• Arion subfuscus;
  • carabids;
  • natural regeneration;
  • post-dispersal seed predation;
  • Pterostichus spp;
  • seedling mortality


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1. Forest management involving live tree retention and natural regeneration after cutting is currently increasing in boreal areas. This calls for further analysis of the optimal conditions for seedling establishment from seed following stand disturbances due to logging.

2. We studied post-dispersal predation on seeds and juvenile seedlings of Pinus sylvestris over 3 years in 32 north Swedish boreal forest stands with different levels of stand disturbance by logging. The aims were to identify the most important predator species and to quantify the damage inflicted upon seeds and seedlings in relation to disturbance.

3. In most stands and years, seed predation resulted in < 20% seed mortality, although occasionally it reached 60%. Predation on juvenile seedlings ranged from 5% to 100%, with > 70% mortality in 10 cases and < 30% in 44 cases (of a total of 79 observations).

4. The most important seed predators were the carabids Pterostichus oblongopunctatus and Calathus micropterus, and seed predation was correlated with the number of seed-eating carabids caught in pitfall traps. Microtine rodents caused high damage levels only on a single occasion.

5. Logging affected both catches of carabids and seed predation levels, but the relationship between tree stand density and predation was not linear and, generally, seed predation decreased in the order shelterwood > unlogged forest > clear-cut.

6. The most important predator on juvenile seedlings was the slug Arion subfuscus, which attacked seedlings during the first weeks after germination. Pitfall trap catches of Hylobius abietis, which commonly damage planted (1–3-year-old) conifer seedlings, were not related to the levels of seedling predation.

7. Seedling predation was negatively related to stand disturbance, with the highest predation levels by slugs in unlogged forests and the lowest in clear-cuts. Seedling predation was higher in wet than in dry summers, probably because slugs are moisture-limited.

8. There was a large between-site variation in both seed and seedling predation, but predation was not strongly related to forest site types. The fact that predation was strongly affected by logging operations indicates that there may be opportunities to reduce damage through modification of the silvicultural practices.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

In European boreal forest, post-dispersal predation on conifer seeds and juvenile seedlings may influence the success of seedling establishment (Vaartaja 1950; McVean 1961; Kielland-Lund 1963). Ground-feeding generalists such as carabids, microtine rodents, birds and slugs have been identified as consumers of seeds and seedlings (Forsslund 1944; Vaartaja 1954; Heikkilä 1977; Nilsen 1986). Despite its potential importance, there is a general lack of knowledge concerning the levels of predation in boreal forest and the factors regulating it. The results from earlier studies are variable but studies in other ecosystems indicate that a large variation in time and space is characteristic of post-dispersal predation on seeds and seedlings. Several predator taxa may be involved simultaneously, with different habitat and food preferences, and predator population sizes will vary between years (Willson & Whelan 1990; Myster & Pickett 1993; Hulme 1994; Edwards & Crawley 1999).

In recent years, concern for biodiversity has prompted forest owners throughout the boreal region to change their afforestation practices (Angelstam, Majewski & Bondrup-Nielsen 1995; Riley 1995). A general trend is to retain live trees and to encourage seeding or natural regeneration instead of clear-felling and planting (Fries et al. 1997). Trees retained after harvesting can facilitate regeneration in two ways: by sheltering the ground and by providing seeds for restocking (Hagner 1962; Ottosson Löfvenius 1993; Gray & Spies 1997). However, the structural changes generated by tree harvesting are also likely to influence the size and activity of the animal populations, and seed and seedling predation could therefore be affected by the density and structure of the residual tree stand. In the fauna of the boreal coniferous forest some common carabids are known to respond positively to logging (Niemelä, Langor & Spence 1993). Other animals, e.g. slugs, are favoured by humid conditions (Nystrand & Granström 1997a) and may therefore be negatively affected by removal of the tree overstorey. Due to these relationships, we would expect a complex effect of logging on seed and seedling predation levels in boreal forest.

In this study, we quantified the level of post-dispersal predation on Pinus sylvestris L. seeds and juvenile seedlings in forest vegetation, separating the effect of different animal groups as far as possible. To determine if predation is affected by the degree of tree retention following logging, we included three types of management: clear-cut stands, shelterwood stands and unlogged older forests. We accounted for the spatial and temporal variation by including several sites of different forest site types (Cajander 1926), and by quantifying predation over 3 consecutive years. To support the data on predation, we also trapped invertebrate seed eaters in some of the study sites.

In earlier studies, predation levels have been estimated indirectly by observing seedling emergence in densely seeded spots with different types of predator exclusion. In these cases, predation may be confused with other factors affecting seed and seedling survival. There is also a potential bias due to the locally elevated seed densities that may attract (or satiate) predators. To avoid this, we monitored individual seeds and seedlings deployed at very low densities. Damage due to predation was assessed through analysis of wounds on seed coats and seedlings. The wounds were classified based on laboratory studies of several animal species. This procedure gives a direct estimate of the predation level, while local seed and seedling densities remain practically unchanged.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study sites

Experimental region

The experiments were performed in boreal coniferous forests located within 50 km of Umeå in northern Sweden (Fig. 1). The tree stands in these forests consist mainly of Scots pine P. sylvestris and Norway spruce Picea abies (L.) Karst. in various proportions, with some broad-leaved species (typically birches Betula L.) intermixed (Table 1). Conifer seed dispersal occurs during late winter and spring (Heikinheimo 1937). None of the tree species has a persistent seed bank and most seeds usually germinate between late June and early August. Seedling growth ceases in late August.


Figure 1. Location of the experimental sites in northern Sweden. Site numbering corresponds to Table 1.

Download figure to PowerPoint

Table 1.  Forest site types and stand characteristics for the experimental sites. The first five sites were on moist soils with thin peat layers, covered mainly by Sphagnum spp. and Polytrichum spp. mosses. For these sites, site type classification follows Eurola, Hicks & Kaakinen (1984). Classification of the remaining sites follows Cajander (1926). Sites 6–10 had mesic soils with well-developed mor layers, the moss layer dominated by Pleurozium schreberi (Brid.) Mitt. and Hylocomium splendens (Hedw.) B.S.G. feather mosses. The last four sites were on xeric, sandy soils with thin mor layers, covered mainly by Cladina spp. lichens. Light transmission is the proportion of light 1 m above ground in the stand, relative to a large open field. A blank space indicates that the stand treatment was not included at the particular site
Light transmission (%)
Site no.Forest type*Tree species composition (P.s./P.a./Bet.)UnloggedShelterwoodClear-cut
  • *

    Vmsm, Vaccinium myrtillus spruce mire; Tshf, thin-peated spruce heath forest; Tpf, thin-peated pine forests; MT, Myrtillus type; VT, Vaccinium type; MClT, MyrtillusCladina type; ClT, Cladina type.

