Rejuvenation and restoration measures foster specialised and threatened carabid beetle species in montane heathland ecosystems

Montane heathlands are among the most threatened semi‐natural ecosystems across Central Europe. Nevertheless, empirical studies on the effects of rejuvenation and restoration of montane heathlands have been scarce thus far. The aim of our study was to analyse the long‐term effects of heathland rejuvenation and restoration on carabid beetle assemblages. Our study took place in the Rothaar Mountains, one of the most important strongholds of montane heathlands in Central Europe. We considered four different successional stages: (i) early‐successional heathlands as a result of rejuvenation measures (EARLY), (ii) restored heathlands (RESTORED), (iii) late‐successional heathlands (LATE) and (iv) windthrows (WIND). Our study revealed that the composition of carabid beetle assemblages differed across the gradient. From the earlier to the later stages, beetle biomass, richness of macropterous, threatened and heathland species decreased, whereas moisture and shading indicator values increased. The number of indicator species peaked in EARLY and RESTORED. Solely brachypterous species had the highest species richness in the later seral stages. Overall, rejuvenation and restoration measures fostered specialised and threatened carabid beetle species of montane heathlands. In contrast, carabid beetle assemblages of the two later and dense successional stages consisted mainly of non‐threatened habitat generalists. Vegetation structure and the interrelated microclimate are assumed to be the key drivers of assemblage composition. For the long‐term conservation of montane heathlands and their specialised carabid beetle assemblages, we recommend rejuvenation by sod cutting and choppering in a mosaic‐like manner and at intervals that clearly exceed more than two decades.

tion and restoration of montane heathlands have been scarce thus far.
2. The aim of our study was to analyse the long-term effects of heathland rejuvenation and restoration on carabid beetle assemblages. Our study took place in the Rothaar Mountains, one of the most important strongholds of montane heathlands in Central Europe. We considered four different successional stages: (i) earlysuccessional heathlands as a result of rejuvenation measures (EARLY), (ii) restored heathlands (RESTORED), (iii) late-successional heathlands (LATE) and (iv) windthrows (WIND).
3. Our study revealed that the composition of carabid beetle assemblages differed across the gradient. From the earlier to the later stages, beetle biomass, richness of macropterous, threatened and heathland species decreased, whereas moisture and shading indicator values increased. The number of indicator species peaked in EARLY and RESTORED. Solely brachypterous species had the highest species richness in the later seral stages. 4. Overall, rejuvenation and restoration measures fostered specialised and threatened carabid beetle species of montane heathlands. In contrast, carabid beetle assemblages of the two later and dense successional stages consisted mainly of nonthreatened habitat generalists. Vegetation structure and the interrelated microclimate are assumed to be the key drivers of assemblage composition. 5. For the long-term conservation of montane heathlands and their specialised carabid beetle assemblages, we recommend rejuvenation by sod cutting and choppering in a mosaic-like manner and at intervals that clearly exceed more than two decades.

K E Y W O R D S
biodiversity conservation, ground beetle assemblage, insect biomass, semi-natural habitat, species richness, successional gradient

