Taxon-specific responses to different forestry treatments in a temperate forest

There are only few studies that explore the ecological consequences of forest management on several organism groups. We studied the short-term effects of four forestry treatments including preparation cutting, clear-cutting, retention tree group and gap-cutting in a temperate managed forest on the assemblage structure of understory plants, enchytraeid worms, spiders and ground beetles. Here we show, that the effect of treatments on the different facets of assemblage structure was taxon-specific. Clear-cutting and retention tree group strongly impoverished enchytraeids assemblages. Even if the species richness and cover of plants increased in clear-cutting and gap-cutting, their species composition moderately changed after treatments. For spiders only their species composition was influenced by the treatments, while the response of ground beetles was slightly affected. Short-term effect of forest management interventions on biodiversity might be compensated by the dispersal (spiders, ground beetles) and resilience (plants) of organism groups, however sedentary soil organism showed high sensitivity.

. In many cases their alpha diversity and cover showed positive response to management and was higher in intensively managed forests than in old-growth ones, because management usually increased light availability, which is a limiting factor for many species 12,18,19 . However, following the higher structural complexity, beta and gamma diversities were generally higher in old-growth than in managed stands 19,20 . Management intensity also effects species composition favouring non-forest, light demanding species, whereas many shade tolerant ancient woodland species disappeared after clear-cutting [21][22][23] .
Animals are more mobile and less resilient than plants, thus they can respond very fast to the alterations of habitat conditions. The effect of forest management on true soil organisms has been rarely studied, even though these organisms, such as enchytraeid worms (Annelida: Clitellata: Enchytraeidae) often play a crucial role in litter decomposition 24 . Changes in soil conditions such as moisture and temperature may mediate management effects on these groups of organisms 25 . A previous study revealed slight management effects on the abundance and composition of these worms in boreal forests, suggesting a beneficial effect of gap-cutting on their abundance as compared to closed stands 26 . Ground-dwelling arthropods such as spiders (Aranae) and ground beetles (Coleoptera: Carabidae) are less bound to the substrate and are more mobile, occupying a large variety of niches 27 . They respond to stand structural complexity at various temporal and spatial scales 28 and they are strongly influenced by natural and anthropogenic disturbances [29][30][31] . They have been suggested as bioindicators of commercial forestry management 32,33 , being sensitive to environmental factors such as temperature, humidity and vegetation structure [34][35][36][37] .
To compare the effect of different forest management approaches on biodiversity, we studied the response of a range of organism groups with varying sensitivity to site conditions modified by management. In a randomised block experiment we investigated the short-term effect of four forestry treatments (each with a control -C) commonly used in rotation forestry system (preparation cutting -P, clear-cutting -CC, retention tree group -R) and selection forestry system (gap-cutting -G) on four organism groups before-and two years after implementation. Our major aim was to explore how the assemblages of these organism groups (composition, species richness and abundance) react to the different treatments. We also recorded the microclimatic conditions in the treatments, which enable to link the responses of the assemblages to the treatment induced by environmental changes.