  • Tree species composition is shown as the proportions (in tenths) of the basal area of the three dominant species (P.s., Pinus sylvestris; P.a., Picea abies; Bet., Betula spp.).

  • Not recorded.

Moist soils1Vmsm1/9/01462 
5Tpf7/3/0–  100
Mesic soils6MT3/6/111 100
Xeric soils11MClT8/2/02674 
12MClT9/1/038 100
Average   225698
Sites, stands and management types

In June 1993, we selected 14 forest sites for experiments. Parts of the area at each site had been logged during the previous winter so that each site contained a set of differently managed stands. The stands were of three management types: clear-cuts, shelterwoods and unlogged forests. All sites included an unlogged forest stand and one or both types of logged stands. Four sites contained all three management types, seven sites contained shelterwood and unlogged forest, and three sites contained clear-cut and unlogged forest. Thus, a total of 32 stands were used for the study.

Stand size was at least 5 ha and for the experiments we selected a 2-ha trial area in each of the stands used at each site. Figure 2 shows examples of an unlogged forest and a shelterwood. The management types were defined as follows: in clear-cut stands, < 10 trees ha−1 were retained after logging; in shelterwoods, 150–300 trees ha−1 were retained, most of which had been dominant trees in the original stand, the tree species composition was not altered by logging; unlogged forests contained > 400 trees ha−1, with an average age of at least 90 years, and no logging operations had taken place for the past 15 years.


Figure 2. Typical example of (a) an unlogged forest stand and (b) a shelterwood stand used for the study. The pictures were taken in the central part of the stands at site 8 in June 1995.

Download figure to PowerPoint

During logging, harvesters and forwarders were used for felling and transport, and the logging slash was left dispersed on the ground. After the logging, no silvicultural operations (such as mechanical scarification, seeding or planting) were performed during the trial period.

Detailed investigations were carried out to ensure that all stands within a site had the same forest characteristics before the logging operations. Forest inventory data, aerial photographs and personal information from the forest managers were used. Field observations were made of the cover and species composition of the tree-, field-, and bottom-layers, tree age (ring counts on stumps in the logged stands) and the soil moisture conditions. It was confirmed that the stands within each site had been similar prior to logging.

The predation experiments were performed during 1993–95. As two of the sites had to be abandoned before the experiments were completed, some observations are lacking for those sites. At site 5 the clear-cut was mechanically scarified by the land manager in autumn 1993, and at site 14 the shelterwood was further logged during winter 1995.

Between-site variation

The sites were selected to cover an array of forest site types. Forest site types and proportions of the dominant tree species are shown in Table 1. The site type classification system used here is based mainly on vegetation composition but indicates soil moisture and site productivity (Cajander 1926; Eurola, Hicks & Kaakinen 1984). Five sites were on moist soils, five were mesic and four had xeric soils. Soil fertility was less variable and was generally low to moderate. However, sites 4, 6, 7 and 8 were slightly more productive than the other sites, as shown by slightly higher tree growth rates and higher species diversity in the field layer vegetation.

The transmission of photosynthetically active light through the tree canopies was measured on cloudy days in 1995 (Table 1). The readings were taken approximately 1 m above the forest floor, with at least 10 points within each stand, and with parallel readings in a nearby large open area. Light transmission was calculated as the proportion of light in the stand relative to the open area.

Weather conditions and seed production

For each year, summer weather data (total precipitation and daily mean temperature from May 1 to July 31) were obtained from two meteorological stations in the region (Vindeln and Umeå; Fig. 1). The summer of 1993 was cool and rainy (total precipitation 299 mm, mean temperature 14·1 °C), while 1994 and 1995 were warmer and drier (precipitation 107 mm and 136 mm, mean temperatures 17·6 °C and 17·7 °C, respectively).

We made ocular estimates of cone numbers at the sites in early spring each year. In 1993, there were moderate to high numbers of P. abies cones but very few of P. sylvestris, while in 1994–95 cone numbers were very low for both species. Thus there was a rather intense conifer seed rain at spruce-dominated sites (sites 1–4 and 6–9) in 1993, but almost no conifer seeds were shed in 1994 and 1995.

Seed predation in the field

To facilitate observation of predation rates on P. sylvestris seeds on the forest floor, single seeds were glued to the ends of 25-cm long pieces of thin black polyester thread. The glue was cellulose acetate dissolved in acetone, which leaves a clear hard film after drying. The seeds came from P. sylvestris stands in the region and were obtained from a commercial supplier. The seeds had 96% germinability, weighing 4·61 g per 1000 seeds. Only a small drop of glue was applied (always to the hilum side) in order to minimize the possible effects on seed germination and palatability. In a greenhouse trial, tethering did not affect germination rate or total germination percentage (O. Nystrand & A. Granström, unpublished data).

The threads were attached to 15-cm tall plastic sticks, which were then placed out in rows with 2 m between the sticks. The seeds were dropped on the forest floor but if a seed became suspended by the vegetation it was moved to a more natural position of rest. In 1993 experiments started in early July (20 seeds placed out in each stand), whereas in 1994 and 1995 experiments started in late May, soon after snowmelt (30 seeds in each stand). In the autumn of each year, we recovered the seeds by carefully following the thread from the stick to the seed. Seeds were examined in the field using a 10× pocket lens. If a seed had been killed by predators, we studied the marks on the seed coat to determine which predator was responsible. In doubtful cases, seed coats were brought to the laboratory and analysed under a 40× stereoscope. If a seed had germinated and released the seed coat, the immediate surroundings were carefully examined for living or dead seedlings likely to have emerged from the deployed seed. The seed coat wounds were classified based on laboratory experiments with several animal species (see Laboratory studies below; for pictures of P. sylvestris seeds damaged by insects, birds and voles, see Heikkilä 1977).

Seedling predation in the field

For the study of predation on newly germinated seedlings, seeds (from the same sample as in the seed predation experiment) were sown individually in small peat-filled open-ended paper cylinders, 5 cm tall and 1 cm wide (Paperpot; Nipponbeet Sugar Mfg. Co. Ltd, Japan). Most seeds opened after 4–5 days, and 8–10 days after sowing the containers were placed out at the experimental sites. The seedlings were then approximately 30 mm tall and about to unfold their cotyledons. A hand-held rod was sunk into the soil to prepare a hole into which the small seedling container was fitted. The terminal bud of the seedling was then located close to the uppermost part of the loose moss layer, which is the typical position of naturally emerged seedlings. The use of containers did not affect the predation level, as shown in a previous field experiment where seedling predation was quantified for containerized and uncontainerized transplanted seedlings (Nystrand & Granström 1997a).