INTRODUCTION
We are living in a time of continuous and dramatic biodiversity loss on a global scale (Butchart et al., 2010). Hence, scientists have already warned that we are heading towards a sixth global mass extinction (Barnosky et al., 2011;McCallum, 2015). In this context, the loss of insects has gained public attention, especially in the last few years, and international researchers are calling for urgent and increased efforts in habitat conservation and restoration (Cardoso et al., 2020;Harvey et al., 2020;Samways et al., 2020). In terrestrial biomes, landuse change is considered to be the main driver of this biodiversity crisis Foley et al., 2005;IPBES, 2019;Stoate et al., 2009).
Semi-natural habitats shaped by traditional low-intensity management, such as the vast majority of European heathlands (Halada et al., 2011), have been particularly affected by land-use change and have dramatically declined since the mid-19th century (Keienburg & Prüter, 2004;Symes & Day, 2003). While many lowland heathlands have been converted to arable fields, montane heathlands have largely disappeared because of abandonment of traditional land use (Hahn, 2007;Schubert et al., 2008) and afforestation (Symes & Day, 2003;Walker et al., 2004). In general, heathlands are characterised by highly specialised and, thus, often threatened species (Buchholz et al., 2013;Schirmel & Fartmann, 2014). As a result, they are of great value for biodiversity conservation and are protected by the EU Habitats Directive (EC, 2007).
Plant and insect communities of montane heathlands differ strongly from those of lowland heathlands due to a wet and cold mountain climate (Britton et al., 2005). However, nowadays, the remaining montane heathlands mostly consist of small and isolated patches with species-poor old-growth heath (Borchard et al., 2013;Streitberger et al., 2021a). Therefore, they are among the most threatened ecosystems across Central Europe (Finck et al., 2017;Hoffmann, 1998). Nevertheless, empirical studies on the effects of rejuvenation and restoration measures in montane heathlands have been scarce thus far. The few available Central European studies solely investigated choppering (removal of biomass and the organic layer) as a rejuvenation measure  or only studied the short-term effects of restoration (4-5 years after application of the measures) (Borchard et al., 2013(Borchard et al., , 2017. Recently, however, research on the long-term effects of rejuvenation and restoration of montane heathlands on phytodiversity (vascular plants, bryophytes and lichens) has been conducted (Streitberger et al., 2021a(Streitberger et al., , 2021b. The aim of our study was to analyse the long-term effects of montane heathland rejuvenation and restoration on carabid beetle assemblages (Coleoptera: Carabidae). Carabid beetles are excellent bioindicators for environmental alterations in heathland ecosystems (Buchholz et al., 2013;Bargmann et al., 2016). They (i) feature typical heathland species, (ii) respond rapidly to environmental change, (iii) are abundant, (iv) are easy to sample and (v) represent essential elements of the food chain Kotze et al., 2011;Rainio & Niemelä, 2003). Our study took place in the Rothaar Mountains, one of the most important strongholds of montane heathlands in Central Europe . We compared carabid beetle assemblages across a gradient from early-to late-successional stages. These successional stages contained (i) rejuvenated heathlands that originated either from sod cutting 10-20 years ago or regular disturbance by skiing, (ii) restored heathlands with the application of seed transfer on former spruce forests 9-10 years after restoration, (iii) old-growth heathlands without any rejuvenation measures for at least 30 years and (iv) windthrows of spruce forests with salvage logging as a reference for vegetation development without seed transfer (cf. Streitberger et al., 2021a).
In particular, we addressed the following questions: • How do the four successional stages differ in environmental conditions?
• What were the effects of rejuvenation and restoration measures on species richness and biomass of carabid beetles, and how did assemblage composition differ between the four successional stages?
• What management recommendations can be derived from our findings for rejuvenation and restoration of montane heathlands?

Study area
The study area is located in the Rothaar Mountains, a low mountain 5 C) and wet (mean annual precipitation: 1.450 mm) montane climate (Borchard et al., 2013) with snowy winters (mean snow-cover duration: 100 days/year) (German Weather Service, pers. comm.). The prevailing soils are nutrient-poor cambisols on acidic bedrock (Geologisches Landesamt NRW, 1998). The landscape is dominated by woodland, especially non-native spruce forests (Picea abies) and grassland . However, on many mountain peaks, remnants of montane heathlands still regularly occur (Streitberger et al., 2021a(Streitberger et al., , 2021b. Altogether, we examined eight different subareas with occurrences of montane heathlands having a minimum size of 4.6 ha per subarea (cf. Streitberger et al., 2021b).
In the years 2008-2009, large-scale heathland restoration adjacent to the remaining heathlands was conducted in the study area (Borchard et al., 2013;Streitberger et al., 2021a). Spruce forests on former heathlands were deforested and, additionally, remaining branches and most of the organic layer were removed. Subsequently, seeds from a nearby nature reserve ('Neuer Hagen') with representative montane heathland vegetation were transferred to the restoration sites. For this purpose, two different procedures were applied:

Sampling design
For each of the four successional stages, we randomly selected eight plots (N = 32). Each plot had a size of 20 m Â 25 m (500 m 2 ). The number of studied successional stages within a subarea corresponded to their occurrence and was at least two. The minimum distance between two plots ranged between 105 and 535 m (mean distance to nearest plot: 168 m). Subareas were 2660 to 6410 m apart (mean distance to nearest subarea: 4080 m).

Environmental parameters
For each plot, we sampled different parameters of habitat structure and macroÀ/mesoclimate (Table 1). During mid-June 2018, we surveyed habitat structure and sunshine duration. The cover of vegetation in the plot was estimated in 5% steps (2.5% steps below 10% and

Habitat structure
Cover (%) five randomly selected spots within each plot and then averaged. The duration of sunshine in June was recorded using a horizontoscope (Scherer et al., 2021). Data on mean annual temperature were derived

Classifications
Classification of heathland species were based on GAC (2009) and Trautner (2017). Accordingly, all species for which heathlands are among the main habitats were considered heathland specialists. Threatened species were assigned according to the red data book of North Rhine-Westphalia (Hannig & Kaiser, 2011). Biomass of carabid beetles was calculated for each plot based on body length of the species using the formula provided by Szyszko (1983). Body length of each species was averaged based on minimum and maximum values extracted from Klaiber et al. (2017). Additionally, we determined the wing length (macropterous vs. brachypterous, dimorphic species were considered macropterous; Klaiber et al., 2017) as a measure for the dispersal ability of the species (Kotze & O'Hara, 2003;Ng et al., 2018). Ecological preferences of carabid beetle assemblages were analysed by averaging indicator values for moisture and shading (Irmler & Gürlich, 2004) across all species of a plot.