Our hypotheses were the followings: (i) for all studied community parameters (species composition, species richness and abundance) we may expect strong responses from enchytraeid worms due to their increased sensitivity for environmental parameters by their relatively low mobility. For plants, the short term response will be intermediate due to the limited growth rate of newly established plants and survival of the original vegetation. Spiders and carabids might be less sensitive to treatment effects due to their higher mobility as compare to other studied organisms. In relation to treatment types, (ii) we expect the strongest responses of all assemblages in clear-cutting because it causes the most drastic changes of forest site conditions. Retention tree group (in the clear-cuts) can buffer more effectively light condition than temperature changes, thus we are expecting stronger responses from animals (mainly soil organisms) than from plants. Gap

Results
Assemblage composition. We observed a strong compositional difference in the case of enchytraeids and spiders, and a slight, but still significant difference in the case of plants and ground beetles across treatments ( Fig. 1, Table 1). For plants (Fig. 1A), the plots representing different treatments overlapped considerably in the multivariate ordination space. When comparing the difference in species composition of plants across treatments, measured as the Bray-Curtis (BC) dissimilarity of species composition of a plot before and after treatments, we found that the gap-cutting was significantly different from control and retention tree group (Fig. 2). The enchytraeid worms of retention tree group separated from other treatments, while gap-cutting and clear-cutting (bottom of Fig. 1B) diverged from control and preparation cutting (upper part of Fig. 1B). Similar pattern was revealed by BC dissimilarities, the retention tree group was significantly different from gap-cutting, preparation cutting and control (Fig. 2). Spider assemblages in treatment plots were separated in the ordination space from those in control plots (Fig. 1C). Assemblages in the clear-cutting plots were the most distinct from control. Spider species composition in other treatments was intermediate between control and clear-cutting, largely overlapping with each other. The dissimilarity analysis showed that the control was different from all the other treatments, expect preparation cutting (Fig. 2). For ground beetles the plots of control, retention tree group and preparation cutting formed relatively homogeneous groups, but the treatments showed considerable overlap (Fig. 1D). The BC dissimilarities of control and preparation cutting were significantly lower than those of other treatments (Fig. 2). The compositional changes were significantly different between 2014 and 2016 in each taxa (Fig. 2). Species richness. The effect of the treatments for species richness was the strongest for enchytraeids, intermediate for plants and slightly significant for spiders and ground beetles ( Table 2). Plants' species richness increased in clear-cutting and gap-cutting two years after the treatments, whereas it did not change in control, preparation-cutting and retention tree group (Fig. 3). Enchytraeids' species richness increased in control and preparation-cutting during the experiment (significant year effect), while it was the lowest in retention tree group. The species richness of spiders was not different between years or treatments, except for the control treatment, where a significant drop in species richness could be observed from 2014 to 2016. The number of ground beetle species decreased in clear-cutting and gap-cutting between years, while it was the significantly the highest in the retention tree group. Abundance. The effect of treatments on between-years abundance differences was the strongest for enchytraeids, intermediate for plants and was not significant for spiders and ground beetles (Table 3). There is an overall increase in the cover of plants in all treatments between years (Fig. 4). However the abundance was significantly higher in clear-cutting, gap-cutting and preparation cutting than in other treatments. There was a concordance in the response of enchytraeids between the years and treatments, their abundance significantly decreased in clear-cutting and retention tree groups for both effects. The only change in the abundance of spiders was a decline between years in the control treatment. Although there is an overall decline in the abundance of ground beetles between years, the treatment effect was not significant.

Discussion
The integration of biodiversity conservation in forest management (and its planning) may requires effective monitoring of spatial and temporal changes of forest biotas including selected organisms with management-specific responses. Although this paper focused on the direct effects of treatments on the communities, these treatments affects through the changes of environmental conditions, which are detailed in electronic supplement (ESM- Fig. 2). The strongest environmental changes were detected in clear-cuttings: light conditions considerably increased, air temperate increased, air humidity decreased with high diurnal variance, soil temperature  and humidity also increased. Gap-cutting is characterised by high light conditions, but buffered air temperature and humidity, it has the highest soil moisture and lowest soil temperature among treatments. In retention tree group light conditions were similar as in the control, but air temperature and humidity was similar to the clear-cutting (only the diurnal variance was buffered), it is characterised by low soil moisture and high soil temperature. Preparation cutting was the most similar to the control, it showed only slight increase in light, air and soil temperature. Considering all groups, the strongest treatment effects were detected in clear-cuttings which can be expained by the drastic changes in all environmental variables (ESM- Fig. 2). Gap-cutting had a positive effect on the herbaceous vegetation, possibly as a result of the increased ground-floor light availability. Retention tree groups where soil moisture decreased and soil temperature increased considerably, had a negative effect on sedentary enchytraeids, but a slightly positive effect on ground beetles. Preparation-cutting, which caused only moderate environmental changes, brought limited changes in the studied organism groups. In addition, these effect of treatments on the studied organisms group might be explained by their taxon-specific mobility. This was best exemplified by the sedentary, non-resilient enchytraeids, for which all the three studied community parameters changed considerably. The cover of non-mobile plants was affected, but their species composition responded weakly. In ground beetles, which have moderate mobility and resiliency, richness and composition showed relatively weak responses. In spiders, which can be regarded the most mobile group, species composition changed gradually along the management intensity gradient from clear cutting toward control treatments, while species richness and abundance showed difference between control and all other treatments.
