Timing and mulching frequency affected the number of nests of cavity-nesting wasps that hunt for aphids in forest meadows

Mulching, cutting of the vegetation without plant biomass removal, is a common alternative to mowing. The aim of our study was to find out if the mulching of forest meadows at different time points affects cavity-nesting bees and wasps. We exposed trap nests for cavity-nesting bees and wasps at 24 forest meadows in south-western Germany over 2 years and applied four experimental mulching treatments with six replicates: (i) mulching in June, (ii) mulching in September, (iii) mulching in June and September, and (iv) no mulching as control. Nests were collected throughout the growing period. The insects were sorted and analyzed according to functional groups. Mulching in June and September reduced the nest number of all cavity-nesting insects in the second but not in the first year. The separation of insects into three functional groups (bees, herbivore-hunting wasps and carnivore-hunting wasps) showed that the number of herbivore-hunting wasp nests was reduced by mulching in September in both years and by mulching in June and September in the second year. Specifically, aphid-hunting wasps were influenced by mulching in September or mulching twice in the second year. Aphid-hunting wasps likely find their larval food in the vegetation of the forest meadows, while the other studied groups likely find their main larval food in the surrounding forests and are therefore not negatively affected. For maintaining the reproductive success of cavity-nesting wasps that hunt for aphids, we recommend mulching once in June rather than mulching in September or twice a year.


Introduction
The biodiversity of meadows is influenced by several factors, for example the timing and frequency of cutting, the amount and type of fertilization and the geographic location (Gilhaus et al. 2017;Rannap et al. 2017;Schuch et al. 2012;Sullivan et al. 2018). Many insects depend on species-rich grasslands (Ebeling et al. 2018;Knops et al. 1999;Schuch et al. 2012). A special type of meadows are forest meadows, characterized by small grassland patches (around 1 ha) surrounded by forests. Forest meadows were e.g., established to increase the food supply for roe deer (Capreolus capreolus) and red deer (Cervus elaphus), for hunting and to protect tree seedlings and saplings in forests from harmful browsing (Erlacher and Völk 2003;Petrak 2003). The focus on game animal control and the exclusion of livestock is reflected by their alternative name "game meadow". Towards the end of nineteenth century, these meadows increased in number and most still persist as managed forest meadows (Völk 1999). Due to the uniformity of the forests surrounding these meadows they provide a suitable study system. Compared to other grasslands, forest meadows are infrequently studied in the scientific literature (but see Aboling 2003;Petrak 2003;Petrak et al. 2015;Tomić et al. 2010). Only Aboling (2003) mentioned game meadows as interesting for game animals and vegetation studies, Petrak (2003) and (2015) gave practical instructions how to provide food for game animals and integrate nature conservation in game meadow management and a study by Tomić et al. (2010) described the plant communities on game meadows in Serbia.
Insects are currently declining in many parts of the world (Hallmann et al. 2021;Sánchez-Bayo and Wyckhuys 2021;Seibold et al. 2019;Wagner et al. 2021). The main explanations for insect decline are the loss of habitat through increasing urbanization, climate change, landscape homogenization, high use of pesticides and fertilization and intensive farming methods for example frequent mowing (IPBES 2019;Dicks et al. 2021;Mupepele et al. 2019;Potts et al. 2010).
Cavity-nesting bees and wasps comprise around 5-10% of all solitary bee and wasp species in different parts of the globe (calculation based on species numbers given in Krombein (1967) and Tscharntke et al. (1998). However, they were shown to be sensitive to habitat changes (Staab et al. 2018;Tscharntke et al. 1998). This is backed up by many studies showing the negative influences of habitat change and land-use intensification for cavity-nesting bee and wasp communities (e.g. Steffan-Dewenter 2002;Tylianakis et al. 2006). Steffan-Dewenter (2002) showed, for example, that the structure and composition of the landscape adjacent to the exposed trap nests influenced the diversity of cavity-nesting bees and wasps, while Tylianakis et al. (2006) showed changes in the cavity-nesting bee and wasp communities along a gradient of land-use intensification involving rice fields, grasslands, agroforestry, and forest habitats. Bees and wasps are involved in different ecosystem functions like herbivore control (Harris 1994) and pollination (Klein et al. 2007). Evaluating the effects of forest meadow management on the number of established bee and wasp nests is crucial for guiding future insect conservation in forest meadows and beyond.
