced environmental care within slash-harvested stands. Leaving more tree clusters, and creating and protecting large woody debris would be especially important in these stands, and would also improve the habitat for other organisms.


Converting fossil-based energy systems into CO2-neutral systems such as those based on bioenergy is an essential strategy for reaching targets set by the Kyoto Protocol (Gustavsson et al. 1995; Gururaja 2003). For those nations with forestry industry there is potential to exploit slash (i.e. logging residues composed of treetops, branches and twigs) for bioenergy purposes (Egnell & Valinger 2002). In fact, wood fuels produced as by-products of clear-cutting in coniferous forests are already an important energy resource (Doherty, Nilsson & Odum 2002). In Sweden and other timber-producing nations, slash harvesting is likely to become more widespread, both for bioenergy production and because it facilitates regeneration of commercial timber species (National Board of Forestry 2001; Hakkila 2003).

Disturbance dynamics in managed boreal forest ecosystems are constrained by the exclusion of natural disturbances such as fire and insects outbreaks (Esseen et al. 1997). Forestry practices based on clear-cutting, thinning and short rotation periods produce small amounts of decomposing wood compared with natural disturbances (Fridman & Walheim 2000; Siitonen 2001). In this context, slash harvest after clear-cutting can be considered an intensification of forest management, further reducing the amount of decomposing wood in the forest landscape. This practice may counteract efforts to enhance conditions for wood-inhabiting organisms in managed boreal forests. Therefore, it is important to assess the impacts of commercial slash harvesting.

In experimental studies, slash harvesting has been reported to affect post-harvest vegetation dynamics (Olsson & Staaf 1995; Bråkenhielm & Liu 1998; Bergquist, Örlander & Nilsson 1999) and to reduce tree growth (Egnell & Leijon 1999; Egnell & Valinger 2002) and the abundance of soil arthropods as a result of decreased input of organic material and nutrients (Bengtsson, Persson & Lundkvist 1997). Removal of woody debris also reduces habitat and substrate for many species of plants, animals and fungi (Samuelsson, Gustafsson & Ingelög 1994). Although most studies have focused on the importance of coarse woody debris (Loeb 1999; Lohr, Gauthreaux & Kilgo 2002), fine fractions of decomposing wood are important for many wood-living fungi, lichens and bryophytes (Kruys & Jonsson 1999; Nordén et al. 2004), especially in areas with little coarse woody debris (Kruys & Jonsson 1999). In clear-cuts, slash provides shelter, reducing wind velocity and fluctuations in ground surface temperature (Mahendrappa & Kingston 1994; Proe, Dutch & Griffiths 1994). It contributes to structural heterogeneity, and may affect species richness and abundance of organisms such as small mammals (Ecke, Löfgren & Sörlin 2002) and ground-active beetles (Gunnarsson, Nittérus & Wirdenäs 2004). Slash may also shelter plants sensitive to desiccation, especially during the clear-cut phase (cf. McInnis & Roberts 1994; Bråkenhielm & Liu 1998).

Previous studies of the effects of slash harvesting on vegetation dynamics have been experimental, restricted to few sites and small plots, with the result that effects were evaluated only for a few abundant species (Olsson & Staaf 1995; Nykvist 1997; Bergquist, Örlander & Nilsson 1999). We present a more realistic and comprehensive study, assessing the impact of commercial slash harvest on entire communities of liverworts, mosses and vascular plants using large plots in many clear-cut stands. By comparing matched pairs of conventional and slash-harvested stands, we investigated effects on species composition and species richness. We also analysed the effects on mosses and liverworts grouped according to their habitat and substrate associations.


The study was conducted in south central Sweden, between latitudes 58°40′ and 59°35′N. Study sites were distributed over the western and southern parts of Örebro County in the southern boreal zone, described by Ahti, Hämet-Ahti & Jalas (1968). The forests in the area have been modified over several centuries by repeated extraction of timber, initially to supply the mining industry (Blomberg & Holm 1902), and later the forestry industry. Clear-cutting has been the main mode of timber extraction for the last 50 years (Wastenson & Nilsson 1995). More recently, the area has increasingly been subjected to slash harvesting. The length of the growing season ranges from 180 to 190 days and annual precipitation is 700–800 mm (Sjörs 1999). Mean temperature in July is 15–16 °C (Sjörs 1999). The parent soil material at our study sites consists of moraine with underlying bedrock of gneiss or granite. The soils of coniferous forests are chiefly podsolic (Fredén 1994).

