Introduction

Natural disturbances are one of the key drivers of forest dynamics (Svoboda et al. 2012). Large-scale, high-severity disturbances such as windstorms and insect outbreaks are greatly important for the structure and functioning of montane spruce forests in Central Europe. The Šumava Mountains constitute one of the biggest forested areas in Central Europe and the history of its forests is shaped with big windstorms and subsequent bark beetle (Ips typographus) outbreaks (Svoboda et al. 2012).

A large area of natural and semi-natural forests on the top part of the Šumava Mts. suffered a bark beetle outbreak in the 1990s that killed almost all mature trees. The area was left to regenerate naturally without salvage logging, because it is part of the core zone of the Šumava National Park where natural processes are supposed to govern its development. Studying these natural processes can help us to understand the responses of the ecosystem to disturbance and adapt forestry techniques to make forest management sustainable (Marshall 2000).

Besides the detrimental impacts on vegetation cover, this kind of disturbance also seriously affects the belowground part of the ecosystem. The soil and litter are warmer with more temperature extremes during the day in the later phases of the forest dieback (Hais and Kučera 2008; Griffin et al. 2011). The litter depth increases immediately after outbreak and the subsequent decomposition of needles releases high amounts of nutrients to the soil (Kaňa et al. 2013). Tree death likely leads to elimination of root-mycorrhizal connections and decline of the fungal community (Siira-Pietikäinen et al. 2001; Brant et al. 2006).

In our study, we focused especially on two major groups of soil microarthropods (Oribatida and Collembola) that are expected to be sensitive to the factors described above and have an important role in ecosystem functioning. They are highly sensitive to changes in moisture and temperature (Niedbala 1980) and concurrently are not able to escape unfavorable conditions due to their limited long-distance dispersal (Ojala and Huhta 2001). In addition, they are connected with primary decomposers through complex bottom-up and top-down effects (Marshall 2000; Osono 2007) and participate in organic matter decomposition and nutrient cycling (Luxton 1981; Takeda 1995). The structure of soil food web was found to govern the processes of C and N cycling and related ecosystem services (de Vries et al. 2013).

The aim of our study was to describe the microarthropod community in terms of (1) density and diversity; (2) species composition and (3) life-history and functional traits. We hypothesized that the microarthropod community would be dominated by disturbance-tolerant, opportunistic, detritivorous species at the expense of sensitive, specialized, fungivorous ones.

Methods

Site description

The study area is located in the central part of the Šumava Mountains (Czech Republic; 48°57′N, 13°28′E) close to Březník and the border with Germany. The altitude in the study area ranges between 1210 and 1310 m a.s.l. The climate is cold and wet, with mean annual temperature about 4 °C and mean annual precipitation about 1300 mm. The bedrock is gneiss covered predominantly with a mountain humus podzol. The forest in the area is naturally dominated by Norway spruce (Picea abies), with Calamagrostis villosa, Avenella flexuosa, Soldanella montana, Luzula sylvatica and Vaccinium spec. div. in the herb layer (Albrecht 2003).

The area is a part of the core zone of the Šumava National Park. It suffered a massive bark beetle outbreak in the 1990s that gradually caused the forest dieback. Because the forest in the area was supposed to be natural and uninfluenced by humans, it was left without any active human intervention to regenerate naturally. At the time of sampling, the area was covered with a mosaic of standing dead trees, rare old spruce trees that survived, and numerous spruce, birch and rowan saplings of various ages. The herb layer was well developed and consisted of all the typical species of a climax montane spruce forest.

Microarthropods sampling and abiotic parameters measurement

Soil samples were taken in late October 2007 and November 2008 (10 cm2 surface area and 8 cm depth) in the prevailing microhabitat of Calamagrostis villosa patches. We took 20 samples for microarthropod extraction on each sampling date. Soil arthropods were extracted in high-gradient MacFadyen funnels for 6 days. Collembola, Oribatida, Gamasida, Prostigmata and Astigmata were counted. Collembola from all samples and Oribatida from half of them (10 samples in October 2007 and 10 in November 2008) were determined to species level. When possible, juveniles were also determined, counted and added to the respective adults.