Cavity-nesting bees and wasps can be separated, based on their natural history, into three major functional groups. The three major functional groups are: (1) bees provisioning their larvae with pollen provided by flowers; (2) herbivorehunting wasps provisioning their larvae for example with moth and beetle larvae or aphids which they collect from vegetation; (3) carnivore-hunting wasps, in our study mainly represented by spider-hunting wasps, collecting spiders in the surroundings of their nests (Westrich 1996). By grouping cavity-nesting bees and wasps into these three functional groups, we can test the individual response of these major functional groups to the mulching management.
The aim of the study is to find out if the mulching of forest meadows at different times affects cavity-nesting bee and wasp nest numbers in trap nests and if the responses differ between major functional groups, to provide management recommendations. Therefore, we have the following three hypotheses: Meadow mulching directly reduces short-term flower availability (Cizek et al. 2012). We therefore expect mulching twice a year to decrease the number of cavity-nesting bee nests (hypotheses 1). Mulching also directly reduces herbivorous insect larvae attached to meadow plants (Humbert et al. 2010), which are suitable food resources for herbivorehunting wasps (Westrich 1996). Especially in August and September insect larvae and aphid abundance is high (Chung et al. 1980;Holmes et al. 1979). Hence, we assume mulching twice a year and mulching in September will reduce the number of herbivore-hunting wasp nests compared to the control (hypotheses 2). In our study, carnivore-hunting wasps depended mainly on spiders as a food resource. Spider abundance increased with the succession of grasslands to forest (Brauckmann 2013) and spider-hunting wasps are known to find their prey predominantly in forests (Rypstra et al. 2007). Therefore, if these wasps hunt in the surrounding forest and not on the meadows, we expect there will be no effect of mulching on the number of carnivore-hunting wasp nests (hypotheses 3).

Study region
The 10,000 ha Black Forest National Park is located in Baden-Württemberg and was founded in January 2014 (Förschler 2015). The national park is divided into three zones: core, transition and management zone (Förschler 2015) and the examined 24 forest meadows were distributed over all three zones located in the southern part of the Black Forest National Park (Fig. 1A). The national park belongs to the regions with the highest annual precipitation rates between 1400 and 2200 mm of Germany (DWD 2019). The annual mean air temperature ranges between 5 and 7 °C (Landesanstalt für Umwelt Baden-Württemberg [LUBW] 2006) and days with a maximum temperature of > 25 °C occurs 5-15 times annually (LUBW 2006) resulting in a short growing season.
The 24 examined forest meadows were located between 570 and 1005 m.a.s.l. and are predominantly surrounded by coniferous forests (dominated by spruce, Picea abies), which were managed for timber production prior to 2014 (Fig. 1B). The 24 forest meadows were created from 10 to more than 150 years ago for hay making, hunting purposes and for use as tree nurseries. In the last 5-30 years, forest meadows in the study region were founded and managed mainly for hunting. Since 2014 the meadows were managed by mulching in June and no fertilization. Management information before 2014 was not available for each meadow in detail (Tschöpe et al. 2017(Tschöpe et al. , 2018. Soil nutrient availability on all 24 examined forest meadows can be considered as poor (Buse et al. 2018).
The selection of the forest meadows was based on the following criteria: (1) Even distribution across the area of the national park (measured in QGIS using distance measurement, the mean minimum distance was 1112 m (± 401 m), the mean maximum distance was 10,617 m (± 2529 m), (2) uniform slope of the meadows with a preference for flat meadows (cut-off slope value was 10°), (3) surrounded by forest.
To test the effect of mulching time and no mulching (control) on cavity-nesting bees and wasps, we established four different treatments in 2017 and 2018. In both study years the same 24 meadows were used and the same treatment was applied for both years on each meadow. We chose this design as management effects are not always detected in the same year but rather in consecutive years (Eckerter et al. 2021). Mulching was conducted as cutting and shredding the vegetation and its direct deposition on the meadow ). This procedure commonly applied by local authorities is considered cost efficient management.
The 24 meadows were divided into four management groups (i) six meadows with mulching in June, (ii) six meadows with mulching in September, (iii) six meadows with mulching twice, once in June and once in September and (iv) six meadows without mulching treatment (control). The mulching was conducted at beginning of the month mentioned above at all sites within 5 days, during dry weather between 10 am and 6 pm. A Fendt mulching machine (Marktoberdorf, Germany) was used. The machine cut the grass above 5 cm, shredded the grass, and then distributed it evenly over the forest meadow.