Organism specific assemblage responses to management. Sustainable forest management demands a better understanding of the environmental drivers affecting assemblage composition and diversity across taxonomic groups to ensure the multifunctionality of forests 38,39 . Thus, it is important to explore the interactions between forestry treatments and the different taxonomic groups unravel possible ecological consequences. We found that plants and enchytraeids provided the most drastic response two years after the implementation of the treatments. Thus, we suggest that these groups can act as early warning signalers of undesired effects of  forest management due to their short-time response. The direction of changes of the two groups was different: in clear-cutting the richness and abundance of plants increased while those of enchytraeids decreased, retention tree group did not seem to affect plants but had a strong negative effect on enchytraeids. It can be assumed that the apparently low dispersal capabilities of enchytraeids make this group especially sensitive to habitat alteration, because local extinctions are not compensated by immigration. Previous findings 26 , for instance, proved that enchytraeid density responded negatively to forestry treatments in Finland, however, three years after the implementation some management-specific density increase was observed. This study also pointed out that decomposer communities such as enchytraeids are not solely controlled by soil resources, but also by abiotic soil conditions which are greatly affected by forest management 26 .
The species composition of plants, changed slightly, which is possibly linked to the presence of perennial species and to the seed bank in the soil. However, the improved light conditions and soil moisture in treated plots might lead to an increased cover of existing species, and to the appearance of immigrant and dormant species, resulting in an increase in cover and species richness 40 . The sensitivity of spiders and ground beetles to forest management was inconsistent; the abundance of both groups was mostly unaffected by the treatments; however, Full circles shows the mean, white space between the circles the standard error for mean, while vertical lines denote the standard deviations. The stars denote the real difference from zero (value of 2014) based on a regression through the origin, the letters designate the significant differences among treatments, significance level were set at 0.05 for both cases.   41 demonstrated that the high mobility of orthopteran species can mask spatial heterogeneity between habitats, whereas sedentary species contributed more to maintaining beta diversity between habitats. We may argue that similarly to the orthopteran study, the high mobility of spiders and ground beetles might mask the response to forest management, a pattern that can be identified in the almost equal abundance in the various treatments.