Chemical analyses were performed on the <2 mm fraction of the dried soil samples after microarthropod extraction. Soil moisture was calculated as the difference between weights of fresh and dried soil samples (dried at 105 °C; 20 samples on each sampling date), and pH was determined in a 1 M KCl solution (20 replicates on each sampling date). Carbon and nitrogen contents were measured using a CN analyzer (NC 2100 Soil Analyzer, ThermoQuest Italia S.p.A.) in composite soil samples (consisting of five subsamples, 4 replicates per sampling date).

Oribatida and Collembola community data treatment

The Oribatida and Collembola communities were described by density, species richness (species number per sample), Shannon’s diversity index (H, using log10) and the Berger-Parker index (contribution of the most dominant species; Caruso et al. 2007). In addition, eight morphological and life-history traits were used in the study (sensu Violle et al. 2007; Table 1). Reproductive mode (sexual/parthenogenetic) is connected with reproduction rate and was shown to be important for recovery after disturbance (Lindberg and Bengtsson 2005) as well as vertical distribution of the species (Malmström 2012). Feeding guilds should provide insight into the roles in ecosystem functioning that the microarthropods fulfill. Information on feeding preferences of different species was collected from multiple sources which are shown in Tables 2 and 3 in the “Appendix”.

Table 1 Description of Oribatida and Collembola traits
Full size table
Table 2 Oribatida species arranged according to their mean abundance (±SE) and their trait attributes (Feeding: 1—fungi and animals; 2—fungi, in part litter; 3—predominantly litter; 4—algae and lichens; Reproduction: S sexual, P parthenogenetic)
Full size table
Table 3 Collembola species arranged according to their mean abundance (±SE) and their trait attributes (Feeding: 1—fungi; 2—litter and fungi; 3—litter and algae; Reproduction: S sexual, P parthenogenetic; Furca: 0—absent, 1—reduced; 2—developed; Life-form: Eu euedaphic, H hemiedaphic, Ep epedaphic)
Full size table

Statistics

A preliminary one-way ANOVA was performed at p < 0.05 to verify whether density, the diversity indices and trait distribution differed between the sampling dates. Because we did not find any significant differences, data from both years were pooled and treated together.

The densities of functional groups (feeding guilds, reproduction, life-forms, species with absent/reduced/developed furca) were analyzed by one-way ANOVA. Comparison of means was performed with the Tukey HSD test (p < 0.05) when significant differences appeared. Species with unknown trait attributes were omitted from the trait assessment (their abundances are given in Figs. 2, 3).

The results obtained in the regenerating forest were compared with data from unaffected, mature spruce forest and from bark-beetle gradation. Data for comparison were taken from literature (Siira-Pietikäinen et al. 2001; Starý 2003; Materna 2004; Matějka and Starý 2009; Čuchta et al. 2012; Lóšková et al. 2013; Farská et al. 2014). We used principal component analysis (PCA) to show main gradients in the communities. A linear response model was selected as recommended for small gradients by Lepš and Šmilauer (2003). Analyses were carried out using Canoco, Version 5. Species data were centered.

Results

Soil throughout the study area had a low pH (3.17 ± 0.03; all values represent mean ± SEM) and high moisture (69.0 % ± 0.9). Carbon content was 29.8 % ± 2.7 and nitrogen content 1.5 % ± 0.1.

Oribatids were the most dominant group of soil microarthropods with a density of 167250 ± 20248 individuals per square meter (mean ± SEM). In addition, we found high numbers of Astigmata and Prostigmata (59375 ± 7436 ind. × m−2), Collembola (49475 ± 4729 ind. × m−2) and Gamasida (15150 ± 1510 ind. × m−2).