To select pairs of similar clear-cut stands, one of which had been subjected to slash harvest, we used a forest database provided by Sveaskog (Sweden's publicly owned forest company) that contained information describing approximately 8000 stands harvested between 1990 and 1997. Stand pairs were selected subjectively to minimize between-plot distance as well as differences in time since logging, area, altitude, temperature sum, slope and surface structure, as well as pre-harvest vegetation type, productivity and soil moisture. Using these criteria, we initially selected 79 pairs with high intrapair similarity. All stands were then examined in the field in 2002 and 2003, and pairs showing large differences in past non-forestry activities, green tree retention or topographical variability were excluded. In clear-cuts subjected to commercial slash harvesting, site preparation for planting is normally made 1 year earlier than on clear-cut areas where slash is not harvested. This practice is part of the treatment and therefore not accounted for in the matching procedure. After exclusion, 14 pairs (28 stands) remained for further analysis (Table 1). All 28 stands were on mesic sandy till and were formerly dominated by Norway spruce Picea abies L. Karst. Site indices (dominant height in metres at age 100 years) ranged from 24 to 33 m. The stands were all clear-cut between 1993 and 1997, site-prepared with similar methods and planted with spruce. There is no record of recent liming or fertilization in any of the stands selected. Intrapair distances ranged from 0·4 to 10·1 km, with a mean of 3·8 km. Within each stand we established a representative 20 × 50-m (0·1-ha) sample plot. Sample plots were situated at least 20 m from the nearest forest edge and were placed to obtain matching plots with high degrees of pairwise similarity with respect to cover of stones, boulders, moist ground, coarse woody debris (CWD) and small trees (1·5 < height < 5 m). Cover of slash and mesic ground was not included in the matching process, as these variables are strongly affected by slash harvesting.

Table 1.  Properties of conventionally harvested and slash-harvested stands. Data are means and ranges for each variable and treatment, and are derived from the forestry database of the company Sveaskog. Deviation describes mean and maximum differences (absolute values) between stands in a pair. n equals the number of pairs for which data are available
Variable n Conventional harvestSlash harvestDeviation
  • *

    Pre-harvest vegetation type ordered with decreasing productivity. 1, high herb; 2, low herb; 3, broadleaved grass; 4, thin-leaved grass; 5, blueberry.

  • Productivity expressed as dominant height (m) at age 100 years.

  • Pre-harvest ground moisture. 1, dry; 2, mesic; 3, moist; 4, wet.

  • §

    1, very smooth surface; 2, smooth; 3, somewhat coarse; 4, coarse.

  • 1, 0–5%; 2, 5–15%; 3, 15–33%; 4, 33–50%.

Years since cutting14   7·75·6–9·7   7·95·4–10·3 0·2 1·6
Area (ha)14   7·42·8–14·8   8·52·5–16·4 3·4 7·7
Altitude (m a.s.l.)14 16995–260 170100–23018·680
Temperature sum1413001173–143613001200–143614·572
Vegetation type* 9   3·82–5   3·82–5 0·9 2
Site index14  2826–32  28·824–33 1·9 6
Soil moisture14   2·02–2   2·02–2 0·0 0
Surface structure§11   2·22–4   2·22–3 0·2 1
Slope11   2·51–4   2·01–3 1·0 2

Data collection in the field was conducted in July 2002 and 2003, when the clear-cuts were 5–10 years old. The 0·1-ha sample plots were divided into five equally sized subplots of 0·02 ha (10 × 20 m).