The containers were placed in rows with 1 m between individual seedlings. In 1993, 50 seedlings were placed out in each stand, and in 1994 and 1995, 75 seedlings were planted out. The experiment was started in July each year, when germination of conifer seeds normally takes place in northern Scandinavian forests (Yli-Vakkuri 1961; Nystrand & Granström 1997a). In 1994, we also planted seedlings in late May in a random subset of sites (site nos 3, 6, 7, 13 and 14).

During remeasurements we recorded all damage and signs of specific predators. In most cases the seedlings were revisited on three occasions and the last remeasurement was always done approximately 60 days after deployment. In the later remeasurements of 1994, many seedlings suffered from drought but causes of mortality could still be determined. In 1995 all seedlings were first inspected after approximately 12 days. Thereafter, a dry-spell followed by heavy rainfall destroyed many seedlings, making further observations difficult and unreliable. For 1995 we therefore present data for the 12-day observation period only. Practical problems in obtaining seedlings for deployment forced the exclusion of sites 2 and 9 in 1994, and sites 3 and 4 in 1995.

Seed predator fauna

At nine randomly chosen sites we screened the fauna of ground-moving invertebrates using a set of 10 pitfall traps in each stand. The traps were 20 cm deep and had square 10-cm wide openings. They were dug into the soil so that the rim was flush with the top of the humus layer, and were filled with 5 cm of ethylene glycol : water (1 : 1; for preservation), bittering agent (to discourage vertebrates from drinking the solution) and a few drops of detergent (to reduce surface tension). To protect from rain, the traps were covered with a 20 × 20-cm hardboard roof, supported 5 cm above the traps by wooden sticks. Within each stand, the traps were placed in a row with approximately 5 m in between. The traps were emptied at least twice during each summer, in most cases three or more times. In the laboratory all Coleoptera and molluscs were determined to species level. To determine if the seed predation level was dependent on the activity of seed-eating carabids we compared trap catches of seed-eating carabids with carabid seed predation data.

Pitfall trap data provide estimates of the ‘activity density’ (discussion and references in Thiele 1977) of ground-moving arthropod species in the habitat, i.e. the trap catches of a species will depend not only on its absolute population density but also on its mobility and degree of activity (Baars 1979; Topping & Sunderland 1992). For molluscs, trappings only serves to detect the presence of a species. Although molluscs are frequently caught in pitfall traps, data are probably biased because molluscs are able to climb on the trap walls.

Laboratory studies

To obtain a reference material for the classification of seed coat wounds, and to study the behaviour of animals when encountering tethered seeds, we performed several laboratory experiments with captured animals. The experiments were carried out as follows:


Six Clethrionomys glareolus Schreber were trapped in coniferous forests near Umeå, and kept singly in 1-m2 experimental arenas. Tethered and untethered P. sylvestris seeds (five of each) were placed at predetermined random locations on 0·25-m2 sheets of forest substrate. The animals had access to the sheets for approximately 30 min, during which time we recorded their activity with a remote-controlled video camera placed above the arenas. Subsequently, we searched the sheets for seeds and seed remains. If untethered seeds were missing but no remains were found, we used the video recordings to determine if the seeds had been eaten. Each animal was presented with two types of substrate: (i) intact forest vegetation with dense Pleurozium feather moss, and (ii) humus surface (moss layer removed).


Individuals of Pterostichus oblongopunctatus F. (n = 3) were allowed to feed on three tethered and three untethered seeds placed at random locations on humus surfaces. The animals were kept singly in experimental arenas, consisting of 20 × 20-cm plastic boxes with a small shelter near the centre of each box. The arenas were kept under constant observation during the first few hours and, thereafter, further damage to seeds was recorded daily for 3 days. Three individuals of Calathus micropterus Duft. were tested in the same way, except seeds were allowed to germinate (radicle protruding approximately 10 mm) before the start of the experiment. C. micropterus were also offered ungerminated seeds.


Two species of finches were tested. Nine Fringilla coelebs L. and five F. montifringilla L. were presented with tethered and untethered seeds (n = 10 of each category) placed randomly on a mineral soil surface arranged in a 60 × 40 cm tray. The birds were kept singly in large aviaries and allowed to feed in the trays for 30 s. Afterwards, the trays were removed and searched for seed remains. Experimental procedures followed Nystrand & Granström (1997b).

Statistical analyses

We used anova to detect differences in seed predation (proportion of seeds killed) depending on the stand treatments. The experiment had a randomized incomplete block design (Montgomery 1991). Because the seeds were deployed at new spots within the stands in each year, and because the set of stands was not exactly the same in all years, we made a separate analysis for each year instead of performing a repeated measures anova. Diagnostic tests on the residuals showed no marked deviations from the assumptions of normality, constancy or independence. Arcsine-transformation, which is often used for proportional data, did not change any conclusions drawn from the statistical tests, and here we present results for the untransformed data. Within each year, post-hoc tests were made according to the sequentially rejective Bonferroni procedure (Holm 1979). All analyses and calculations were made in systat (SPSS Inc. 1996)

To study the between-site differences, we standardized the predation levels over each treatment and year. The standardized scores have a mean of zero and a standard deviation of one, and were calculated by subtracting the mean (for each treatment and year) from each value and dividing the difference by the standard deviation (z-score; SPSS Inc. 1996).

For seedling predation, the anova, multiple tests and standardized scores were calculated as described for seed predation. The last remeasurement in each year was used, and for 1994 only the July deployment was included.

The relationship between the estimates of seed and seedling predation was tested for each year with Pearson product-moment correlation. Student's t-test was used to determine if the correlation coefficients deviated significantly from zero.

To compare trap catches of seed-eating carabids and carabid seed predation data, we used logarithmic regression with seed predation as the dependent variable. Zero values (no seed-eating carabid caught) were set to one to allow logarithmic transformation (four cases in 1993, two in 1994, and one in 1995). In the laboratory studies of voles and finches, the proportion of tethered and control seeds killed by the animals was compared using t-tests for paired samples.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Laboratory studies


When seeds were presented on a bare humus surface, C. glareolus consumed similar amounts of tethered seeds and control seeds (average seed retrieval 73% vs. 63%; t = 1·46, d.f. = 5, P = 0·20). However, when seeds were hidden in a dense moss layer, the voles consumed more seeds with threads than without (33% vs. 7%; t = 3·16, d.f. = 5, P = 0·025), indicating that observations on tethered seeds in dense vegetation may overestimate seed predation due to voles.


Pterostichus oblongopunctatus readily attacked and consumed seeds, while C. micropterus only managed to consume seeds that had already opened for germination. However, we never observed C. micropterus feeding on fully emerged seedlings, where the seed had been released from the cotyledons. For both carabid species, the location of seeds in the arena seemed more important than whether or not the seeds were tethered. Generally, seeds located close to the shelter were taken first, while those placed farther away were detected later. In no case did the tethering prevent an animal from consuming a seed. We did not observe carabids taking interest in the threads, and threads never seemed to guide beetles towards the seeds.


