Introduction

Microclimates in forests are determined by canopy cover (De Frenne et al. 2019; Zellweger et al. 2020) and strongly impact both biodiversity and ecosystem processes (Penone et al. 2019). The frequency of forest disturbances is expected to increase, leading to a decrease in canopy cover (Seidl et al. 2017; Senf et al. 2018) and thus to changes in forest microclimates (Thom et al. 2020). In the current era of climate change, the role of microclimatic conditions, and thus of canopy cover, in maintaining biodiversity has been increasingly recognized (De Frenne et al. 2021). Forest disturbances that create high canopy openness are accompanied by the accumulation of deadwood (Bouget and Duelli 2004), which in forest ecosystems supports a large number of species, including deadwood-dependent (saproxylic) beetles, a species-rich and functionally important taxonomic group (Grove 2002; Seibold et al. 2021). Assemblages of saproxylic beetles depend on numerous drivers associated with their habitat, such as forest microclimate, tree species, and object size and decomposition stage (Stokland et al. 2012; Seibold et al. 2015; Ulyshen 2018; Vogel et al. 2020a).

Forest stands with high canopy openness were commonly observed to host more saproxylic beetle species than stands with low canopy openness and the respective assemblages differ in their species composition (Vodka et al. 2008; Bouget et al. 2013; Seibold et al. 2015, 2016; Müller et al. 2015; Heidrich et al. 2020). The reasons for this difference are unclear, but four potential mechanisms have been proposed: (1) the activity of saproxylic beetles correlates with temperature (Taylor 1963; Liu et al. 1995). Due to increased beetle activity, more individuals and thus more species will be found in sunny than in shady forest stands when activity traps, such as flight-interception traps, are used. (2) High temperatures may allow for quicker insect development, leading to higher densities and ultimately more species per sun-exposed deadwood volume. This corresponds to the “more individuals” hypothesis (Clarke and Gaston 2006) and in the context of saproxylic beetles has been termed the “temperature-dead wood compensation” hypothesis (Müller et al. 2015). (3) Sun exposure leads to a higher temperature variability within and between deadwood logs, such that sunny forests offer a larger number of microclimatic niches. Considering that saproxylic beetle species differ in their microclimatic preferences, a higher microclimatic heterogeneity might promote a higher species richness in sun-exposed deadwood logs, and thus in sunny than in shady forests (Seibold et al. 2016; Heidrich et al. 2020). (4) Finally, other environmental factors characteristic of sunny areas, such as the higher cover and diversity of herbaceous plants and bloomers, cause a higher richness of saproxylic beetles but without necessarily increasing their abundance (Haddad et al. 2001).

Studies of beetles in forest stands are facilitated by the use of flight-interception traps, which sample the habitat of interest in a circle of about 30–40 m (Leutner 2018); their success depends on the flight activity of the insects (Birkemoe and Sverdrup-Thygeson 2015). In a previous study, Seibold et al. (2016) found that canopy openness had a positive effect on beetle abundance and a marginally positive effect on beetle richness (accounting for abundance), as determined from samples obtained in flight-interception traps. However, which of the potential above-described mechanisms best explained the observed pattern could not be determined. In contrast to flight-interception traps, stem-emergence traps are largely independent of beetle activity and reflect the habitat conditions and beetle densities of single objects (Vodka et al. 2008; Bouget et al. 2011; Müller et al. 2015; Seibold et al. 2018; Hagge et al. 2019; Vogel et al. 2020b). Consistent with studies based on flight-interception traps, their use revealed higher species number of saproxylic beetle in sun-exposed than in shaded deadwood (Vodka et al. 2008; Müller et al. 2015, 2020; Gossner et al. 2016; Vogel et al. 2020b), suggesting that mechanism (1), i.e., higher flight activity, is not the driving mechanism. The lower diversity of saproxylic beetles in artificially shaded logs placed adjacent to sun-exposed logs in herb-rich forest openings (Vogel et al. 2020b) indicates that the higher beetle numbers in sunny forests are also not explained by mechanism (4), i.e., a richer herb layer. This leads to the conclusion that the higher species numbers of saproxylic beetles in sunny forests are due either to mechanism (2), i.e., fewer required resources (the “more-individuals” and “temperature-dead wood compensation” hypotheses) or to mechanism (3), i.e., a higher intrinsic habitat heterogeneity of deadwood.