Cavity-nesting bee and wasp survey
We surveyed the cavity-nesting bee and wasp community using trap nests in 24 different forest meadows. In each forest meadow, we installed four posts equipped with two trap nests each. The posts were placed along the forest edge and orientated towards southeast. Following Tscharntke et al. (1998) the trap nests consisted of 20 cm long PVC tubes, each with a diameter of 13 cm. Each of these tubes contained about 180 internodes of common reed grass, Phragmites communis. In April, July and October 2017 and 2018, all occupied reed internodes (reed internodes in which a bee or wasp made provisioned a nest) were collected from the trap nests and stored in test tubes closed with cotton until March at 4 °C, to simulate hibernation. The walk-in freezer used for hibernation had a dehumidifier to prevent mold. During each winter, the reed internodes were opened and the functional bee and wasp groups and their larval food resource were determined based on nest structure and content (Tscharntke et al. 1998). The reared imagos were stored in the insect collection of the Black Forest National Park. Species level identification was conducted for a subset of individuals but not further used for analyses.

Statistical analyses
For each meadow and sampling year, we pooled the data from the eight trap nests of each meadow and the three sampling dates to calculate the total number of nests, and the number of nests, separated into the three functional groups: bees, herbivore-hunting wasps and carnivore-hunting wasps on each meadow and separated into the nest number of the aphid-hunting wasps, larvae-hunting-wasps, barklice-hunting wasps, spider-hunting wasps, and fly-hunting wasps.
Generalized linear mixed models [GLMMs, package 'lme4' (Bates et al. 2018)] were used to test the effect of the time of mulching on the total number of nests and the number of nests of the different functional groups separately and for each year separately (Eq. 1). We evaluated the model assumptions using the package 'DHARMa' (Hartig 2017) and used an observation level random effect (olre), to account for overdispersion.
The statistical analyses were performed in R, version 3.6.2 (R Core Team 2019).
In 2017 the herbivore-hunting wasp nests comprised 84% of aphid-hunting wasp nests and in 2018, 76% of the herbivore-hunting wasps were aphid hunters. The other nests from herbivore-hunting wasps were built from caterpillar-, beetle larvae-and barklice-hunting wasps. The carnivorehunting wasp nests were mainly spider-hunting wasp nests and a few fly-hunting wasp nests (< 2% in 2017 and < 1% in 2018) (Table 1).
While the nest numbers of the herbivore and spider-hunting wasps stayed similar in both years, the bee nest number was around 5 times higher in 2018 compared to 2017. Also, the beetle-larvae hunting wasp nests and the barklicehunting wasp nest numbers were more than 3 times higher in 2017 than in 2018. We did not identify all imagos at the species level but for an approximation of the species found in the observed community see Supplementary Table 1 (for  a list of flowering plant species see Supplementary Table 2).
(1) Model < − (number of nests ∼ mulching time + (1|olre), family = Poisson, data) First, we examined the total number of nests in both years. In 2017, the total number of nests was not affected by mulching. In 2018, the total number of nests was reduced in meadows which were mulched twice, in June and in September, (Estimate = − 0.64, std. error = 0.32, z-value = − 2.04, p = 0.04) (Fig. 2). All other mulching times had no effect different from the control.
As the herbivore-hunting wasp nests in our study mainly consisted of aphid-hunting wasp nests and carnivore-hunting wasp nests made mainly out of spider-hunting wasp nests, we analyzed, in a third step, these two groups in more detail: They were separated into aphid-hunting wasps, larvae-hunting-wasps and spider-hunting wasps. We could not analyze the barklice-and fly-hunting wasp nests and we had to sum up the caterpillar-and beetle larvae-hunting wasp nests to larvae-hunting wasp nests because we had too few nests of that type. The result was, that only aphid-hunting wasps reduced their nest numbers in the second sampling year through mulching in September and mulching twice (2018 mulching in September; Estimate = − 1.08, std. error = 0.54, z-value = − 1.99, p = 0.05 | 2018 mulching twice; Estimate = − 1.44, str. error = 0.54, z-value = − 2.66, p = 0.01). Spider-hunting wasps and larvae-hunting wasps were not affected by our treatment.