Management affects organism groups through different environmental variables. Plant under-
story often responds strongly to forest management 37,42,43 , because management practices change the microclimatic conditions of the forest habitat. These patterns may explain the high species richness and cover of plant understory in clear cutting and gap cutting, as well as the high cover in preparation cutting. We found that clear-cutting and gap-cutting had the highest light and soil moisture levels (ESM- Fig. 2), while soil temperature and vapor pressure deficit was the highest in the clear-cutting as compared to other treatments. This is in agreement with a previous study 42 which suggests that post-treatment microclimatic conditions might be responsible for changes in plant communities, chiefly for the appearance of non-forest plants in clear-cuts. We argue that in spite of these changes the weak differences in the species composition of plants in relation to management can be explained by the survival of perennial species 44 . Similarly to plant communities, we found that the enchytraeids responded promptly to the treatments and associated changes in environmental parameters. Drastic increase in soil temperature (ESM- Fig. 2) in retention tree group and clear-cutting treatments may contribute to the rapid decrease in both spcies richness and abundance in this strictly sedentary group 26 . However these issues may require further investigations as two recent global meta-analyses revealed that there is no evidence that retention forestry in temperate forests would be a safeguard of biodiversity 9,45 , however one of the major function of this treatment type to conserve the native forest biotas for the recolonisation of the logged area during forest regeneration. In addition, these meta-analyses also suggest that the size and the spatial arrangement of retention trees may influence the effectiveness of this treatment, in which regard further investigations might be necessary. Ground-dwelling predatory arthropods are among the best indicators of forest management 37,43 . These groups have a short lifespan, have a higher position in the food web and give a complex response to changes in their abiotic and biotic environment 35 . These groups give a good example that evoking environmental filtering is often insufficient to explain assemblage responses. As argued in a recent opinion paper 46 , abiotic environment determines assemblage composition not only directly via survival, but also by affecting biotic interactions. Spider abundance and species richness did not change significantly in the treatments, however their species composition was very sensitive to treatments induced environmental changes. Previous studies showed that spider assemblages react quickly to changes in vegetation structure 28,47,48 and vascular plant richness being species turnover enabled by the high mobility of spider species. Ground beetles, a group with moderate mobility, gave weaker and somewhat less specific response manifesting in a general decline in abundance. This is presumably caused by the short-term response of these invertebrates, which is explanied by the loss of forest specialists due to treatments 36 . Ground beetles reacted positively to the retention group treatment, where the higher species richness of ground beetles might be explained by the appearance of open-habitat species 36 . Some recent findings 38 revealed that functional diversity of ground beetles are not influenced only by the diversity of forest ground vegetation, other indirect drivers may occur. All issues mentioned above prompted us to emphasize that treatment effect was the strongest in retention tree groups and clear-cutting for all studied organisms in the present study.
Conclusions. Multiple ecosystem functions in forests require sets of species forming communities.
Particularly in the light of global climatic change scenarios, which predict more frequent disturbances and extreme climatic events, it is important to explore the relationship between biodiversity and forest management practices in forestry since the forest can be a tool for mitigation of climate change 49 . Yet, the majority of research on the impact of forestry on biodiversity has focused on a specific relation between a certain management type and the response of a selected organism group. Although our results do not provide direct evidence, we suggest that dispersal dynamics may play a crucial role in the taxon-specific responseto various forest management practices. Enchytraeid worms, the least mobile animal group in the study, gave the most pronounced short-term response to the treatments. Thus, from all studied invertebrate organisms, they can be candidate early-warning signalers in Central European forests. In contrast, the more mobile ground-dwelling predatory arthropods, spiders and ground beetles, did not exhibit any easily interpretable mass effect, presumably due to their quick colonization by dispersal which enabled them to adapt to changes in the forest environment. In our multi-taxon approach we could identify organism groups that gave easily detectable mass response, while others gave taxon-specific responses, that might allow the detection of slight environmental changes. Such as in the present study, where we could prove that gap-cutting had the least adverse effect on forest dwelling organisms, the approach presented here might be useful in monitoring the ecological effects of various forestry treatment practices simultaneously. These results can contribute to the assessment of forest management's effect on biodiversity in Central European oak-hornbeam forests. However there are sporadic information available about this forest type and its region, which demands further investigations in the future.

Study area. The study area of the Pilis Experiment was in the vicinity of Pilisszántó village in the Pilis
Mountains (N47°40′ and E18°54′), Northern part of Hungary ( Supplementary Fig. 1). The elevation is 370-470 m a.s.l., average annual mean temperature is 9.0-9.5 °C, mean annual precipitation of 600-650 mm 50 . The bedrock consists of limestone and red sandstone with loess, the most common soil type is lessivage brown forest soil (luvisol). The experiment was carried out in a structurally homogenous, 80 years old, 40 ha sized, managed sessile oak-hornbeam forest stand 5 (Natura 2000 code: 91G0 5 ). The stand was cut many times in the past, recently it has been managed by shelterwood silvicultural system resulting into an even-aged, structurally homogenous stand. The microclimatic conditions (photosynthetic active radiation, relative diffuse light, air temperature, air humidity, vapor pressure deficit, soil temperature, soil water content) of the treatments were systematically measured before and after the implementation of the treatments. The methodology and major results are detailed in the Electronic Supplementary Material (Fig. S2).