The role of microclimate in beetle diversity at the scale of forest stands and deadwood objects was investigated in this study as part of an experiment set up in autumn 2011, in which 800 m3 of fresh deadwood logs of two tree species (Fagus sylvatica L. and Abies alba Mill.) were placed on plots located either in open, and therefore, sunny, forest stands or in shady forest stands with a closed canopy cover (Seibold et al. 2016). For the present work, flight-interception traps were used to determine the activity densities of beetles at the stand scale, and stem-emergence traps to obtain measures of beetle density in deadwood logs and thus at the deadwood-object scale. Deadwood surface temperature, as a proxy for microclimatic heterogeneity, was measured to test whether the habitat heterogeneity associated with microclimatic conditions was higher in sunny than in shady plots. The following hypotheses were tested:

  1. 1.

    Microclimatic heterogeneity within deadwood logs and between deadwood logs is higher in sunny forests than in shady forests.

  2. 2.

    Beetle abundance is higher in sunny forests compared to shady forests at both scales.

  3. 3.

    Beetle diversity is higher in sunny forests compared to shady forests at both scales.

  4. 4.

    Higher beetle diversity in sunny forests than in shady forests is linked to higher microclimatic heterogeneity.

Materials and methods

Study area and experimental design

This study was part of a larger experiment conducted in the management zone of the Bavarian Forest National Park, in south-eastern Germany (Seibold et al. 2016). The forest in the study region is montane-mixed forest dominated by Norway spruce (Picea abies L.), European beech (Fagus sylvatica L.), and silver fir (Abies alba Mill.). The experiment was initiated in autumn 2011. Logs of European beech or silver fir (coarse woody debris; diameter 25–35 cm, length 5 m), branches (fine woody debris; diameter 3–5 cm, length 2–3 m), or both were deposited on 190 plots of 0.1 ha (Seibold et al. 2016). Our study used data from a subset of 60 plots which contained only logs which allowed the use of stem-emergence traps (see below). To sample different microclimatic conditions, half of the plots were located under a closed canopy with shady microclimatic conditions, and half in forest gaps with sunny microclimatic conditions. Coarse woody debris was deposited in either low amounts of four logs (about 10 m3 ha−1) or high amounts of 40 logs (about 100 m3 ha−1) per plot and comprised wood of either one or two tree species. The plots comprised six different treatments representing independent gradients of deadwood amount and diversity, each replicated five times on sunny and shady plots and arranged in a random block design: (1) low amount of beech logs, (2) low amount of fir logs, (3) low amount of beech and fir logs, (4) high amount of beech logs, (5) high amount of fir logs, and (6) high amount of beech and fir logs. Deadwood diversity was considered low when the logs were from the same species and high when the logs were from different species.

Beetle sampling

On every plot, two flight-interception traps were installed to allow sampling at the stand scale. In addition, one stem-emergence trap was installed on a plot when one tree species was present and two traps when two tree species were present. Flight-interception traps were installed in the spring of 2012, and stem-emergence traps in the spring of 2013. The delay in the latter allowed the logs to be left undisturbed for the first year and thus colonized by insects without interference. Arthropod sampling from a total of 120 flight-interception traps was conducted from May to August 2012–2014, and from a total of 80 stem-emergence traps from May to August 2013–2017. The longer sampling period of the stem-emergence traps was necessary to control for the time-displaced catch of those saproxylic beetle species with a larval development period of 2 or 3 years. The stem-emergence traps were moved yearly to a new section of the same deadwood log to avoid disturbance of the colonization process. The traps cover an area of ≈ 0.35 m2 and their samples represent emerging arthropods as accurately as achieved by in-situ rearing from experimental logs (Hagge et al. 2019). Sampled beetles were determined to the species level (by Boris Büche and Alexander Szallies), and strictly saproxylic species were selected for further analyses (Schmidl and Bussler 2004). To make the sampling methods comparable, only species from the flight-interception traps that were also present in the stem-emergence traps were included in the analyses. This ensured the exclusion of species that originated from deadwood outside the plots (Fig. S1, Table S3).

Temperature measurements

The temperature variability of a single deadwood log, and thus at the object scale, was assessed by measuring the wood surface temperature of the 80 logs sampled with the stem-emergence traps at 17 fixed positions using an infrared thermometer (Fig. S2a). Temperature variability between logs, reflecting the temperature variability at the stand scale, was accounted for by measuring the temperature of the top side of all 1320 deadwood logs present in all 60 plots (Fig. S2b). Measurements were carried out on five sunny days in August 2018, completing one block per day. Large logs buffer internal temperature fluctuations more effectively, whereas for medium-sized and smaller logs, used in our study, temperature variability inside logs and on the wood surface is similar (Brackebusch 1975; Pouska et al. 2016).