In total, we recorded 49 Oribatida and 48 Collembola species. Cumulative species richness curves (Fig. 1) illustrated higher richness of Oribatida community compared to Collembola. Similarly, mean number of species per soil sample was 15.15 ± 0.77 (mean ± SE) for Oribatida and 10.73 ± 0.54 for Collembola. Berger-Parker index of Oribatida community was 0.38 ± 0.03 and that of Collembola was 0.35 ± 0.02.

Fig. 1
figure 1

Cumulative speciess richness curves of Oribatida and Collembola communities

Full size image

The communities were dominated by Tectocepheus velatus (28 %), Platynothrus peltifer, Scheloribates initialis and Oppiella falcata for Oribatida; and Isotomiella minor (22 %), Folsomia sensibilis, Tetracanthella stachyi and Parisotoma notabilis for Collembola (Tables 2, 3 in the “Appendix”).

Principal component analysis was used to illustrate variability in communities from unaffected, affected and regenerating forests. The main gradient, characterized by the first ordination axis, clearly separated oribatid assemblages according to the disturbance status (Fig. 2). Collembola assemblages were primarily arranged according to their geographic origin (Fig. 3).

Fig. 2
figure 2

Principal component analysis ordination diagram of Oribatida community, the first axis explained 62.21 % of variability. Samples are presented with black circles (unaffected forests), empty circle (regenerating forest) and empty triangles (bark-beetle gradation). Data sources: 1–6 Farská et al. (2014), 7–8 Lóšková et al. (2013), 9–10 Starý (2003), 11 our results, 12–13 Matějka and Starý (2009)

Full size image
Fig. 3
figure 3

Principal component analysis ordination diagram of Collembola community, the first axis explained 70.16 % of variability. Samples are presented with black circles (unaffected forests) and empty circle (regenerating forest). Data sources: 1–6 Farská et al. (2014), 7 Materna (2004), 8 our results, 9 Siira-Pietikäinen et al. (2001), 10–12 Čuchta et al. (2012)

Full size image

Oribatid mean size was 0.43 mm and collembolan 1.10 mm. Parthenogenetic species reached significantly higher abundances than those sexually reproducing (F 1,38 = 12.494, p < 0.01; F 1,78 = 5.287, p < 0.05; for Oribatida and Collembola respectively; Fig. 4). We also found a higher occurrence of detritivorous oribatids than fungicarnivorous and detritofungivorous species (F 2,57 = 4.357; p < 0.05; Fig. 5). In the Collembola community, fungivores and detritofungivores were highly abundant while detritivores were missing (F 2,117 = 29.176; p < 0.001; Fig. 5). In terms of Collembola life-forms, euedaphic species dominated (F 2,117 = 40.952; p < 0.001; Fig. 6) while species with developed furca were also highly abundant (F 2,117 = 12.915; p < 0.001; Fig. 6).

Fig. 4
figure 4

Densities (±SE) of parthenogenetic and sexually reproducing oribatids and collembolans. Significant differences between groups are marked with bold letters (one-way ANOVA and post hoc Tukey test)

Full size image
Fig. 5
figure 5

Densities (±SE) of oribatids and collembolans feeding on various food sources. Significant differences between groups are marked with bold letters (one-way ANOVA and post hoc Tukey test)

Full size image
Fig. 6
figure 6

Densities (±SE) of euedaphic, hemiedaphic and epedaphic collembolans and densities of Collembola species with absent, reduced and developed furca. Significant differences between groups are marked with bold letters (one-way ANOVA and post hoc Tukey test)

Full size image

Discussion

Species composition

The soil microarthropod community in the area still seemed to be influenced by forest dieback and differed from that of undisturbed natural forest. Tectocepheus velatus, the most abundant species recorded, is one of the few oribatids capable of tolerating strong anthropogenic stress; it occurs at high abundance in habitats impacted by human activity like arable soil and colliery dumps after brown coal mining (Frouz et al. 2001; Luptáčik et al. 2012). It is tolerant to mechanical disturbance (Maraun et al. 2003) and to harsh environmental conditions (Maraun and Scheu 2000). Similarly, Platynothrus peltifer, the second most abundant microarthropod, is resistant to mechanical perturbation (Maraun et al. 2003) and is able to reproduce very quickly after disturbance (Travé et al. 1996). Maraun et al. (2003) assumed that this species is a weak competitor that is able to dominate only under conditions of strong disturbance when sensitive species are suppressed. It probably can take an advantage of its parthenogenetic mode of reproduction and high number of eggs laid (Luxton 1981). Our results confirm that disturbance can decrease the density of sensitive species and thereby allow an increase in a few opportunistic species (Caruso et al. 2007).