In each subplot, we quantified cover of CWD, including stumps (diameter > 10 cm) and logs (mid-bole diameter > 10 cm, length > 1 m). Observed CWD was classified according to degree of decay as: (i) wood hard; (ii) wood starting to soften; (iii) wood soft with pieces lost but outline of stump or log still clearly visible; and (iv) wood soft and deformed so outline of stump or log not clearly visible (modified after Söderström 1988). Cover was calculated using cut surface diameter (cross-measured) and circle approximation for stumps, and length, mid-bole diameter and rectangle approximation for logs. Cover of slash with a diameter < 10 cm mid-bole was estimated visually, as was the cover of stones (diameter 0·2–0·6 m) and boulders (diameter 0·6–2·0 m) not overgrown by understorey vascular plants but occasionally overgrown by bryophytes. Soil surface not covered by slash was classified as mesic or moist ground (the only types present in this study), according to the system of Hägglund & Lundmark (1987), and the cover of each type was estimated visually. Using the subplot data, we calculated total cover of each substrate (logs and stumps of different stages of decay, slash, boulders, stones, mesic and moist soil) within the entire 0·1-ha plot. Together these substrates covered 100% of the 0·1-ha plot. Substrates considered in the matching process (i.e. all substrates except slash and mesic ground) were integrated into an index of substrate heterogeneity using the Shannon index of diversity (Shannon & Weaver 1949). This index incorporates the number of substrates at each 0·1-ha plot, weighted by their relative cover.

Each 0·02-ha subplot was inventoried for species of mosses, liverworts and vascular plants. From the subplot records we calculated the number of species present in each 0·1-ha plot. Frequency within the 0·1-ha plots was expressed as number of subplots where a species was present (0–5). For mosses and liverworts, notes were taken regarding each species’ occurrence on different substrates. The cover in each 0·1-ha plot was estimated visually for (i) small trees (1·5 m < height < 5 m), (ii) shrubs and tree saplings (0·5 m < height < 1·5 m), (iii) dwarf shrubs, (iv) herbs plus ferns, (v) graminoids and (vi) bryophytes.

ecological classification of bryophytes

Using personal experience from the region cross-checked with information found in the literature (Hallingbäck 1996), mosses and liverworts were sorted into ecological groups based on their habitat (forest or non-forest) and substrate (organic or inorganic) associations (see Appendices S1 and S2). Non-forest species mainly inhabit open habitats such as road- and riverbanks, open mires and rocky outcrops. Organic substrate species primarily grow on logs, stumps, branches, litter and humus, whereas inorganic substrate species are mainly associated with boulders, stones and exposed mineral soil. Three taxa, Hypnum cupressiforme Hedw., Pellia sp. and Cephalozia bicuspidata (L.) Dum., could not be classified because of the breadth of their habitat or substrate preferences. These taxa were excluded from all analyses involving ecological groups.


The nomenclature for vascular plants and bryophytes follows Karlsson (1997) and Söderström & Hedenäs (1998), respectively. A few species groups were treated as a single aggregate species: Hieracium subsect. silvaticiformia, Hieracium subsect. vulgatiformia, Taraxacum sp. (vascular plants), Brachythecium oedipodium (Mitt.) Jaeg. + Brachythecium rutabulum (Hedw.) Schimp., Brachythecium reflexum (Starke) Schimp. +Brachythecium starkei (Brid.) Schimp., Dicranum fuscescens Sm. + Dicranum flexicaule Brid., Plagiothecium laetum Schimp. + Plagiothecium curvifolium Limpr. (mosses), Cephaloziella sp., Cephalozia lunilifolia (Dum.) Dum. +Cephalozia affinis Steph., Lophozia silvicola Buch +Lophozia ventricosa (Dicks.) Dum. and Pellia sp. (liverworts). Voucher specimens of bryophyte species have been deposited in the herbarium UME at Umeå University, Umeå, Sweden.

statistical analyses

We used ordination analysis (canoco for Windows Version 4·0; ter Braak & Šmilauer 1997, 1998) to analyse differences between 0·1-ha conventionally harvested plots (CHP) and slash-harvested plots (SHP) in (i) vascular plant, moss and liverwort species composition and (ii) composition of substrates besides slash and mesic ground. We first applied detrended canonical correspondence analysis (DCCA) to test gradient length. As the longest gradient was shorter than 3 and we were interested in treatment effects, we then used partial redundancy analysis (pRDA, with default settings), which is a direct ordination method assuming a linear relationship across ordination axes (Lepš & Šmilauer 2003). We applied treatment (conventional vs. slash-harvested, transformed into a dummy variable) as the explanatory variable and used frequency (0–5) and cover (m2) as input data for species and substrates, respectively. Because of the pairwise design, we used pair numbers as covariables in the analysis. A Monte Carlo permutation test was used (with 500 permutations) to test if the explanatory variable significantly influenced species or substrate composition. In an anova we tested if there was any difference in the pRDA scores (i.e. the response to the treatment) between species of different habitat (forest or non-forest) or substrate (organic or inorganic) associations.