The surveys of vascular plants was carried out in a 2 × 2 m sized quadrat in each plot within a fenced 6 × 6 m area excluding the effect of grazing. The cover of species was estimated in percentage, only arboreal individuals under 0.5 m height were included in the sampling. The survey was carried out in spring (April) and summer (June). Data of the two aspects were merged using the maximum cover values of each recorded species. We sampled enchytraeids (Annelida: Enchytraeidae) through soil samples taken with a soil corer (diameter 5 cm, depth: 12 cm resulting 235 cm 3 sample volume). Three samples were taken in the plots in spring and autumn, every year, and mixed into an average sample (ca. 235 cm 3 ), the worms were extracted by wet funnel method 52 . The datasets from the two sampling occasions were pooled by year.
To sample ground dwelling predatory arthropods (spiders (Aranae) and ground beetles (Coleoptera: Carabidae)), four pitfall traps were installed in every plot around the fenced area in each direction. Two sampling intervals (one month in spring and one in autumn) was set, corresponding to the highest activity regime of the beetles 53 and spiders 54 . The traps were made of 85 mm diameter plastic cups; each containing approximately 250 cm 3 of a 50% solution of propylene glycol and water, saturated with salt and with a drop of odorless detergent to reduce surface tension. A dark green plastic roof protected the solution from litter and rain. The data of the pitfall traps of the same plots were merged resulting 30 sampling units for both study years. The traps were checked monthly.
Ethical Approval. The observation of plant communities in this study were non-invasive, while the field sampling of invertebrates such as enchytraeid worms, spiders and ground beetles were conducted under the license from the respective Hungarian authority (Közép-Duna-Völgyi Környezetvédelmi és Természetvédelmi Felügyelőség KTF:30362-3/2014).

Data analysis. Non-metric Multidimensional Scaling (NMDS) and Permutational Multivariate Analysis
of Variance (PERMANOVA) with square-root transformation was applied to test for dissimilarities in species composition among treatments using the Bray-Curtis index of dissimilarity 55 . Data before and after treatments were analyzed separately, but only the post-treatment analyses (dataset from 2016) are presented here. In order to explore the direction of changes in assemblage composition, we calculated Bray-Curtis dissimilarities between the same plots in 2014 and 2016, the effect of treatments on the dissimilarities were compared using Generalized linear mixed-effect models 56 (GLMM) as described below.
The (signed) differences between 2014 and 2016 for species richness and abundance (cover for plants) were used for all organism groups as response variables during the analyses. Positive values indicated increase, while negative values denoted a decrease in the variables after treatment implementation. GLMMs were used to explore the effect of treatments (considered as a fixed effect term) on species richness and abundance difference (as response variables) of the studied organisms, while blocks were used as a random factor. Two families of distribution were applied: "Poisson" for species richness data and "Gaussian" for abundance data and Bray-Curtis dissimilarities. The models were tested with the default Laplace approximation to the log-likelihood. The model diagnostics includes the inspection of model residual's structure (Pearsons's type) versus fitted values and degrees of freedom either in model's output or in graphs (ESM Tables 1-3). For overdispersions, we also compared the Poisson vs. Quasipoisson fit for the same model structure; Gaussian distribution was revelaed by Shapiro-Wilk test and quantile-quantile plots. In case of significant treatment effects, the differences between treatment levels were evaluated by multiple comparisons (with Tukey computed contrast matrices for several multiple comparison procedures). We also tested the significance of changes (cited as year effect in results) between 2014 and 2016 (true difference in estimates from zero) for each treatment by repeating the models with exclusion the effect of intercept 57 .

Data Availability
All data generated or analyzed during this study are included in this published article (and its Supplementary  Information Files).