The oribatid community from the regenerating forest seems to be intermediate between the unaffected and disturbed ones. The mature undisturbed spruce forests studied by Farská et al. (2014), Lóšková et al. (2013) and Starý (2003) were characterized by a more equal distribution of both eurytopic and silvicolous species compared to our results. We found silvicolous oribatids, such as Oppiella falcata, Oppiella neerlandica, Hermannia gibba, Fuscozetes setosus or Sellnickochthonius zelawaiensis, in the regenerating forest as well, but in lower numbers. The total abundances recorded in the disturbed forest are slightly lower compared to the unaffected ones (approximately 200–300 000 ind./m2). Nevertheless, the range in the literature should be considered very carefully since it depends on many variables (e.g. sampling effort or size and heterogeneity of studied forests).

The opposite trend was recorded in forests undergoing bark beetle gradation. There were lower total abundance (approximately 80–100 000 ind./m2) and number of species compared to our results from the regenerating forest, and oribatid community was even more dominated by T. velatus (Matějka and Starý 2009).

The effect on Collembola seems to be less pronounced compared to Oribatida. The Collembola community in our study was similar to those described from unaffected spruce forests by Farská et al. (2014), Materna (2004) and Siira-Pietikäinen et al. (2001). This applies for the number of species, total density as well as species composition. The most abundant species, Isotomiella minor, is a cosmopolitan, eurytopic species capable of being dominant in a wide range of habitats and environmental conditions (Wiwatwitaya and Takeda 2005). The collembolan community from spruce forest in High Tatry Mountains described by Čuchta et al. (2012) was more different as it was dominated with Folsomia penicula.

Generally, Collembola are assumed to follow an r-strategy, compared to K-strategist oribatid mites (Petersen 2002). Collembolans respond rapidly to environmental changes (Chauvat et al. 2003) and concurrently recover faster (Lindberg and Bengtsson 2006). Thus, they may show less consistent changes in community structure than oribatids.

The Berger-Parker index indicated intermediate level of disturbance. We obtained values of about 0.38 for Oribatida and 0.35 for Collembola while Caruso et al. (2007) reported a range from 0.2 for natural ecosystems to 0.5 for intensively disturbed arable fields.

Traits

A remarkable feature was the overall predominance of parthenogenetic species, especially among Oribatida. It is still not certain what role reproduction mode plays during the recovery and disturbance processes since there are various, contradictory data in the literature. On the one hand, parthenogenetic species were found to recover quickly after drought (Lindberg and Bengtsson 2005) and to reach high dominance with management intensification (Farská et al. 2014). Conversely, sexually reproducing microarthropods were faster to recolonize soil in a mesocosm experiment (Domes et al. 2007a) and quickly increased their numbers several years after clear-cut burning (Malmström 2012). Theory predicts sexual reproduction to be superior to parthenogenesis in unstable habitats because high genetic diversity allows for a faster response to a changing environment (Hamilton 1980). However, Norton (1994) assumes parthenogenesis to be an advantageous trait for colonizers, because parthenogenetic species can reproduce faster and establish a population from very few individuals.

This ambiguousness of parthenogenesis is probably caused by the high heterogeneity of life-history tactics within parthenogenetic species. They can be either fast-reproducing and disturbance tolerant (e.g. Tectocepheus velatus) or euedaphic, small-size and slowly-developing species (e.g. Brachychthonidae) (Maraun and Scheu 2000). In our study, the high presence of parthenogenesis was caused especially by the high abundance of Tectocepheus velatus and Platynothrus peltifer which are representatives of the first mentioned strategy.