Differences in substrate abundance, substrate heterogeneity and species growth forms were analysed pairwise for the 0·1-ha plots. Differences in species richness of mosses, liverworts vascular plants and bryophytes (i.e. mosses + liverworts) divided into ecological groups were compared pairwise for the two plot sizes. For the 0·02-ha subplots we used mean values from all five subplots. Differences in frequencies of individual species were analysed pairwise for species present in at least four pairs of 0·1-ha plots. As some differences were not normally distributed we used the Wilcoxon's signed rank test for the pairwise comparisons throughout the study. Although the analyses may include multiple comparisons, the main purpose was exploratory and no posterior correction of P-values was made. Pearson's correlation analysis was used. SPSS for Windows Version 10·0 was used for all statistical analyses except composition analysis. (Norušis 1999).


CHP and SHP were similar in cover of substrates besides slash and mesic ground (Table 2). Slash cover was 79 m2 in CHP and 18 m2 in SHP, which corresponded to a 4·5-fold difference between treatments (P = 0·001, Wilcoxon's signed rank test). More mesic ground was exposed in SHP (935 m2) than in CHP (884 m2; P= 0·003, Wilcoxon's signed rank test). There were significant negative correlations between cover of slash and of mesic ground in both treatments (SHP, R= 0·55, P= 0·041; CHP, R= 0·56, P= 0·039; Pearson's correlation), implying that the observed difference in cover of mesic ground was caused by the slash harvest. For matched substrates (i.e. slash and mesic ground excluded) there was no significant difference in either substrate heterogeneity (P = 1·000, Wilcoxon's signed rank test; Table 2) or substrate composition (P = 0·37, pRDA, F-ratio 1·09).

Table 2.  Cover (m2) and heterogeneity of matched (i.e. controlled for during site and plot selection) substrates in conventionally harvested plots (CHP) and slash-harvested plots (SHP). Deviation is the mean and maximum difference (absolute values) between the two plots in each pair. P-values were derived from Wilcoxon's signed rank test. Plot size is 0·1 ha. n is 14 for each treatment
VariableDecay class*CHPSHPDeviationP
  • *

    Decay class of coarse woody debris (CWD). See the Methods.

  • Substrate heterogeneity index. Cover of the substrates presented in the table (matched substrates) synthesized using the Shannon index of diversity.

Stumps1 4·82·7–6·7 4·82·7–6·70·1 2·60·950
2 0·00–0·2 0·10–0·40·1 0·40·120
3 0·10–0·7 0·10–0·50·1 0·70·937
4 0·30–0·9 0·30–0·80·1 0·60·801
Logs1 1·60–3·8 1·40–7·01·7 5·00·470
2 0·30–1·5 1·10–7·81·2 7·80·484
3 0·40–2·0 0·20–1·60·5 1·80·463
4 0·20–1·4 0·00–00·2 1·40·068
Boulders 160–28161·5–396130·830
Stones  91–18141–378220·170
Moist ground  80–55100–468460·890
Substrate heterogeneity  1·31·0–1·7 1·31·1–1·60·1 0·51·000

Slash harvest reduced bryophyte cover by half and raised graminoid cover by 10% but had no significant effect on other growth forms (Table 3). Cover of small trees, which we controlled for, did not differ between CHP (103 m2) and SHP (121 m2; P= 0·154; Wilcoxon's signed rank test).

Table 3.  Cover (m2) of different growth forms in conventionally harvested plots (CHP) and slash-harvested plots (SHP). Differences were tested using the Wilcoxon's signed rank test to reveal treatment effects. Significant differences (P < 0·05) are boldfaced. Plot size is 0·1 ha. n is 14 for each treatment
Growth formCHPSHP P
Shrubs and saplings22344137190·158
Dwarf shrubs 5110 83210·221
Graminoids7183078726 0·024
Herbs and ferns 6814 66140·272
Bryophytes2163010823 0·003

Species composition of both mosses and liverworts differed significantly between treatments (P = 0·006, pRDA, eigenvalue axis 1 = 0·062, F-ratio 2·6; P= 0·002, pRDA, eigenvalue axis 1 = 0·123, F-ratio 4·96, respectively). Ten species of bryophytes were more frequent in CHP, whereas only one species (Polytrichum commune Hedw.) was more frequent in SHP (P < 0·05, Wilcoxon's signed rank test; see Appendices S1 and S2).