Statistical analysis
Shannon diversity and evenness were calculated for each plot (Fedor & Zvaríková, 2019). In order to detect significant differences in environmental conditions and species data among the successional stages, (generalised) linear mixed-effect models (LMM; GLMM) were applied (R package lme4; Bates et al., 2021) with subarea as a random factor to account for potential spatial autocorrelation. Successional stage was used as a nominal fixed factor and the analysed parameters were used as dependent variables. Depending on the distribution of the variables, either proportional binomial (percentage data), Poisson or linear (for normally distributed and square-root-or log-transformed variables with normal distribution) models were applied. To reduce overdispersion within the models (binomial/Poisson), observationlevel random effects were added as a random factor (Harrison, 2014(Harrison, , 2015. The overall effect of the dependent variables on successional type was analysed by comparing the full models with reduced models without successional type as the fixed factor and applying likelihoodratio tests. Pairwise differences between the successional types were detected by applying Tukey's contrasts (glht function, R package multcomp; Hothorn et al., 2021).
In order to identify indicator species for each successional stage, we conducted an indicator species analysis (ISA) (Dufrêne & Legendre, 1997). All statistical analyses were performed using R 3.6.2 (R Development Core Team, 2021).

Environmental conditions
Macro-and mesoclimatic parameters did not differ between the four studied successional stages (Table 1). In contrast, all habitat-structure variables differedexcept trees, which only occurred in WIND at low cover.
Shrubs were only present in LATE and WIND. Therefore, the cover and height of shrubs increased from EARLY/RESTORED to LATE to WIND.
The cover of the field layer and of mosses were lowest in EARLY/WIND and highest in RESTORED/LATE. Additionally, EARLY had the highest cover of bare ground and the shortest swards compared to the three other successional stages. The cover of grasses peaked in RESTORED/ WIND compared to EARLY/LATE. Litter cover increased across the successional gradient from EARLY to RESTORED/LATE to WIND.

Response of carabid beetle assemblages to environmental conditions
In total, we collected 3283 individuals from 58 species of carabid beetles on the 32 plots (Table A1). Ten of these species were considered threatened, and seven were typical heathland species. The most common species were Poecilus versicolor (34% of all individuals), Carabus problematicus (14%) and Pterostichus burmeisteri (13%). Among threatened species, Cicindela campestris (3%) and Carabus arvensis (2%) were the most frequent ones.
Overall species richness, Shannon diversity and evenness did not differ between the four successional stages (Figure 1)