Statistical analyses

All statistical analyses were conducted in R 3.6.3 (www.r-project.org).

To test our first hypothesis, wood-surface temperature was analyzed as a proxy for microclimatic heterogeneity in deadwood. First, microclimatic heterogeneity within single logs was estimated by calculating the standard deviation of the 17 temperature measurement points on each of the 80 logs. Second, microclimatic heterogeneity between logs of a plot was determined by calculating the standard deviation of the temperature on the top face of the middle of each of the logs located on one plot. Variations in temperature within a log were investigated by fitting linear mixed-effect models with Gaussian errors to test the effect of canopy openness (sunny vs. shady forest) on the mean temperature and standard deviation of the temperature within each log. Due to the correlation between the mean temperature and the standard deviation within single logs, the mean temperature was included as an additional fixed effect in the temperature standard deviation model. Both models included the measurement day and the plot as random effects. The temperature variation between all logs was compared using a linear mixed model with Gaussian errors, with the standard deviation of the temperature between logs on a plot as a response variable, and by fitting canopy openness (sunny or shady forest) as a fixed effect. The model included the measurement day as a random effect.

The second and third hypotheses were tested by analyzing different diversity measures for saproxylic beetles. The abundance (hypothesis 2) and alpha-, beta-, and gamma-diversity (hypothesis 3) of saproxylic beetles were used to analyze the differences between sunny and shady forests at the forest stand and deadwood-object scales. The differences between the two sampling methods, representing the forest stand scale and the deadwood-object scale, were considered following the method of Chao and Jost (2012), in which sample completeness is used to standardize the diversity among different assemblages. Species number was estimated by applying the function estimateD from the iNEXT package (Chao and Jost 2012; Chao et al. 2014). This function computes species diversity by considering Hill numbers with a particular specified level of sample coverage and thus accounts for different trapping methods and allows the relative diversities of multiple assemblages to be determined correctly (Jost 2010; Alroy 2010; Chao and Jost 2012). Due to the extremes in the distribution patterns of emerging saproxylic beetles (particularly bark beetle species [Scolytinae]) from logs; Fig. S1) and the lower sample coverage of stem-emergence traps (Fig. S6), the best estimated number of species obtained with the estimateD function was for a sample coverage of 60%. Therefore, all species number estimations based on the stem-emergence and flight-interception traps were set to a sample coverage of 60%. The effects of canopy openness, and experimentally manipulated deadwood amount, and deadwood diversity on the stand scale diversity of saproxylic beetles were analyzed by fitting generalized linear mixed-effect models using the lme4 package (Bates et al. 2015). A Poisson error distribution was used for the abundance and estimated number of species (i.e., alpha diversity) of saproxylic beetles. For all flight-interception traps, the fixed effects of canopy openness and the interactions of canopy openness, deadwood amount, and deadwood diversity were fitted. Object-scale models included the fixed effects of canopy openness and the interactions of canopy openness, tree species (i.e., the species of the deadwood log hosting the stem-emergence trap), deadwood amount, and deadwood diversity. Abundance and the estimated number of species were fitted as response variable for both scales. All models included the random effects of block, plot, and trap to account for the nested design. Potential overdispersion in the models was taken into account by fitting an additional observation-specific random intercept (Elston et al. 2001) (see Supplement Information Appendix 7 for all model equations). Furthermore, we correlated the abundance and the estimated number of species at the stand scale with the abundance and the estimated number of species at the object scale using Spearman’s rank correlation.

The gamma-diversity of saproxylic beetles sampled with the flight-interception and stem-emergence traps in sunny and shady forest plots was investigated using a sample-based rarefaction-extrapolation approach that considered the rate of increase in species diversity with an increasing number of traps. The calculations were performed using the iNEXT package (Chao et al. 2014) for the three components of species diversity represented by the Hill series (q = 0, species richness weighting all species equally; q = 1, exponential of Shannon’s entropy index weighting typical species; q = 2, inverse of Simpson’s concentration index weighting dominant species).

Beta diversity, i.e., the compositional dissimilarity of emerging saproxylic beetle assemblages between the logs of the sampled plots for the two levels of canopy openness (sunny and shady), was described based on the estimation of Jaccard dissimilarity indices, by applying the function vegdist in the add-on package vegan (Oksanen et al. 2020) to the presence–absence of data from emerging saproxylic beetles. A linear model was fitted in which distance was the response and canopy the explanatory variable. For this analysis, only plots sampled by two stem-emergence traps, such as one on a fir log and one on a beech log (n = 20), were selected.