Fringilla coelebs consumed marginally fewer tethered seeds compared with untethered seeds (35% vs. 56%; t = −2·18, d.f. = 8, P = 0·061). There was no difference between tethered and untethered seeds for F. montifringilla (60% vs. 63%; t = −0·41, d.f. = 4, P = 0·70). On a few occasions, F. coelebs discarded tethered seeds after attempting to pull them off the thread. This behaviour was not seen in F. montifringilla. Thus, tethered seeds may yield underestimations of predation by F. coelebs.

Seed predation in the field

Damage types

Based on the laboratory experiments and information from the literature, we classified damage to the tethered seeds, and to seedlings emerging from them, into four groups according to the shape of the seed coat wounds and other signs of predation: (i) seed predation by voles – bite-marks with clean edges, rounded shape, usually some cracks in the seed coat; (ii) seed predation by carabids – seed coats with irregularly shaped wounds, serrated edges and no cracks; (iii) seed predation by fringillid birds – seed coats split longitudinally in two halves; (iv) post-germination mortality – deceased after radicle protrusion, no wounds on the seed coat.

Bird predation was never seen at the field sites, and hereafter ‘seed predation’ means damage made by voles and carabids. When carabids attacked newly germinated seeds, we classified this as seed predation, as the carabids targeted the megagametophyte rather than the seedling (see Laboratory studies). Of the seeds attacked by carabids, 17% had opened for germination before they were killed (43%, 22%, and 6·9% in 1993, 1994 and 1995, respectively).

In most (72%) cases of post-germination seed mortality, the cause of death could not be established because no remains of seedlings were found. It was difficult to find dead seedlings among moss shoots and litter, and missing seedlings could not be safely considered as eaten. However, there was a strong correlation between mortality of seedlings emerging from the tethered seeds and predation levels on the containerized seedlings (Fig. 3). This indicates that predators were responsible for most cases of post-germination mortality of tethered seeds.


Figure 3. Predation on deployed juvenile Pinus sylvestris seedlings, plotted against post-germination mortality of deployed tethered seeds (wound type 4, see Damage types). Each data point represents a stand, and the plot symbols indicate the stand treatments: squares, clear-cut; triangles, shelterwood; circles, unlogged. Data are for 1993, and include all stands where 10 or more of the tethered seeds germinated. In 1994 and 1995, too few tethered seeds germinated to make the comparison possible. There was a strong correlation between the variables (Pearson r = 0·68, n = 28, P < 0·001).

Download figure to PowerPoint

Most seeds (90%) could be retrieved during the inspections. However, the seeds were fixed with only a small amount of glue, and occasionally they detached. Because the lost seeds were probably not always taken by seed eaters, we chose not to treat them as eaten. Consequently, the seed predation measurement was slightly conservative.

Predation levels

Seed predation typically resulted in < 20% seed mortality, although reaching higher levels (up to 60%) on some occasions. Most seed predation (81% of all wounded seeds) was due to carabids. Vole predation was low, except for one case (the unlogged forest at site 11 in 1994) where 57% of the seeds were eaten by voles. The stand treatments had a significant effect on seed predation in 1993, with decreasing predation levels in the order shelterwood > unlogged forest > clear-cut (Table 2 and Fig. 4a). This tendency was persistent over the years, although the differences were not significant in 1994 and 1995 (Fig. 4a).

Table 2.  Analyses of variance for the effect of stand treatment on seed predation
YearSourced.f.Adjusted MSFP

Figure 4. (a) Predation on Pinus sylvestris seeds deployed in experimental sites during 3 years following different stand treatments. Cut, clear-felled stand; Shlt, shelterwood stand; Unlo, unlogged stand. The bars show average predation by microtine rodents (filled part) and carabids (unfilled part). The error bars indicate ± 1 SEM for the combined carabid and vole predation. Multiple tests were made among treatments within each year, and bars with the same letter were not significantly different at P < 0·05. (b) Predation on juvenile P. sylvestris seedlings deployed in the same experimental sites as above. The slug Arion subfuscus was the dominant predator. Error bars and multiple tests as in (a).

Download figure to PowerPoint

There was considerable between-site variation. Predation differed significantly between sites in 1995, and marginally in 1994 (Table 2). Sites did not differ in 1993, but predation levels were then generally low. In addition, some sites showed consistently higher predation levels than the average for all years and treatments (sites 5 and 7), while other sites (sites 8 and 12) had consistently lower than average levels (Fig. 5).


Figure 5. Standardized values of seed predation on Pinus sylvestris. The data values were standardized over each treatment and year, to have zero mean and standard deviation one (see Statistical analyses). Treatments: Cut, clear-felled stand; Shlt, shelterwood stand; Unlo, unlogged stand. Circles, 1993; squares, 1994; triangles, 1995. Lines connect values for single years, but note that values are relative to the other sites and do not show trends in absolute numbers.

Download figure to PowerPoint

We found no relationship between forest site type and seed predation. The site with the highest overall predation level (site 7) belonged to the Myrtillus forest site type (Table 1), but the other three Myrtillus sites (sites 6, 8 and 9) had low or moderate predation levels (Fig. 5). Also, when grouping the sites according to forest site types from moist to xeric, the standardized predation levels varied greatly between sites within each group (Fig. 5).

Seedling predation in the field

Damage types

In the deployed seedlings that were attacked, the appearance of the wounds varied from complete removal of all parts of the seedling, to hardly visible bite marks on the cotyledons or hypocotyl. We have found that seedlings often survive minor damage, e.g. removal of a cotyledon (O. Nystrand and A. Granström, unpublished data). Therefore, the term seedling predation refers only to seedlings that had suffered severe damage to the hypocotyl or apical meristem. Of all the attacked seedlings, 8% were considered injured but capable of survival, and of the killed seedlings 37% had been consumed entirely. Among the animals known to feed on conifer seedlings (Forsslund 1944; Vaartaja 1954), the only species we encountered on or near seedlings was the slug Arion subfuscus (Drap.). Slugs, or traces of them (slime tracks or faeces), were observed on, or in the immediate surroundings of, 8% of the damaged seedlings. However, traces of slugs are not persistent, so the absence of such traces does not dismiss slugs as responsible for the predation. Most wounds on seedlings conformed to those seen in laboratory experiments with A. subfuscus (Nystrand 1998).

Predation levels

The degree of stand disturbance affected seedling predation, predation levels decreasing in the order unlogged forest > shelterwood > clear-cut (Fig. 4b). The difference was significant between all three treatments in 1993, and between logged and unlogged stands in 1995 (Table 3 and Fig. 4b). Clear-felled stands had consistently low predation levels, while predation in the other stand types varied between years. The range in predation levels (for single stands and years) was 10–50% in clear-felled stands, 5–95% in shelterwoods, and 40–100% in unlogged forests. We found predation levels exceeding 70% in 10 cases (out of 79 observations of single stands and years), all of which were in shelterwoods and unlogged stands in 1993. Levels lower than 30% were found in 44 cases, mainly in 1994 and 1995.