DISCUSSION
Our study revealed that the composition of carabid beetle assemblages clearly differed across the successional gradient in montane heathlands. From the earlier to the later successional stages, beetle biomass, richness of macropterous, threatened and heathland species decreased, whereas the indicator values for moisture and shading increased. The number of indicator species also peaked in EARLY and RESTORED. Solely brachypterous species had the highest species richness in the later seral stages. Since macro-and mesoclimatic conditions did not differ between the successional stages, the differences in habitat structure and the interrelated microclimate have to be considered as the main drivers of assemblage composition.
Threatened and specialised ground beetle species of heathlands are known to vitally depend on such conditions (Desender et al., 2010;Cameron & Leather, 2012;Bargmann et al., 2016). This is especially true for regions that are characterised by a cool climate, like the study area Fartmann et al., 2015). In line with this, among the indicator species, heathland specialists were only detected in EARLY (C. campestris, N. aquaticus) and the number of threatened and heathland species was highest in EARLY and lowest in WIND. The microclimatic gradient was also reflected by increasing moisture and shading indicator values within carabid beetle assemblages from earlier to later successional stages.
In RESTORED, bare ground had widely disappeared 9-10 years after restoration and swards were taller compared to EARLY. Nevertheless, in contrast to LATE and WIND, the number of heathland species still did not differ from EARLY. Characteristic of LATE was a homogeneous vegetation consisting of a dense and tall dwarf-shrub layer mainly structured by V. myrtillus but also by C. vulagris and V. vitis-idaea (see also Comparisons between groups were made by generalised linear mixed models with Tukey's contrasts (see Statistical analysis for details). The absence of concurrent letters indicates significant differences between successional stages (p < 0.05). Statistical differences are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, n.s. not significant T A B L E 2 Results of ISA (Dufrêne & Legendre, 1997)  Note: Only significant species with an indicator value (IV) ≥ 60 are shown. Indicator species are sorted by "IV" for the respective successional stage. The threatened species is highlighted in bold type. N = 8 per successional stage. Statistical differences of Monte-Carlo tests are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001. Abbreviations: HS, heathland species; ISA, indicator species analysis. Streitberger et al., 2021a). Dense dwarf-shrub stands, especially those of the broad-leaved Vaccinium species, build close canopies that hamper solar radiation to reach the near-ground layer; the result is a cool and moist microclimate (Stoutjesdijk & Barkman, 1992). At WIND, even some trees and a shrub layer had established. In contrast, RESTORED had the most heterogeneous vegetation of all four successional stages. The diversity of vascular plants was highest (Streitberger et al., 2021a) and it exhibited an intermediate cover of grasses, dwarf shrubs (C. vulgaris) and mosses, and also some herbs (see also Streitberger et al., 2021a). Such conditions are known to facilitate warming close to the ground much better than dense vegetation (Stoutjesdijk & Barkman, 1992). The differences in shading preference of ground beetle communities between EARLY/RESTORED and LATE/WIND confirm these conclusions. Overall, the carabid beetle assemblages of the two latest successional stages consisted mainly of non-threatened habitat generalists and were poorly characterised by indicator species.
Recent studies have shown that heterogeneous habitats with a high phytodiversity foster overall species richness of ground beetles and abundance of large carabid beetles (Wang et al., 2021;Zou et al., 2019). Additionally, warm ambient temperatures are known to favour reproduction and survival of ectotherms, such as carabid beetles, and, hence, increase their abundance (Speight et al., 2008). These findings are consistent with our study. Overall species richness and carabid beetle biomass were highest in RESTORED, the most heterogeneous and also relatively warm successional stage; for species richness, however, the difference was not significant. In contrast, WIND, exhibiting a dense vegetation, had the lowest beetle biomass. In addition to providing a cool microclimate, dense vegetation is known to act as a spatial barrier, thus impairing carabid mobility and resulting in lower catch numbers (cf. Ekschmitt et al., 1997;Thiele, 1977).

IMPLICATIONS FOR CONSERVATION
As shown by our study, the conducted rejuvenation and restoration measures are an indispensable tool to maintain montane heathlands with their specialised carabid beetle assemblages in the long run.
Montane heathlands rejuvenated by sod cutting provide warm and bare-ground-rich microhabitatssuitable for threatened carabid beetles typical of heathlandsfor at least 10-20 years. These conditions are also attained on ski pistes through the regular disturbance by skiing in winter and mulching in late summer.
Choppering is another, albeit less intensive, way to rejuvenate heathlands Streitberger et al., 2021a). It involves the removal of biomass and the organic layer but not of the mineral soil. Hence, in contrast to sod cutting, choppering is less effective in favouring early-successional plant and arthropod species Streitberger et al., 2021a). However, in choppered montane heathlands, Calluna vulgaris rapidly re-establishes from the seed bank and the two Vaccinium species can re-sprout from the rhizomes with their ericoid mycorrhizas.
Restored heathlands, about 10 years after the transfer of autochthonous seed material (e.g. by the application of chopper material and hydroseeding), were still characterised by a high beetle biomass and represented an important habitat for threatened and heathland carabid beetles. However, both Vaccinium species had hardly established (Streitberger et al., 2021a), since they poorly regenerate from seeds (Borchard et al., 2017;Streitberger et al., 2021b).
As a result, we recommend continuing the enlargement of existing montane heathlands  through the restoration measures applied in this study. Despite the poor establishment of the two Vaccinium species, we do not suggest further measures that may foster the colonisation of restored heathlands by these species (Streitberger et al., 2021a). On the one hand, such methods are often expensive (e.g. the translocation of sods) (Pywell et al., 2011), and on the other hand, both dwarf-shrub species are still widespread and are often dominant within the heathlands of the study area (Streitberger et al., 2021a).
Concerning heathland rejuvenation, sod cutting is clearly the most effective way to foster the threatened assemblages of early-successional heathlands (see also Streitberger et al., 2021aStreitberger et al., , 2021b. However, it produces a lot of waste material resulting in high costs for its disposal (Streitberger et al., 2021a). Consequently, choppering should also be applied for heathland regeneration . A positive side effect of choppering would be the rejuvenation of the Vaccinium species.