One of the primary drivers of biodiversity is climate change (IPBES 2019), while both are intertwined elements of a global environmental crisis. A key solution to address climate change and achieve a sharp decline in emissions is the transition of global energy systems from fossil fuels to renewable energies (Suman 2021; Osman et al. 2023). By 2050, over one-third (35%) of the global electricity demands needs to be covered by onshore and offshore wind energy to achieve the Paris Climate Goals (United Nations 2015; IRENA 2019). However, addressing one environmental issue can inadvertently worsen another (Gasparatos et al. 2021). The effects of wind turbines on a wide range of animal taxa have been shown through literature reviews. Beyond the risk of collisions and barotrauma, wildlife may also suffer from habitat loss or alteration as well as disturbances due to the construction of turbines, associated infrastructure, noise, light pollution and increased human presence (Kuvlesky et al. 2007; Dai et al. 2014; Perrow 2017; Schöll and Nopp-Mayr 2021; Sander et al. 2024). While impacts on birds and bats are comparingly well studied (Laranjeiro et al. 2018; Sander et al. 2024), there is still a lack of studies of potential effects on non-volant mammals (Helldin et al. 2017).

One species of concern is the hazel dormouse (Muscardinus avellanarius), a small arboreal mammal native to many parts of Europe (Juškaitis 2014). Listed in Annex IV of the European Habitats Directive, the hazel dormouse must be carefully considered in any development project (Bright et al. 2006). The growing demand for wind energy (IRENA 2019; Global Wind Energy Council 2025) and technical progress regarding turbine heights have led to wind turbines being increasingly planned and constructed in or close to forested areas (Enevoldsen 2016; Bunzel et al. 2019). In Germany > 2.500 wind turbines (~ 9% of all wind turbines), were situated in forested area by the end of 2024 (Bredemann and Quentin 2025). Each of those wind turbines built in 2024 in forested area required a median of 0.56 ha permanently cleared, with additional 0.76 ha temporarily cleared for storage and assembling. Higher wind turbines and increased rotor size led to the cleared area per turbine doubling between 2010 and 2024 (Bredemann and Quentin 2025). With the amount and height of wind turbines increasing, also the potential impacts on hazel dormice receive increased attention (Büchner et al. 2017). In the natural forest cycle, the hazel dormouse occupies a position at the beginning of the decay phase and the subsequent regeneration of the forest (Sozio et al. 2016). In commercial forests, re-vegetated clear-cut areas, forest edges and young afforestation with a high diversity of tree and shrub species, the species is particularly well-suited (Mortensen et al. 2022; Juškaitis 2024). While the hazel dormouse avoids coniferous forests at lower altitudes, it also colonises pure spruce stands at higher altitudes in the low mountain ranges (Büchner et al. 2017).

Hazel dormouse populations occur in low densities compared to other small mammals (Juškaitis and Büchner 2013). The mean density of adult individuals is approximately one–two per hectare (Juškaitis 2014). Radio-tracking studies over 1 to 13 days have shown hazel dormice to have small home ranges of 0.3 to 0.5 ha (Bright and Morris 1996Goodwin et al. 2018; Mortensen et al. 2022). However, over the course of a year, the foraging areas of individual hazel dormice are approximately 1 ha in size, with different parts exploited seasonally, depending on food availability (Bright and Morris 1996). However, this size is essentially contingent on population density. A reduction in population density is associated with an increase in home range size (Juškaitis et al. 2020). Natural population dynamics in hazel dormice are characterised by relatively slight interannual fluctuations. When observed over decades, the numbers between stable high and stable low periods of abundance can vary greatly (Juškaitis and Mortensen 2025). In an example from Lithuania, when population density declined by more than half, there was a notable shift in the size of the home ranges as home range sizes increased for both males and females. Additionally, the overlap of home ranges decreased for both males and females (Juškaitis et al. 2020).

The complexity of the observations contrasts sharply with the simplified approach currently used in Germany for implementing protection measures for the hazel dormouse. Although two guidelines on species protection in construction projects have been published in recent years (Büchner et al. 2017; LLUR 2018), integrating these into practical applications remains challenging. The prevailing assumption in German construction projects is that clearing one hectare of forest results in the complete loss of one hazel dormouse’s home range.