Table 3.  Analyses of variance for the effect of stand treatment on seedling predation
YearSourced.f.Adjusted MSFP
Treatment2726733·5< 0·001

There was no strong relationship between forest site type and seedling predation level, although two of the four xeric sites (sites 12 and 13) had consistently lower than average predation levels (Fig. 6). Two of the four mesic sites of the Myrtillus type (sites 6 and 8) had consistently high scores, while the scores for sites on moist soils varied greatly between treatments and years (Fig. 6).


Figure 6. Standardized values of seedling predation on Pinus sylvestris. The data values were standardized over each treatment and year, to have zero mean and standard deviation one (see Statistical analyses). Treatments: Cut, clear-felled stand; Shlt, shelterwood stand; Unlo, unlogged stand. Circles, 1993; squares, 1994; triangles, 1995. Lines connect values for single years, but note that values are relative to the other sites and do not show trends in absolute numbers.

Download figure to PowerPoint

The repeated inspections showed that most predation occurred shortly after deployment, regardless of treatment or deployment occasion. On average, 87% of the predation occurred within the first month. The time of deployment (late May 1994 vs. July 1994) did not affect predation levels. The average predation levels were 15·1% and 15·4% for May and July, respectively (n = 12 for both).

Seed and seedling predation levels were not related to each other. Correlation coefficients between the estimates of predation on tethered seeds and containerized seedlings were low and were not significant for any of the years (1993: r = −0·090, n = 32, t = −0·496, P = 0·62; 1994: r = −0·107, n = 24, t = 0·503, P = 0·62; 1995: r = 0·335, n = 23, t = 1·630, P = 0·12). The ‘total’ predation level, calculated as the product of predation losses at both seed and seedling stages (e.g. 50% seed predation and 50% seedling predation would yield a total loss of 75%), exceeded 70% in 16 out of 79 observations, 12 of which were in 1993. The trend among treatments was the same as with seedling predation alone, predation decreasing in the order unlogged > shelterwood > clear-cut.

Seed and seedling predator fauna

Of carabid species, P. oblongopunctatus and C. micropterus were the most abundant in the pitfall traps (Table 4), and were encountered in most stands. In addition, Pterostichus adstrictus Eschscholtz were caught in high numbers in two of the stands. These carabids are generalist feeders, and are able to consume conifer seeds (Heikkilä 1977). Some species in the genera Amara Bonelli, Harpalus Latreille and Agonum Bonelli also attack conifer seeds (O. Nystrand & A. Granström, unpublished data), but they were only encountered occasionally in the traps (Table 4). Most carabids were caught in spring–early summer.

Table 4.  Sum of pitfall trap catches of some seed- and seedling-eating insects and molluscs at nine forest sites, each site containing an older unlogged forest (unlogged, n = 9) and one or two recently logged stands (shelterwood, n = 7; clear-cut, n = 3). There were 10 traps in each stand. The traps were operated from July 1993 to September 1995, except for two of the sites where data were collected for only 1 and 2 years, respectively. ‘Proportion (%) of stands with occurrence’ indicates the proportion of sites where at least one individual was caught
Specimens per 10 trapsProportion (%) of stands with occurrence
Pterostichus oblongopunctatusCarabidae1222711383371100
Calathus micropterusCarabidae6745287810067
Pterostichus adstrictusCarabidae0·2328225767
Agonum mannerheimi DejeanCarabidae0·49011140
Amara spp.Carabidae0·2111222967
C. melanocephalus L.Carabidae0·20·60·7112933
Harpalus quadripunctatus DejeanCarabidae000·70033
Sum of seed-eating carabids 19033020689100100
Otiorrhyncus spp.Curculionidae346565767
Hylobius abietisCurculionidae2611744100100100
Arion subfuscusArionidae578942100100100

The number of trapped seed-eating carabids was positively related to the level of carabid seed predation (Fig. 7). The relationship was statistically significant in 1993 and 1995, and marginally in 1994.


Figure 7. Predation by carabids on deployed Pinus sylvestris seeds plotted against the number of seed-eating carabids caught in pitfall traps in (a) 1993, (b) 1994, (c) 1995. Each data point represents a stand, and the plot symbols indicate the stand treatments: squares, clear-cut; triangles, shelterwood; circles, unlogged. The experimental period ranged from May to September, except for 1993 (July to September). The lines represent logarithmic regressions for all stands within each year: for 1993, R2 = 0·30, n = 19, P = 0·015; for 1994, R2 = 0·20, n = 19, P = 0·054; for 1995, R2 = 0·50, n = 15, P = 0·003.

Download figure to PowerPoint

There was large between-site variation, total trappings of seed-eating carabids ranging from 1 to 3088 individuals (Table 5). Five sites had low numbers (< 100), including all three sites with xeric soils and lichen-dominated vegetation (sites 11, 13 and 14; Table 5). Site 7, which had by far the highest carabid density, was not considerably different from the other mesic sites with respect to vegetation type, tree species composition or light transmission (Table 1).

Table 5.  Pitfall trap catches of the three dominant seed-eating carabids at nine forest sites, each site containing 10 traps. The data shown are the number of individuals caught from July 1993 to September 1995, except for site 5, where data are for 1993 only, and site 14, where data are for 1993–94
1Calathus micropterus166101 
Pterostichus oblongopunctatus72299 
Pterostichus adstrictus013 
3C. micropterus443 
P. oblongopunctatus109187 
5C. micropterus1 0
P. oblongopunctatus0 1
6C. micropterus107 46
P. oblongopunctatus21 61
P. adstrictus1 47
7C. micropterus23713537
P. oblongopunctatus8811408353
P. adstrictus1036
8C. micropterus2062 
P. oblongopunctatus10 
P. adstrictus02 
11C. micropterus2412 
P. oblongopunctatus183 
P. adstrictus05 
13C. micropterus010
14C. micropterus011
P. oblongopunctatus032
P. adstrictus011

Pterostichus oblongopunctatus was more abundant in shelterwood stands than in clear-cuts and unlogged forests (Table 4). For clear-cuts, the results were inconclusive: at site 6, trappings were higher in the clear-felled stand than in unlogged forest, while at site 7, there was the opposite trend (Table 5). Calathus micropterus was the dominant carabid in most unlogged forests (Tables 4 and 5). This species nearly always had lower numbers in the logged stands. In contrast, P. adstrictus was found almost exclusively in logged stands, although never reaching very high numbers (Table 5). In most cases, it appeared 1 or 2 years after the logging event.