Slash harvest did not change species composition of vascular plants significantly (P = 0·15, pRDA, F-ratio 1·45). Only four species differed significantly in frequency between treatments (P < 0·05, Wilcoxon's signed rank test; see Appendix S3). Juncus conglomeratus L., Luzula multiflora (Ehrh) Lej. and Populus tremula L. were more frequent in SHP, whereas Senecio sylvaticus L. was less frequent (see Appendix S3).

species richness

We encountered 65 moss species, 29 liverwort species and 131 vascular plant species in the 28 plots (Appendices S1, S2 and S3). Relative to CHP, species richness of mosses was reduced in SHP, but the reduction was significant only for the smaller plot size (1·9 species or 8%; Fig. 1a). The reduction in species richness was larger for liverworts, both in small plots (2·2 species or 39%) and in large plots (3·6 species or 33%; Fig. 1). Slash harvesting tended to increase species richness of vascular plants but this tendency was not statistically significant either in small plots (2·3 species or 8%) or in large plots (2·3 species or 5%; Fig. 1).

Figure 1.

Species richness of mosses, liverworts and vascular plants. Shaded and unfilled bars are conventionally harvested and slash-harvested plots, respectively. Plot sizes are (a) 0·02 ha and (b) 0·1 ha. n is 14 for each treatment. Data are mean values ± 1 SE. P-values were derived from Wilcoxon's signed rank test.

ecological groups of bryophytes

Forest and non-forest liverwort species differed significantly in their response to slash harvesting (P = 0·024, anova of the pRDA scores; Table 4), where forest species were more negatively affected (Fig. 2a). No significant effect of habitat association was found for mosses (Fig. 2b. P= 0·55, anova; Table 4). Substrate association (organic vs. inorganic) had a nearly significant effect on the response to slash harvesting of moss species (P = 0·052, anova of the pRDA scores; Table 4) but not of liverwort species (P = 0·15, anova; Table 4). Mosses associated with organic substrates tended to be negatively affected by slash harvest, whereas non-forest mosses and liverworts associated with inorganic substrates tended to be favoured (Fig. 2).

Table 4.  Univariate analysis of variance of pRDA scores for moss and liverwort species tested against their habitat (forest vs. non-forest) and substrate (organic vs. inorganic) associations. pRDA scores reflect species’ responses to slash harvest. Habitat and substrate associations are considered fixed factors
 d.f.Mean square F P R 2 inline image
Mosses (n = 64)
Corrected model 30·0922·4140·0750·1080·063
Intercept 10·2085·4530·023  
Habitat 10·0140·3570·550  
Substrate 10·1503·9340·052  
Habitat × substrate 10·0711·8610·180  
Liverworts (n = 27)
Corrected model 30·1795·3240·0060·4100·333
Intercept 10·2778·2250·009  
Habitat 10·1985·8850·024  
Substrate 10·0742·1950·152  
Habitat × substrate 10·0050·1480·704  
Figure 2.

Box-plot for pRDA scores of species classified into different ecological groups of (a) liverworts and (b) mosses. Positive pRDA scores indicate a negative impact of slash harvesting. Boxes and whiskers represent median, interquartile range and the 5th and 95th percentile values. Outliers are indicated. n equals the number of species within each ecological group. For pRDA scores of individual species see Appendices S1 and S2.

Species richness of forest bryophytes associated with organic substrates was reduced in SHP by 20% and 16% in small and large plots, respectively, relative to CHP (Fig. 3). Slash harvesting also reduced the number of non-forest species associated with organic substrates by 14% and 16% in small and large plots, respectively (Fig. 3). In contrast, slash harvesting did not affect the number of species associated with inorganic substrates (neither forest nor non-forest species; Fig. 3).

Figure 3.

Species richness of ecological groups of bryophytes (liverworts + mosses). Shaded and unfilled bars are conventionally harvested and slash-harvested plots, respectively. The classification is based on a combination of substrate and habitat associations. Plot sizes are (a) 0·02 ha and (b) 0·1 ha. n is 14 for each treatment. Data are mean values ± 1 SE. P-values were derived from Wilcoxon's signed rank test.