This paper seeks to challenge and refine this assumption by exploring the following key questions: How do clear-cuts of different shapes and sizes affect the home ranges of hazel dormice in forest habitats? How many home ranges are affected by a single clear-cut? Is there a difference between male and female hazel dormice? The objective of this study was to assess whether the current “one hectare equals one home range” approach is a valid framework or if it needs to be reconsidered.

The collection of concrete data on the size, location and distribution of dormouse home ranges is challenging and, as a result, such data have rarely been published. One of the few examples is that of a long-term study conducted in Lithuania (Juškaitis et al. 2020). The study site has a size of 60 ha and is located in a large forest area of about 3000 ha in the south-west of Lithuania (55°03'15"N, 23°04'12"E). The forest is of medium age (60–70 years) and consists of various mixed tree stands. Either birch species (Betula pendula and B. pubescens) mixed with Norway spruce (Picea abies) or spruce mixed with deciduous trees dominate the study area. Hazel (Corylus avellana) and glossy buckthorn (Frangula alnus) are most common in the understorey (Juškaitis 2014).

In the study site, 272 nestboxes were placed in a grid system at 50 m intervals between boxes (nestbox density of 4 boxes/ha). The nestboxes were checked during daytime twice a month from April until October. All dormice caught were marked with aluminium rings with individual numbers (for details, see Juškaitis et al. (2020)). Additionally, in this study, the nestbox data from Juškaitis et al. (2020) was used to illustrate the size, location and distribution of the hazel dormouse home ranges. Two seasons were chosen, which represent typical numbers of high (2010: 1.4 adults/ha) and low (2013: 0.7 adults/ha) density for the study site (Juškaitis et al. 2020). The size and location of the home ranges of the two seasons were digitised using QGIS 3.38.2. (QGIS Development Team 2024) using dormice found in nestboxes. Juškaitis et al. (2020) calculated home ranges of each animal only if it was captured at least six times in one activity season. For this current study, the nestbox-derived home range was calculated for each animal regardless of how many times it was encountered because excluding individuals captured less than six times would underestimate the impact of clear-cuts. The Minimum Convex Polygon method ("Minimum Bounding Geometry" in QGIS) was then used to replicate the hazel dormouse home ranges and a buffer, representing half the distance to the nearest unused nestbox (25 m), was added as done by Juškaitis et al. (2020). The sizes of the digitised home ranges were afterwards calculated in QGIS via the field calculator. To determine to which percentage the home ranges overlap each other per density and per sex, the “Overlap Analysis Tool” in QGIS was used and the percentage of overlapping area calculated from the amount of total home range area and the amount of overlapping area. The Module “Count Polygon Overlap” was used to calculate how many home ranges overlapped with each other.

As management alternatives are not easily quantitatively evaluated because of difficulties of conducting landscape-level experiments, we opted for a theoretical approach using computer simulations (Li et al. 1993). To simulate the influence of clear-cuts (an area where all trees and shrubs are cleared, to build wind turbines, leaving no suitable habitat for hazel dormice), sixteen different scenarios were created, varying in the density of the hazel dormouse population and the shape and size of the clear-cut area. In each scenario, 1000 points corresponding to the centre of the clear-cut were randomly created. The only constraint was that the outer boundary of a simulated clear-cut could not be further out than the furthest nest boxes. Therefore, all simulated clear-cuts were within the grid where hazel dormice occurred. Each of the following scenarios was created once for the distribution of hazel dormice in low density and once for high density and in the shape of a square and a strip: one hectare (100 × 100 m and 50 × 200 m), two hectares (141.42 × 141.42 m and 70.71 × 282.84 m), four hectares (200 × 200 m and 100 × 400 m) and six hectares (244.95 × 244.95 m and 122.47 × 489.88 m) (Fig. 1). In the scenarios where strips were created, 500 strips were simulated vertically and 500 horizontally and the two layers then merged into one containing again 1000 simulations.