Of the animals known or suspected to feed on juvenile conifer seedlings, the slug A. subfuscus was generally the most common, and was trapped in all stands. Other molluscs were rarely caught. The weevils Otiorrhynchus dubius Strom. and O. scaber L. occurred in small numbers at most of the sites, and Hylobius abietis L. reached high numbers in some of the logged stands (Table 4).


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Predator fauna

The combined evidence from the analysis of seed-coat wounds, laboratory tests and pitfall trappings shows that seed predation in the studied forests was due mainly to the two carabid species P. oblongopunctatus and C. micropterus. These are the two most common and widespread carabids in the forests of northern Europe (Lindroth 1985; Niemelä, Tukia & Halme 1994). P. oblongopunctatus is probably the more important seed predator of the two. Seeds are available to C. micropterus only for a very short period immediately following germination, because it cannot penetrate ungerminated seeds.

Small mammals only rarely consumed large amounts of seeds, but they may be locally and periodically important seed consumers as their populations are largely cyclic and unevenly distributed (Myllymäki 1977; Hansson 1980; Hörnfeldt 1994). Voles Microtinae are the only seed-eating ground-feeding small mammals that are widespread in the north European boreal forest. Of these, C. glareolus is considered the most important seed eater (Hansson 1980). Its population was very low in the region in 1993, increased in 1994, and peaked in 1995, although they were still at a low level when seen over a 20-year study period (B. Hörnfeldt, personal communication). In the single incident of intense seed predation by voles (57% at site 11 in 1994), the deployed seeds were probably within the range of a locally strong vole population.

We observed no seed predation by birds, although some of the passerines that are common in the region are known to feed on conifer seeds on the ground (Vaartaja 1950; Heikkilä 1977). However, in laboratory experiments with two Fringilla species (Nystrand & Granström 1997b), seed detection was restricted in undisturbed moss vegetation, and it is possible that they avoid feeding in such habitats. Birds may be important feeders in situations when seeds are more visible, e.g. on exposed mineral soil or freshly burnt surfaces (Bergsten 1985; Nystrand 1998).

Of the animals known or suspected to eat juvenile conifer seedlings, no species except the slug A. subfuscus seems to have been of importance. This slug is common throughout Europe (Kerney, Cameron & Jungbluth 1983). It tolerates the acidic conditions in conifer litter better than other molluscs (Waldén 1981) and is often the only slug species encountered in European coniferous forests.

Identifying seedling predators from the remains of attacked seedlings is often more difficult than identifying seed predators from seed coat wounds, as wound characteristics are not well defined for different seedling predators. Besides slugs, Otiorrhyncus weevils may have consumed seedlings (Forsslund 1944) but they were scarce in the traps and were never seen near seedlings. The weevil Hylobius abietis has occasionally eaten juvenile P. sylvestris seedlings in the laboratory (O. Nystrand, unpublished data). This species enters forest sites following stand disturbances and may cause considerable damage to planted (1–3-year-old) conifer seedlings (Örlander, Nilsson & Nordlander 1997). In our study, seedling predation was mostly low in stands with large Hylobius populations, and we believe that Hylobius does not normally feed on juvenile conifer seedlings in the field.

Predation levels, predator populations and variation

This study is the first to quantify predation on conifer seeds and juvenile seedlings in a large number of boreal forest stands over consecutive years. Predation was frequently severe, but there was no correlation between seed and seedling predation levels, suggesting that the seed- and seedling-eating faunas are not controlled by the same factors and therefore vary independently of each other.

Seed predation was positively related to the number of trapped seed-eating carabids, indicating that the seed predation level depends directly on the size and activity of the carabid population. Also, carabid activity was highest in the spring and early summer, coinciding with the conifer seed dispersal period (Heikinheimo 1937). As carabids were the main seed predators, the site requirements of various species and their response to disturbance should be considered. Both P. oblongopunctatus and C. micropterus are reported to prefer shaded conditions (Lindroth 1985). The pitfall trap and seed predation data presented here show that populations increased or were maintained in shelterwood stands, but decreased following clear-cutting, for several years after the disturbance. A positive relationship between forest site quality variables (fertility, vegetation and litter composition) and carabid abundance has been shown (Niemelä & Spence 1994; Szysko, Vermeulen & Boer 1996), but in our study we found no relationship between forest site types and seed predation level.

A stronger effect of the logging operations was found for seedling predation. In all 3 years there was a positive relationship between seedling predation level and tree cover. We believe this was a consequence of slugs being the dominant predators. Slugs are extremely sensitive to desiccation, and slug activity is therefore promoted by humid and shaded conditions (Crawford-Sidebotham 1972; Nystrand & Granström 1997a). Because of this, seedling predation was also strongly weather-related. The predation levels were high in 1993, when the weather was cool and very rainy, but low in the extremely warm and dry month of July 1994. In 1995, the weather was cool and not as dry as in 1994 and (for the short observation period) predation approached the 1993 levels. Experimental manipulation of the forest floor moisture regime has shown a strong effect on slug attack rate (Nystrand & Granström 1997a) and it is likely that the moisture content of the forest floor was the ultimate controlling factor for both temporal and spatial variation in seedling predation.

Although forest floor moisture conditions affect slug activity, there was no general difference in seedling predation levels between moist, mesic and xeric forest site types. However, this classification mainly reflects the average ground water level, whereas the moisture content of the uppermost part of the forest floor (where A. subfuscus generally moves) is controlled more by weather conditions and tree shading (Samran, Woodard & Rothwell 1995).

The risk of lethal attack was not constant over time for a seedling cohort. Almost all predation occurred during the first weeks after the seedlings had been placed out. In laboratory studies with the slug A. subfuscus, the palatability of P. sylvestris seedlings to slugs decreased with time, and about 1 month after seed germination the seedlings were no longer attacked (Nystrand 1998). The results show that seedling predation occurs mainly during the first weeks following germination, and is negligible on more developed seedlings. It has been shown that for many plant species slugs prefer juvenile seedlings before adult plants (Fenner, Hanley & Lawrence 1999). In boreal forest, several observations have shown decreasing mortality risks for P. sylvestris seedlings after the first year (Yli-Vakkuri 1961; Ohlson & Zackrisson 1992; Nilsson, Steijlen & Zackrisson 1996), and then mainly due to causes other than predation.