The community is characterized by the high occurrence of detritivores. This generally differs from the community pattern recorded in an old-growth, undisturbed spruce forest where oribatids feeding on fungi prevail (Farská et al. 2014). The high abundance of detritivores corresponds well with the fact that bark-beetle outbreak is usually characterized by increased litter fall and formation of a thick needle layer (Griffin et al. 2011; Simard et al. 2011) and a concurrent decline in the fungal community due to elimination of root-mycorrhizal connections (Siira-Pietikäinen et al. 2001; Brant et al. 2006). All these factors suggest an important shift in soil ecosystem functioning caused by a bark-beetle outbreak and subsequent forest dieback.

The predominance of euedaphic collembolans and very low occurrence of epedaphic species indicate that deeper soil layers provide more suitable conditions in terms of food, habitat-space, etc. In addition, the low abundance of epedaphic species can be due to the severe microclimatic conditions on the surface (Lensing et al. 2005). Nevertheless, the climate in the Březník area is moist and cold with long snow cover and therefore we do not assume that the microclimatic conditions limit soil microarthropods which are much more sensitive to drought and high temperatures.

Many factors may have influenced our results complicating their interpretation. For example, many traits affect soil fauna survival and colonization, but were not included in the study. Habitat specialization and dispersal ability strongly influence post-disturbance succession (Lindberg and Bengtsson 2005; Malmström 2012), but data are lacking for many species. Thus, it is impossible to evaluate them accurately. Additionally, certain features are phylogenetically correlated and combined in the same species, therefore their separation is difficult.

It is also important to pay attention to the natural processes and forest regeneration ongoing in the non-intervention forest. The forest left for natural regeneration differs considerably from the salvage-logged one in terms of abiotic as well as biotic conditions. Generally, the microclimate is less extreme in a decayed spruce forest compared to clear-cuts, probably due to the high reflectance of dead trees and higher density of vegetation (Hojdová et al. 2005; Hais and Kučera 2008). Salvation usually reduces environmental heterogeneity, diversity of microsites, and amount and diversity of coarse woody debris (Kimmins 1997). Furthermore, the harvesting of dead trees often disturbs the forest floor resulting in soil scarification (Bouget and Duelli 2004).

All factors described above affect soil microarthropods. The lying and standing dead trees result in a complicated microrelief that provides sheltered microsites protecting microarthropods against desiccation (Johnston and Crossley 1993). Constant moisture in the deadwood further enhances growth of bacteria and mycelia that serve as food sources (Jabin et al. 2004). Thus, coarse woody debris left on the stand may support the survival of sensitive silvicolous species. The non-intervention windthrown forests were found to be more favorable for recovering edaphic collembolans, whereas harvesting and clearing favoured xeroresistant species able to withstand warmer and drier conditions (Čuchta et al. 2012).

Conclusion

Our results from the regenerating forest demonstrated high abundance of a few tolerant microarthropod species and lower numbers of sensitive silvicolous ones. This is likely the result of the massive bark beetle outbreak and subsequent forest dieback in the 1990s. Disturbance can decrease the abundance of sensitive species and thus allow an increase in a few opportunistic ones (Fountain and Hopkin 2004; Caruso et al. 2007). Although the details which determine the identity of successful species remain unknown, the common features of the most dominant species in our study were parthenogenesis and high reproduction rate (which can be linked together), and detrito- or detritofungivorous feeding. However, the occurrence of sensitive silvicolous species indicates that the ecosystem has the potential to recover. The impact of disturbance was likely mitigated by forest regeneration and the natural processes ongoing in the non-intervention regime (abundant dead wood, complicated microrelief with sheltered microhabitats etc.)

In agreement with the changed species composition, soil functioning seems to be still affected by the bark-beetle outbreak. The impact is especially well pronounced for oribatids where the high dominance of detritivores in their community differs remarkably from the situation in an undisturbed, old-growth forest.