The fourth hypothesis was tested by correlating the temperature standard deviation between and within deadwood logs with the estimated number of species sampled with flight-interception and stem-emergence traps, using Spearman’s rank correlation.

Results

Temperature differences between sunny and shady forests

The mean object-surface temperature was almost twice as high for sun-exposed logs (mean 29.5 ± 5.8 °C) than for shaded logs (mean 16.2 ± 3.0 °C; Fig. S3a). The variability in the object-surface temperature within single sun-exposed logs (mean standard deviation 8.3 ± 3.2 °C) was seven times higher than that of shaded logs (mean standard deviation 1.2 ± 0.8 °C; Fig. S3b). Temperature variability between all logs per plot was higher for sun-exposed logs (mean standard deviation 5.0 ± 2.5 °C) than for shaded logs (mean standard deviation 1.3 ± 1.0 °C; Fig. S3c).

Diversity patterns of saproxylic beetles in forest stands and deadwood logs

From the flight-interception traps (stand scale), 36,546 individuals representing 398 saproxylic beetle species were collected, and from the stem-emergence traps (object scale) 28,094 individuals representing 256 species. Together, 428 species (64,640 individuals) were represented in total, including 226 species collected by both sampling methods, 172 species collected only by flight-interception traps, and 30 species only by stem-emergence traps (Table S3).

The abundance of saproxylic beetles was higher in sunny than in shady forests at the stand scale (Fig. 1a, Table S1) and lower in sunny than in shady forests at the object scale (Fig. 1b, Table S2). The estimated number of saproxylic beetle species (alpha diversity; Fig. 1c, d, Tables S1, S2), as well as beta- and gamma-diversity (Figs. 2, S4, S5) were consistently higher in sunny forest than in shady forest at both the stand scale and the object scale. This pattern was present along the series of Hill numbers, except for q = 0 (all species weighted equally) at the object scale, in which case there were no differences between sunny und shady forests (Fig. 2). There was no correlation of beetle abundance between the stand and the object scales (Fig. 3a), whereas a positive correlation was found for the estimated number of species (Fig. 3b).

Fig. 1
figure 1

Effects of canopy openness (sunny vs. shady forest) on the a abundance and c estimated number of beetle species at the stand scale, sampled with flight-interception traps, and on the b abundance and d estimated number of beetle species at the object scale, sampled with stem-emergence traps. Note the log scale

Full size image
Fig. 2
figure 2

Sample-based rarefaction (solid lines) and extrapolation (dotted lines, up to twice the actual sample size) of the gamma-diversity of saproxylic beetles at a the stand scale and b the object scale. Species diversity was estimated using Hill numbers: q = 0 (species richness), q = 1 (exponential of Shannon’s entropy index) and q = 2 (inverse of Simpson’s concentration index). Solid symbols represent the total number of reference samples, with n = 60 for the flight-interception traps and n = 40 for the stem-emergence traps

Full size image
Fig. 3
figure 3

Spearman correlation between a stand and object scale abundance and b the stand and object scale estimated number of species. Note the log scale

Full size image

Correlation of temperature variability with species diversity

At the stand scale, the estimated species number increased with the increasing variability in both the object-surface temperature between logs (Fig. 4a) and the object-surface temperature of single logs (Fig. 4c). At the object scale, the increase in the estimated species number with the temperature variability of a log was marginally significant (Fig. 4d), but it was not affected by the temperature variability between the logs of a plot (Fig. 4b).

Fig. 4
figure 4

Spearman correlation between a the stand scale estimated number of species (determined from flight-interception traps) and the temperature variability between all logs on a plot and c temperature variability on single logs. Spearman correlation between the object scale estimated number of species and the temperature variability b between logs and d on single logs. Note the log scale

Full size image

Discussion

Canopy openness determines microclimatic heterogeneity

In line with our first hypothesis, wood-surface temperature measurements revealed higher microclimatic heterogeneity within and between deadwood logs in sunny forests than in shady forests. The larger temperature variability in sunny forests can be explained by the buffering effect of the forest canopy (De Frenne et al. 2019; Zellweger et al. 2020; Thom et al. 2020). With a high canopy openness, light penetration and thus solar radiation are higher but temporarily distributed, which leads to variations in the microclimatic conditions in forest stands (Thom et al. 2020) but also between deadwood objects and within single deadwood objects. Considering that saproxylic beetles differ in their microclimatic preferences (Möller 2009), the higher temperature variation of sunny forests offers a higher habitat heterogeneity.