We found that commercial slash harvesting has a significant effect on the species composition and richness of both mosses and liverworts in clear-cut stands in southern boreal spruce forests. There are at least three possible causal mechanisms for the observed differences. First, loss of woody debris influences composition and richness of species able to colonize this substrate in open clear-cut habitats. There was a high degree of post-cutting colonization by bryophytes in piles of slash and litter at the investigated sites (M. Åström, personal observation). The effect of substrate association on the response of moss species (Table 4) and the reduction in richness of bryophytes on organic substrates (Fig. 3) also suggest that slash may be an important substrate affecting bryophyte composition and richness after clear-cutting.

Secondly, slash may help reduce wind velocities and ground temperature fluctuations (Mahendrappa & Kingston 1994; Proe, Dutch & Griffiths 1994; Proe, Griffiths & McKay 2001). Many forest bryophyte species and a particularly high proportion of liverwort species are negatively affected by reduced shelter (Kurschner 1999; Hylander, Jonsson & Nilsson 2002; Hylander et al. 2005). In an experimental study of a relatively nutrient-poor Pinus sylvestris L. stand, bryophytes declined faster and showed lower rates of survival in plots where slash had been removed (Bråkenhielm & Liu 1998). Bråkenhielm & Liu (1998) suggested that slash may provide shelter from environmental extremes and may protect species intolerant to desiccation. The significant influence of habitat association (forest vs. non-forest) on liverwort response to slash harvest suggests that changes in liverwort composition are at least partly linked to microclimatic effects (Table 4). Thus the shelter effect of slash may be important, particularly for liverwort communities during the clear-cut phase.

Thirdly, slash harvest requires additional machinery and results in greater mechanical disturbance of the ground surface (Kardell 1992). Moreover, reduced slash cover in SHP enables more effective site preparation that increases soil disturbance (Hakkila 2003). Increased mechanical disturbance may cause extinction of remnant populations (Jalonen & Vanha-Majamaa 2001) but may also stimulate establishment of new populations (Granström 1986), thereby contributing to changes in bryophyte cover, composition and richness. In some cases the cover of common moss genera such as Polytrichum increases as a result of experimental slash harvesting (Kardell 1992; Olsson & Staaf 1995). In our data, all three Polytrichum species had higher mean frequency occurrence in SHP (although the difference was statistically significant at α= 0·05 only for Polytrichum commune Hedw.), probably reflecting reduced slash cover and higher mechanical disturbance of the ground. This is an example of the tendency of some species, particularly pioneers on disturbed ground, to be favoured by slash harvesting (Figs 2 and 3). The smaller effects on moss relative to liverwort species richness, along with the marginally significant explanatory power of the substrate association on the response of moss species, probably reflects a relatively high ability of many moss species to tolerate changes in microclimate and/or to colonize disturbed ground or marginally decomposed dead wood in clear cuts (Table 4 and Fig. 2).

Previous experimental studies have reported both reductions and increases in cover of common vascular plant species in response to slash harvesting (Olsson & Staaf 1995; Nykvist 1997; Bråkenhielm & Liu 1998). It has been suggested that the effects of slash harvesting on vascular plants result from reduced nitrogen inputs (Olsson & Staaf 1995; Bråkenhielm & Liu 1998), fewer physical barriers preventing regeneration (Fahey et al. 1991; Olsson & Staaf 1995) and reduced survival of remnant forest vascular plants (Bergquist, Örlander & Nilsson 1999). Our results, being more representative of the effects of commercial slash harvesting, suggest that the vascular plant community is not seriously affected, even though minor changes may occur. The higher graminoid cover in SHP is most probably caused by the removal of slash as a physical barrier preventing graminoid expansion (Fahey et al. 1991; Olsson & Staaf 1995) but also by mechanical disturbance favouring establishment of pioneer graminoid species (Granström 1986). Increased disturbance is also the most likely reason for the higher frequencies of Juncus conglomeratus L., Luzula multiflora (Ehrh) Lej. and Populus tremula L. in SHP, a result that corroborates previous studies suggesting better establishment of pioneering taxa following slash removal (Fahey et al. 1991; Olsson & Staaf 1995).