For each scenario, the 1000 simulated clear-cuts were intersected with the existing home ranges. One clear-cut could overlap with several home ranges and, each time it was calculated in QGIS, how much of the area of a home range was overlapped by a clear-cut. The hectares that did not overlap with any home ranges were included with zero overlap.

The data were further analysed using R Statistical Software (v.4.5.2; R Core Team 2025). All following analyses were performed separately for high and low densities. To analyse the number of affected home ranges, a threshold of 5% was defined and only home ranges with an affected area exceeding this threshold were included. While 0% overlap would represent the best scenario for the species, it would make development measures nearly impossible. The medians and interquartile ranges of the number of affected home ranges per scenario were visualised as boxplots. Afterwards, generalised linear models (GLM) using Poisson distribution were applied on the high- and low-density datasets. The number of home ranges affected was the response variable, while the predictor variables included size and shape of the simulated clear-cut as well as the sex of the dormouse whose home range was affected (affected_hr ~ (size + shape) × sex). Using the “emmeans” package (Lenth and Piaskowski 2026), pairwise contrasts for the Poisson GLMs were computed on the link (log) scale and then reported on the response scale as rate ratios (RRs). In the results section, the RRs are displayed as target/reference (e.g. 6 ha/1 ha, strip/square) and percentage changes are expressed using (RR-1) × 100. Due to the high number of simulations (1000 per scenario), most of the p-values are very low (p < 0.0001). The presentation of the results will, therefore, focus on the effect sizes and only report p-values if they differ from p < 0.0001.

The percentages to which an individual home range area was overlapped by a clear-cut were visualised and the influence of the different parameters tested with a second GLM. The response variable overlap_pc represented the proportion to which the home ranges were overlapped by the simulated clear-cuts and was scaled to the range [0,1]. The predictors included size and shape of the clear-cut scenario and the sex of the hazel dormouse whose home range was affected (overlap_pc ~ (size + shape) × sex). Due to the nature of the response variable, a beta regression was used with a logit link function. The GLM was calculated separately for high density and for low density data. Following that also for the beta regressions, the function “emmeans” (Lenth and Piaskowski 2026) was used to compare the different levels of size and shape of the clear-cut and the sexes of the hazel dormice. The results are reported as absolute percentage differences between each two compared levels. As in the analysis of affected home ranges, also in this analysis p-values will only be displayed if they differ from p < 0.0001.

In the season with low density, 45 hazel dormice were recorded. The 16 females had an average home range of 0.7 ha (min: 0.2, max: 2.2) as did the 29 males (min: 0.2, max: 5.5). The female home ranges overlapped each other by 3.6%, the individual male home ranges overlapped by 8.7%. When looking at the home ranges of the total population, male and female combined, 26.9% of the home ranges overlapped. In 40 cases, two home ranges overlapped, with a median overlapping area of 0.19 ha (min: 0.0, max: 1.13). In 11 cases, three home ranges overlapped by an average of 0.11 ha (min: 0.0, max: 0.27).

At high density, 92 individuals were recorded. The 44 adult females had an average home range of 0.5 ha (min: 0.2, max: 1.6) and the 48 males of 0.9 ha (min: 0.2, max: 3.4). The individual home ranges of females at high density overlapped by 7%, the ones of the males by 25.2% and, for both sexes together, the overlap was at 52.1%. When taking the whole season into account at high density, 75 times two home ranges overlapped, 41 times three home ranges overlapped and 12 times four home ranges overlapped each other. The median overlapping area when two home ranges overlapped was 0.15 ha (min: 0.0, max: 0.70), with three home ranges, it was 0.11 ha (min: 0.0, max: 0.54) and for four home ranges, the median overlapping area was 0.09 ha (min: 0.0, max: 0.21).

The median number of affected home ranges per clear-cut increased with clear-cut size and were higher at high densities. At low density, medians rose from two affected home ranges at 1 ha to six at 6 ha (Fig. 2). This would be an increase from about 4–13% of the population. The Poisson GLM showed that the shape of the clear-cut did not make a significant difference to the number of affected home ranges (p = 0.908). Neither did increasing the clear-cut size from 1 to 2 ha (p = 0.245). Statistically significant were the differences of clear-cuts the size of 4 and 6 ha, each compared to 1 ha and the difference between the sexes (p = 0.003).