Effects on conifer regeneration

High levels of conifer seed and seedling predation can have severe silvicultural consequences. In artificial seeding, higher seed numbers will be needed for acceptable results, increasing costs substantially. Pinus sylvestris seeds are expensive (price range in Scandinavia Euro 800–Euro 1600 (£400–£1000) kg−1) and seed cost has a large influence on the total cost of direct seeding (Wennström, Bergsten & Nilsson 1999). In natural regeneration, the rate of seedling establishment could be reduced, resulting in a longer regeneration period and uneven age distribution in the seedling population. Here, we show that high seed and seedling predation levels (> 70%) occur in north Swedish forests, especially in unlogged forests and dense shelterwoods. Provided that the predation causes a corresponding reduction in seedling establishment, our results suggest that predation will frequently impede the regeneration process following shelterwood cutting. Yet seedling establishment is restricted also by the seed's physical and chemical environment (Bjor 1971; Zackrisson et al. 1997), and it is conceivable that the mortality risk from predation is lower for seeds in favourable spots for germination. If this is the case, even high predation levels can have little effect on establishment (Andersen 1989; Crawley 1992). However, in several experiments in northern coniferous forests, predator exclusion has had a positive effect on conifer seedling recruitment 1–3 years after sowing (Vaartaja 1954; McVean 1961; Kielland-Lund 1963; Nystrand 1998).