Under current commercial slash harvest practices in Sweden approximately 70% of the needle-mass is retained, mitigating negative impacts on the soil nitrogen budget (Egnell et al. 1998). This suggests that nitrogen-mediated effects (i.e. lower nitrogen inputs in SHP), particularly on nitrophilous vascular plants, such as Rubus idaeus L. and Epilobium angustifolium L., should be less severe in practice than has been indicated by previous experimental studies (Olsson & Staaf 1995; Bråkenhielm & Liu 1998; Bergquist, Örlander & Nilsson 1999). We found no significant effect of slash harvesting on the frequency of these species (see Appendix S3). In fact, Epilobium angustifolium showed a tendency to be more frequent in SHP, which supports the findings of Kardell (1983). Effects of nitrogen inputs are mostly short-term (< 10 years; Kellner 1993) and differences between CHP and SHP in soil nitrogen content may disappear within 4 years because of rapid expansion of field layer vegetation (Staaf & Olsson 1994). Hence our study, dealing with the 5–10 years of succession following clear-cutting, should have detected nitrogen-mediated effects of slash harvest on vascular plant communities. Furthermore, it is not likely that such effects will appear in later stages of succession in our studied stands. However, nitrogen-mediated effects of slash harvest may vary geographically because of differences in anthropogenic nitrogen deposition (cf. Bergquist, Örlander & Nilsson 1999). In our study area, nitrogen deposition is high (Lövblad et al. 1995) and may counteract nitrogen losses and subsequent effects on vascular plants in SHP. We cannot rule out the possibility of short- or long-term nutrient effects on plants in regions with lower rates of nitrogen deposition.

Slash harvesting also results in reduced pH (Egnell et al. 1998) that may negatively affect plant species, including bryophytes, whose establishment is stimulated by high pH (Olsson & Kellner 2002; Wiklund & Rydin 2004). The germination of Senecio sylvaticus L., often encountered in piles of slash, is stimulated by high pH (Olsson & Kellner 2002). This species was more frequent in CHP, which is in accordance with the findings of Kardell (1983) and Bergquist, Örlander & Nilsson (1999). The effects of slash harvesting on vascular plant species richness have only received cursory study. Kardell (1992) reported a slight increase in species number, whereas Bergquist, Örlander & Nilsson (1999) found a minor reduction. The tendency towards increased species richness in our data is in accordance with Kardell's (1992) result. However, our plots were considerably larger than in the cited studies (200–1000 m2 compared with 50 and 12 m2). Our relatively large plots make it possible to evaluate many species. However, we cannot dismiss the possibility that commercial slash harvesting might have negative effects on rare species not included, or occurring at low frequencies, in this study.

management applications

We did not find any rare species or species listed in the Swedish Red Data book (Gärdenfors 2000) in our clear-cut study plots. Rare forest mosses and liverworts are typically confined to old forest stands characterized by high heterogeneity, humid microhabitats and a large volume of broad-leaved trees and/or CWD (Gustafsson, Hylander & Jacobson 2004). It is unlikely that their future depends on slash harvest in clear-cuts. However, slash harvests during later stages of succession (e.g. the thinning stages) may constitute a substantial threat to some rare or Red-listed bryophyte species.

Until further research has been conducted on the long-term landscape scale consequences of slash harvesting on bryophytes, we recommend that management practices should include retention of clustered trees with intact undergrowth and protection and creation of CWD during clear-cutting (cf. Fries et al. 1997; Larsson & Danell 2001). Groups of trees and intact undergrowth will facilitate survival of moisture-dependent species (cf. Hazell & Gustafsson 1999), which subsequently may colonize the rest of the stand. In contrast to CWD, few species are specifically tied to fine woody substrates such as slash (Kruys & Jonsson 1999). It is therefore likely that protection and creation of CWD may mitigate the loss of slash. Unfortunately, commercial slash harvests may also result in loss of CWD (Rudolphi, Gustafsson & Weslien 2005).


We are grateful to the forest company Sveaskog for providing the database of conventionally harvested and slash-harvested clear-cut stands, and for providing field sites. We thank Patrik Blomberg and Sandra Karlsson for valuable assistance in the field. We also thank James Helfield and three anonymous referees for valuable comments on the manuscript. We are grateful to Carl-Gustaf and Viveka Åkerhielm for providing lodging at Dyltabruk. The study was supported by grants from the Swedish Energy Agency (to C. Nilsson).