Canopy openness drives the activity of saproxylic beetles

Our comparison of beetle assemblages in sunny and shady forests revealed higher abundances of saproxylic beetles in sunny than in shady plots at the stand scale using flight traps, but the opposite pattern was found at the deadwood-object scale using stem-emergence traps. The second hypothesis is therefore only true for the forest stand scale. These results are in line with the abundance patterns previously determined at the stand scale (Sverdrup-Thygeson and Birkemoe 2009) but contradict those found at the object scale (Vodka et al. 2008; Müller et al. 2015). Two potential mechanisms explain the higher species numbers in sunny forests, as observed in our study, involve higher abundances. The first attributes higher abundances to higher flight activity, based on data from flight-interception traps (Taylor 1963), while the second predicts higher abundances per unit deadwood volume due to quicker insect development at high temperatures (“temperature-dead wood compensation” hypothesis (Müller et al. 2015). However, the higher abundances in sunny forests measured from the flight-interception traps but not the stem-emergence traps suggest that they were due to higher flight activity, not to a higher density of individuals developing in sun-exposed deadwood logs. Furthermore, the sample coverage of the flight-interception traps was higher in the sunny than in the shady forests (Fig. S6a), a further indication of the activity-driven sampling effect of flight-interception traps (Birkemoe and Sverdrup-Thygeson 2015).

Microclimatic heterogeneity drives the species diversity of saproxylic beetles

The estimated number of species (alpha diversity) was higher in sunny than in shady forests at the stand and the object scale, consistent with our third hypothesis and with the other studies (Vodka et al. 2008; Bouget et al. 2013; Müller et al. 2015; Gossner et al. 2016; Vogel et al. 2020b). Small organisms such as beetles are directly influenced by micro-environmental conditions (De Frenne et al. 2021), including the temperature variability in sunny forests. In our study, temperature variability, and thus the associated microclimatic habitat heterogeneity, correlated positively with beetle diversity at the stand scale. At the object scale, however, the correlation with the estimated number of species was marginally significant only for temperature variability within a log, not between logs. Hence, our fourth hypothesis is only partly supported. Microclimatic habitat heterogeneity may be higher when all logs on a plot are considered, which might explain the stronger correlation at the stand scale. This was further supported by the analysis of gamma-diversity across several logs sampled with stem-emergence traps. Those analyses showed that gamma-diversity was similar or even higher in sun-exposed logs, if typical and dominant species were more heavily weighted. As a summary of the assemblages of several logs, gamma-diversity provides a coarse representation of stand scale estimates based on samples from flight-interception traps. In our study, gamma-diversity at the object scale may be overestimated, as it included logs from different plots. However, the dissimilarity in the assemblage composition of logs of the same plot was higher in sunny than in shady forests. This suggests that the gamma-diversity was also higher at sunny than at shady plots when calculated for logs on the same plot. Summarized, the observed patterns indicate that the higher species diversity of saproxylic beetles at sunny stands is not caused by higher abundance but by higher microclimatic habitat heterogeneity both within and between deadwood logs.

Microclimatic habitat heterogeneity is the product of differences not only in temperature but also in the parameters affected by temperature, such as the composition of fungal assemblages (Pouska et al. 2016) and deadwood decomposition rates (Gossner et al. 2016; Krah et al. 2018). Accordingly, the correlations between temperature and saproxylic beetle diversity will be supported by the direct effects of temperature and by the indirect effects of fungi and deadwood decomposition. Our findings demonstrate the crucial contribution of microclimatic habitat heterogeneity in deadwood objects to the diversity of saproxylic species at the stand scale.

Conclusion

The results of our study provide insights into the reasons for the higher diversity of saproxylic beetles in sunny than in shady forest stands. The higher abundance of saproxylic beetles observed at the stand scale, determined using flight-interception traps, was not caused by a higher abundance per deadwood log but rather by a higher insect activity. Thus, as this finding demonstrates, the data obtained in insect diversity studies using activity traps must be carefully interpreted. Our linkage of coverage-based species diversity estimates to temperature measurements revealed that microclimatic habitat heterogeneity within and between deadwood logs is the major driver of the higher beetle diversities in sunny forests with low canopy cover. Thus, a strategy to conserve and promote saproxylic beetle diversity is to maintain natural sun-exposed deadwood, such as in gaps after natural disturbances or tree-collapse, and to combine active deadwood enrichment with the creation of canopy gaps (Vogel et al. 2020b). These actions should be accompanied by the retention of deadwood also under a closed canopy, to protect the full range of saproxylic species.