The further analysis of the significant pairwise contrasts of the low density data-set revealed that, for female home ranges, a rise in the clear-cut size from 1 ha to 2 ha showed no significant difference (rate ratio, RR 2/1 = 0.96, p = 0.65). A clear-cut of 4 ha was associated with affecting 17% (RR 4/1 = 1.17) more home ranges than a clear-cut of 1 ha, while a clear-cut of 6 ha would affect 46% (RR 6/1 = 1.46) more home ranges. This difference between the sizes was stronger for males. Between 2 ha and 1 ha, 19% (RR 2/1 = 1.19) more home ranges would be affected, 4 ha compared to 1 ha would affect 72% (RR 4/1 = 1.72) more home ranges and 6 ha would affect 125% (RR 6/1 = 2.25) more home ranges than clear-cuts of 1 ha. Across all clear-cut sizes, more male home ranges were affected than female home ranges, with the sex difference widening at larger clear-cut sizes. By square-shaped clear-cuts of 1 ha, 12% (RR m/f = 1.12, p = 0.003) more male home ranges were affected than female home ranges, increasing to 73% (RR m/f = 1.73) at 6 ha. For strip-shaped clear-cuts, the difference increased from 16% (RR m/f = 1.16) more male home ranges affected by a clear-cut of 1 ha to 79% (RR m/f = 1.79) at 6 ha.

At high density, the medians of affected home ranges rose from four at 1 ha to 12–14 at 6 ha (≈ 4–14% of the population) (Fig. 2) and the Poisson GLM showed that all sizes compared to 1 ha were statistically significantly different. The differences between the two clear-cut shapes were also statistically significant, with strip-shaped clear-cuts consistently affecting 1–2 home ranges more than squares. Additionally, as in low density, also sexes were affected statistically differently. The pairwise contrast calculated with emmeans showed that the number of affected home ranges increased sharply with the size of the simulated clear-cut area. For female home ranges, 2 ha vs. 1 ha clear-cuts corresponded to approximately a 42% increase (RR 2/1 ≈ 1.42), 4 ha vs. 1 ha to about + 136% (RR 4/1 ≈ 2.36) and 6 ha vs. 1 ha to about +217% (RR 6/1 ≈ 3.17), with closely comparable increases in males (i.e. RR 2/1 ≈ 1.43, RR 4/1 ≈ 1.58, RR 6/1 ≈ 3.06). In contrast to the low-density simulations where shape made no significant difference, strip-shaped clear-cuts at high density affected significantly more home range than square-shaped clear-cuts. Females home ranges were 9% more affected by strip-shaped clear-cuts (RR square/strip ≈ 0.91) than by square-shaped ones, while for male home ranges, the effect of strip-shaped clear-cuts was 12% higher compared to square ones (RR square/strip ≈ 0.88). Across all sizes and both shapes of clear-cuts, more male home ranges were affected than female home ranges. The RRs m/f ranged from 0.82 to 0.89 implying 12–22% higher affectedness in males.

The GLMs with the beta regressions and the afterwards calculated emmeans show that the percentages to which an area is overlapped by a clear-cut increase with the size of the clear-cut for both sexes and both shapes. At low density, medians of 12.2% (1 ha) and of 33.0% (6 ha) were affected by square-shaped clear-cuts versus 12.2% and 30.4% by strip shaped clear-cuts. At high density, square shaped clear-cuts affected medians of 17.7% (1 ha) and 46.0% (6 ha) versus 16.8% and 39.3% by strip-shaped clear-cuts. In both densities, the interquartile range and the presence of outliers increased with the size of the clear-cut, particularly for square-shaped clear-cuts, indicating a greater variability in their impact (Fig. 3).