In the last decades in Scandinavia, regeneration directly from seeds has mainly been employed in xeric P. sylvestris forests with seed tree stands having less than 100 trees ha−1. Our results show that predation is often low in these situations. Today, sowing and regeneration from seed trees is practised also in other forest site types, and with a larger diversity in the structure and density of the residual tree stands. In this case, our results indicate a considerably increased risk of predator damage to seeds and seedlings. These risks need to be considered in future use and development of shelterwood systems in the northern coniferous forests. By modifying the density of the residual tree stands, it should be possible to balance the effects of seed and seedling predation, seed rain and physical properties, to find optimum conditions for seedling establishment. Actions such as mechanical scarification, prescribed burning and removal of logging slash are also likely to affect predation through effects on predator populations, seed germination rate and seed detectability. These relationships remain to be studied, and recommendations for specific actions against seed and seedling predators must await more detailed analyses of the population dynamics of the predators, their choice of habitats and feeding behaviour.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Birger Hörnfeldt for information on vole population densities. Thanks also to Roger Petterson and Bengt Ehnström, who guided us regarding carabid species determination, to Leif Nilsson for statistical advice, and to Lars Eliasson, Morgan Karlsson, Anna Lundberg, Anna Mårtensson and Martin Persson for technical assistance. We are grateful to MoDo Skog, Skogssällskapet and the private forest managers who gave access to the study sites. The study was supported by the Swedish Council for Forestry and Agricultural Research and by WWF of Sweden.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Andersen, A.N. 1989) How important is seed predation to recruitment in stable populations of long-lived perennials? Oecologia, 81, 310315.
  • 2
    Angelstam, P., Majewski, P. & Bondrup-Nielsen, S. 1995) West–east cooperation in Europe for sustainable boreal forests. Water, Air, and Soil Pollution, 82, 311.
  • 3
    Baars, M.A. 1979) Catches in pitfall traps in relation to mean densities of carabid beetles. Oecologia, 41, 2546.
  • 4
    Bergsten, U. 1985) A Study of the Influence of Seed Predators at Direct Seeding of Pinus sylvestris L. Department of Silviculture Report 13. Sveriges Lantbruksuniversitet, Umeå, Sweden.
  • 5
    Bjor, K. 1971) Forstmeteorologiske, jordbunnsklimatiske og spireøkologiske undersøkelser [Summary: Forest–meteorological, soil–climatological and germination investigations]. Meddelelser Fra Det Norske Skogforsøksvesen, 28, 429526.
  • 6
    Cajander, A.K. 1926) The theory of forest types. Acta Forestalia Fennica, 29, 141.
  • 7
    Crawford-Sidebotham, T.J. 1972) The influence of weather upon the activity of slugs. Oecologia, 9, 141154.
  • 8
    Crawley, M.J. 1992) Seed predators and plant population dynamics. Seeds: The Ecology of Regeneration in Plant Communities (ed. M.Fenner), pp. 157191. CAB International, Wallingford, UK.
  • 9
    Edwards, G.R. & Crawley, M.J. 1999) Herbivores, seed banks and seedling recruitment in mesic grassland. Journal of Ecology, 87, 423435.
  • 10
    Eurola, S., Hicks, S. & Kaakinen, E. 1984) Key to Finnish mire types. European Mires (ed. P.D.Moore), pp. 11117. Academic Press, London, UK.
  • 11
    Fenner, M., Hanley, M.E. & Lawrence, R. 1999) Comparison of seedling and adult palatability in annual and perennial plants. Functional Ecology, 13, 546551.
  • 12
    Forsslund, K.H. 1944) Något om djurlivets inverkan på barrskogens naturliga föryngring. Svenska Skogsvårdsföreningens Tidskrift, 42, 366376.
  • 13
    Fries, C., Johansson, O., Pettersson, B. & Simonsson, P. 1997) Silvicultural models to maintain and restore natural stand structures in Swedish boreal forests. Forest Ecology and Management, 94, 89103.
  • 14
    Gray, A.N. & Spies, T.A. 1997) Microsite controls on tree seedling establishment in conifer forest canopy gaps. Ecology, 78, 24582473.
  • 15
    Hagner, S. 1962) Naturlig föryngring under skärm. En analys av föryngringsmetoden, dess möjligheter och begränsningar i mellannorrländskt skogsbruk [Summary: Natural regeneration under shelterwood stands. An analysis of the method of regeneration, its potentialities and limitations in forest management in middle north Sweden]. Meddelanden Från Statens Skogsforskningsinstitut, 51, 1263.
  • 16
    Hansson, L. 1980) Small Rodents: Population Ecology, Damage to Swedish Forests and Ways of Avoiding this Damage. Concluding Report on Small Rodent Research at the Faculty of Forestry in 1971–1979 . Department of Animal Ecology Report no. 5. Swedish University of Agricultural Sciences, Umeå, Sweden.
  • 17
    Heikinheimo, O. 1937) Metsäpuiden siementämiskyvystä II. [Summary: Über die Besamungsfähigkeit der Waldbäume II]. Communicationes Instituti Forestalis Fenniae, 24, 167.
  • 18
    Heikkilä, R. 1977) Eläimet kylvetyn männyn ja kuusen siemenen tuhoojina Pohjois-Suomessa [Summary: Destruction caused by animals to sown pine and spruce seed in northern Finland]. Metsäntutkimuslaitoksen Julkaisuja, 89, 135.
  • 19
    Holm, S. 1979) A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 6, 6570.
  • 20
    Hulme, P.E. 1994) Post-dispersal seed predation in grassland: its magnitude and sources of variation. Journal of Ecology, 82, 645652.
  • 21
    Hörnfeldt, B. 1994) Delayed density dependence as a determinant of vole cycles. Ecology, 75, 791806.
  • 22
    Kerney, M.P., Cameron, C.A.D. & Jungbluth, J.H. 1983) Die Landschnecken Nord- und Mitteleuropas. Paul Parey, Hamburg and Berlin, Germany.
  • 23
    Kielland-Lund, J.J. 1963) Noen faktorer som påvirker spireresultatet for gran [Summary: Some factors affecting germination of spruce]. Tidsskrift for Skogbruk, 71, 5878.
  • 24
    Lindroth, C.H. 1985) The Carabidae (Coleoptera) of Fennoscandia and Denmark. Fauna Entomologica Scandinavica, 15. E. J. Brill Scandinavian Science Press, Copenhagen, Denmark.
  • 25
    McVean, D.N. 1961) Experiments on the direct sowing of Scots Pine. Empire Forestry Review, 40, 217227.
  • 26
    Montgomery, D.C. 1991) Design and Analysis of Experiments. John Wiley & Sons, Singapore.
  • 27
    Myllymäki, A. 1977) Outbreaks and damage by the field vole Microtus agrestis (L.) since World War II in Europe. European and Mediterranean Plant Protection Organization (EPPO): Report of the Joint FAO/WHO/EPPO Conference on Rodents of Agricultural and Public Health Concern, pp. 177207. EPPO Bulletin no. 7. EPPO, Geneva, Switzerland.
  • 28
    Myster, R.W. & Pickett, S.T.A. 1993) Effects of litter, distance, density and vegetation patch type on post-dispersal tree seed predation in old fields. Oikos, 66, 381388.
  • 29
    Niemelä, J.K. & Spence, J.R. 1994) Distribution of forest dwelling carabids (Coleoptera): spatial scale and the concept of communities. Ecography, 17, 166175.
  • 30
    Niemelä, J., Langor, D. & Spence, J.R. 1993) Effects of clear-cut harvesting on boreal ground-beetle assemblages (Coleoptera, Carabidae) in western Canada. Conservation Biology, 7, 551561.
  • 31
    Niemelä, J., Tukia, H. & Halme, E. 1994) Patterns of carabid diversity in Finnish mature taiga. Annales Zoologici Fennici, 31, 123129.
  • 32
    Nilsen, P. 1986) Tap av frø og spireplanter av gran (Picea abies (L.) Karst.) etter såing på forskjellige vegetasjonstyper i fjellskog og lavlandsskog [Summary: Loss of seeds and seedlings of Norway spruce (Picea abies (L.) Karst.) after sowing on different vegetation types in mountain forest and lowland forest]. Meddelelser Fra Norsk Institutt for Skogforskning, 39, 130145.
  • 33
    Nilsson, M.C., Steijlen, I. & Zackrisson, O. 1996) Time-restricted seed regeneration of Scots pine in sites dominated by feather moss after clear-cutting. Canadian Journal of Forest Research, 26, 945953.
  • 34
    Nystrand, O. 1998) Post-dispersal predation on conifer seeds and juvenile seedlings in boreal forest. Acta Universitatis Agriculturae Sueciae: Silvestria, 61. Doctoral Thesis. Swedish University of Agricultural Sciences, Umeå, Sweden.
  • 35
    Nystrand, O. & Granström, A. 1997a) Forest floor moisture controls predator activity on juvenile seedlings of Pinus sylvestris. Canadian Journal of Forest Research, 27, 17461752.
  • 36
    Nystrand, O. & Granström, A. 1997b) Post-dispersal predation on Pinus sylvestris seeds by Fringilla spp: ground substrate affects selection for seed color. Oecologia, 110, 353359.
  • 37
    Ohlson, M. & Zackrisson, O. 1992) Tree establishment and microhabitat relationships in north Swedish peatlands. Canadian Journal of Forest Research, 22, 18691877.
  • 38
    Örlander, G., Nilsson, U. & Nordlander, G. 1997) Pine weevil abundance on clear-cuttings of different ages: a 6-year study using pitfall traps. Scandinavian Journal of Forest Research, 12, 225240.
  • 39
    Ottosson Löfvenius, M. 1993) Temperature and radiation regimes in pine shelterwood and clear-cut area. Doctoral Thesis. Department of Forest Ecology, Swedish University of Agricultural Sciences, Umeå, Sweden.
  • 40
    Riley, L.F. 1995) Criteria and indicators of sustainable forest management in Canada. Water, Air, and Soil Pollution, 82, 6770.
  • 41
    Samran, S., Woodard, P.M. & Rothwell, R.L. 1995) The effect of soil water on ground fuel availability. Forest Science, 41, 255267.
  • 42
    SPSS Inc. (1996) Systat 6.0 for Windows: Statistics. SPSS Inc., Chicago. MA.
  • 43
    Szysko, J., Vermeulen, H.J.W. & Boer, P.J.D. 1996) Survival and reproduction in relation to habitat quality and food availability for Pterostichus oblongopunctatus F. (Carabidae, Col.). Arthropod Natural Enemies in Arable Land. II. Survival, Reproduction and Enhancement (eds C.J.H.Booij & L.J.M.F.D.Nijs), pp. 2540. Acta Jutlandica, 71. Aarhus University Press, Aarhus, Denmark.
  • 44
    Thiele, H.-U. 1977) Carabid beetles in their environments. Zoophysiology and Ecology, Vol. 10 (ed. D.S.Farner). Springer Verlag, Berlin, Germany.
  • 45
    Topping, C.J. & Sunderland, K.D. 1992) Limitations to the use of pitfall traps in ecological studies exemplified by a study of spiders in a field of winter wheat. Journal of Applied Ecology, 29, 485491.
  • 46
    Vaartaja, O. 1950) On factors affecting the initial development of pine. Oikos, 2, 89108.
  • 47
    Vaartaja, O. 1954) Factors causing mortality of tree seeds and succulent seedlings. Acta Forestalia Fennica, 62, 131.
  • 48
    Waldén, H.W. 1981) Communities and diversity of land molluscs in Scandinavian woodlands. I. High diversity communities in taluses and boulder slopes in SW Sweden. Journal of Choncology, 30, 351372.
  • 49
    Wennström, U., Bergsten, U. & Nilsson, J.E. 1999) Mechanized microsite preparation and direct seeding of Pinus sylvestris in boreal forests – a way to create desired spacing at low cost. New Forests, 18, 179198.
  • 50
    Willson, M.F. & Whelan, C.J. 1990) Variation in post-dispersal survival of vertebrate-dispersed seeds: effects of density, habitat, location, season, and species. Oikos, 57, 191198.
  • 51
    Yli-Vakkuri, P. 1961) Kokeellisia tutkimuksia taimien syntymisetsä ja ensi kehityksestä kuusikoissa ja männiköissä[Summary: experimental studies on the emergence and initial development of tree seedlings in spruce and pine stands]. Acta Forestalia Fennica, 75, 112.
  • 52
    Zackrisson, O., Nilsson, M.C., Dahlberg, A. & Jäderlund, A. 1997) Interference mechanisms in conifer–Ericaceae–feathermoss communities. Oikos, 78, 209220.

Received 21 February 1999; revision received 22 January 2000