Female home ranges are 6.7% more overlapped by clear-cuts of 2 ha, than of 1 ha. Four ha overlap 15.7% more than 1 ha and 6 ha overlap 24.3% more. For strip-shaped clear-cuts, the numbers are similar: 2, 4 and 6 ha overlap in comparison to 1 ha by 6.2%, 14.9% and 23.5% more, respectively. For male home ranges, the increases were smaller, but still highly significant: 3.9%, 9.8% and 14.5% on squares and 3.7%, 9.4% and 13.9% on strips. At every clear-cut size and for both sexes, square-shaped clear-cuts affected higher percentages of the home ranges than strip-shaped clear-cuts. For female home ranges, square-shaped clear-cuts overlapped 4.6–5.4% more and, in males, 2.4–2.9%. Female home ranges were to a higher percentage overlapped than male ones at all sizes and by both shapes. For squares, the differences were 5.0%, 7.7%, 10.8% and 14.7% at 1, 2, 4 and 6 ha and, on strips, 2.7%, 5.2%, 8.2% and 12.3%.

At high density, size-related increases were larger than at low density. For females, moving from 1 to 2, 4 and 6 ha raised overlap by 7.9%, 19.5% and 25.3% on squares and by 7.6%, 19.3% and 25.5% on strips, respectively. For males, the corresponding increases were 5.0%, 11.8% and 17.2% on squares and 4.8%, 11.4% and 16.7% on strips. Square-shaped clear-cuts consistently exceeded strips by roughly 6.5–7.1% for female and 3.7–4.2% for male home ranges across sizes. Females again had higher overlap than males at every size and shape, with differences from 7.9% to 15.9% on squares and from 4.8% to 13.6% on strips as the clear-cut size increased.

The conducted simulations strongly challenge the accuracy of the one hectare equals one home range approach for management decisions. At higher population density, the individual home ranges were smaller (Juškaitis et al. 2020) and half of the total home range area consisted of areas where two to four home ranges overlapped. With decreasing population density, the individual home range increased and overlapped less than at high density. Still more than one quarter of the total home range area overlapped at low density. When looking at the number of affected home ranges, this means, at low density, a median of two home ranges is affected by a clear-cut of one hectare, while at high density, it is a median of four home ranges. The results showed that a maximum of twelve home ranges could have been affected by a square clear-cut of one hectare at high density.

The results furthermore show that both the number of home ranges affected by a clear-cut as well as the fraction of each home range area overlapped depend strongly on clear-cut size and population density. Across densities, larger clear-cuts affect more home ranges and overlap a larger share of any given home range. Density amplifies these effects: at high density, the total number of affected home ranges and the percentage overlapped increased much more steeply with clear-cut size than at low density. Typically, the interannual fluctuations in hazel dormice populations are rather slight, but can vary more greatly in the long term. In our example, 2010 was the last year in which the population had stable high abundance numbers, before the number of individuals decreased and hit a stable low phase from 2018–2024 with a minimum of 0.4 individuals per hectare (Juškaitis and Mortensen 2025). In the present study, in both densities (high in 2010 and low in 2013) about 4% of all home ranges of the population were affected by a clear-cut the size of 1 ha. However, the total number of home ranges affected is twice as high at high density.

Clear-cuts can have several impacts on small mammal populations. One of them is death as a direct cause of the felling operations due to being, for example, crushed by vehicles or felled trees (Escobar et al. 2015). In a study carried out by Goodwin et al. (2018), the majority of hazel dormice did not flee when chainsaws were used in their proximity for small-scale felling. While this might differ at larger scale disturbances, it indicates the risk of hazel dormice not seeking refuge. Especially high would be the risk of direct death if the felling happens during winter times when hazel dormice are in torpor (Trout et al. 2012; Büchner et al. 2017; Goodwin et al. 2018) or between May and October, when juveniles are to be expected (Büchner et al. 2017) and breeding females can abandon their litters because of disturbance during tree felling and wood transportation (Juškaitis 2008).

If a hazel dormouse survives the felling and construction process, a part of their home range is gone and it must move to adjacent suitable habitats, if enough suitable habitat is available in the surrounding area. If the surrounding habitats are occupied by other dormice, which is very likely at high densities or the surrounding habitats are not suitable, the hazel dormouse cannot establish a new home range. Small-scale felling and thinning generally have the potential to improve hazel dormouse habitats by creating stages of succession that form highly suitable habitats after a few years (Trout et al. 2012; Sozio et al. 2016; Mortensen et al. 2022; Juškaitis 2024). Additionally, forest edges created around clear-cuts could be suitable habitats for hazel dormice after a few years. In the case of clear-cuts for wind turbines, however, the positive effects of succession stages are only true for the area that is temporarily affected; the rest is permanently lost to the dormice. The suitability of edges is also likely to depend on the age, size and orientation of the clear-cut, as has been demonstrated for birds (Brazaitis et al. 2005; Villard et al. 2007).

Due to areas permanently lost, the size of the forest is also very relevant. Generally, a minimum size of 20 ha is seen as a prerequisite for a stable and long-term hazel dormouse population (Bright and Morris 1994; Keckel et al. 2012). Even high density populations face an extinction risk if the population size is small due to a small available patch size (Mortelliti et al. 2014). Clear-cuts and larger roads both can act as barriers between individuals or between an individual and its resources (Büchner et al. 2017). Both males and females become more mobile at lower population densities and increase their home range sizes to a certain extent. It is important that they are still able to encounter each other for mating (Juškaitis et al. 2020), as otherwise population recruitment will lower.

Sex-specific effects were in this study consistent through densities and clear-cut sizes. Typically, females have smaller home ranges than males (overview in Juškaitis (2014)) which was also true in this study. Therefore, less female than male home ranges are affected by clear-cuts of all sizes and both shapes. On the other hand, the percentage of an individual home range area that would be overlapped by a clear-cut is significantly higher for females, which is an important ecological factor. Female hazel dormice are critical for the population stability. If the number of adult females falls below a critical threshold, for example, through increased mortality due to clear-cuts, it can lead to low numbers of breeding events and low population recruitment in autumn (Juškaitis and Mortensen 2025).

Given these dynamics, habitats should not be reduced below 20 hectares and generally clear-cuts should be kept as small as possible. While the shape of the clear-cut will be dependent on the surroundings, it should be kept in mind that square-shaped clear-cuts affect a higher percentage of the individual home range. This increases the risk of direct death during the felling operations and would especially put females at higher risk. Directional felling could be used to try and guide hazel dormice towards intact, suitable habitat (Trout et al. 2012). Most importantly, the area where clear-cuts are planned should be thoroughly checked for hazel dormice and timing of the felling should be adjusted to avoid torpor and peak breeding phases. We highly encourage to accompany clear-cuts that are being executed by scientific monitoring projects with a BACI (Before-After, Control-Impact) design (Smith 2014) and to investigate the actual impact of clear-cuts on home ranges in the field.

The methods used in the presented study have limitations. The used home range sizes are likely underestimated due to two reasons: i) Nest-box monitoring is a reliable detection method for hazel dormice throughout one season as well as over generations (overview of survey methods in Büchner et al. (2017)) and represents their home ranges well (Bright and Morris 1991). Nevertheless, hazel dormice can also use boxes and the surrounding area without being detected; ii) All detected individuals were included, regardless of their capture frequency. Underestimating the home range sizes leads to overestimating the percentage overlap, while the number of affected home ranges is understated. Moreover the 5% that were used as a threshold to define a home range as affected by a clear-cut are an operational choice to prevent the results from being skewed by numerous negligible intersections. No literature was found to which extent a home range could be cleared without disturbing the dormouse significantly. Finally, our simulations also did not account for habitat fragmentation by new or upgraded roads, built for construction and maintenance purposes (Perrow 2017) additionally to the here estimated habitat loss. These limitations likely make the estimates conservative regarding the number of home ranges affected and the effects of clear-cuts are probably even greater.

Our findings showed that, across densities and clear-cut shapes, a 1 ha clear-cut typically intersects not one, but multiple home ranges, especially at higher population densities. Therefore, the currently used approach underestimates the actual effects of habitat loss on hazel dormouse populations.