Discussion
We confirmed the expected negative effect of mulching on aphid-hunting wasps and showed that mulching in September or twice a year is responsible for this pattern. However, bees, larvae-hunting wasps and spider-hunting wasps were not affected by mulching. Therefore, the results show that even highly mobile, flying insects nesting above the meadow vegetation are affected by the removal and mulching of living plant biomass, but depending on the functional group and food resource, species respond differently. Nevertheless, this study on three distinct functional groups provides specific management recommendations for forest meadows that can be beneficial for cavity-nesting wasps hunting for aphids.
Bee nesting was not affected by mulching and the removal of all flowering vegetation in the meadows (hypotheses 1). The lack of response of bees may indicate that these highaltitude and forest adapted bees foraged either at times that were less affected by mulching (before June or in July or August), on fast regrowing flower species or outside the meadows, potentially in the surrounding forests, as was shown by previous studies in different landscape contexts Fig. 2 Effect of mulching time on the number of total bee and wasp nests in 2017 and 2018. Each mulching treatment was compared to the control, significant differences are indicated by different letters (Rodríguez and Kouki 2015;Winfree et al. 2007). The species we found confirm these assumptions. Although we did not identify all reared imagos, we found specialists like Osmia parietina, foraging on fast regrowing Fabacea, Megachile lapponica, foraging on Epilobium sp., that may grow in patches not affected by the management commonly occurring along forest roads. Additionally, we found Osmia truncorum, a specialist on Asteracea which likely occurs at the highest abundance in forests (e.g. on Senecio ovatus). Bees in the genus Hylaeus can also do their foraging in trees/ forests (Westrich 2018). Additionally, Eckerter et al. (2021) found, that cavity-nesting bee abundance in the same study area (northern black forest) was not affected by the amount of meadows area in the surroundings. Also, cavity-nesting bees depend on some nesting material (resin, leaves) which is obtained from forests (Rodríguez and Kouki 2015). Future studies, which examine the proportions of nesting resources and pollen collected from forest plants and meadow plants in bee nests could reveal the importance that both meadow and forest flowers have for the cavity-nesting bee communities in forest meadows.
The nest number of carnivore-hunting wasps (and more specifically spider-hunting wasps) was not affected by the mulching treatments (hypotheses 2). This might be explained by a lack of foraging activity of spider-hunting wasps in the meadows. It is more likely that spider-hunting wasps forage amongst forest vegetation, as especially forests providing complex vegetation structures, which is important for spiders (Brauckmann 2013;McDonald 2007). Although we did not identify all reared imagos the spider-hunting wasp Trypoxylon figulus is known to be strongly associated with forest habitats (Osorio et al. 2015) and was often found in our trap nests. Even though, spiders are also abundant in grasslands they might be non-target species to our observed wasp community, as Brauckmann (2013) found a negative effect on spider abundance in the herb layer by mulching compared to not mulching and Bornholdt et al. (1997) also found more spiders on not mulched meadows. A future study directly investigating the effect of mulching on predator species that rely on spiders as prey might shed some light on this question.
The number of herbivore-hunting wasp nests was reduced both by mulching once in September and 2 times during the growing period in the second sampling year (hypotheses 3). The wasp genus Passaloecus (hunting aphids) was most abundant and the genera Pemphredon (hunting aphids), Ancistrocerus (hunting caterpillars), Symmorphus (hunting beetle larvae), Nitela (hunting mainly barklice) and Rhopalum (hunting mainly barklice) contributed fewer individuals. Therefore, aphids were the main food resource of the observed community. As aphid-hunting wasps are highly mobile, we argue that the mulching did not kill the adult wasps directly but likely their larval food resources. The changing nest numbers suggest that meadow aphids were the preferred food resource for the observed aphid-hunting wasps which is supported by a study of Osorio et al. (2015). That means, the aphid-hunting wasps could probably not escape to the forest for larval food provisioning or produced fewer nests due to longer foraging flights. Therefore, mulching in September and twice a year, which can be expected to decrease aphid abundance, likely also disturbed aphid-hunting wasp nesting. As without disturbance through mulching, especially in August and September aphid abundances remain high (Chung et al. 1980). The spider abundance (Blick 2014, Bornholdt 1997 and the availability of pollen (Frankie et al. 2019, Salisbury 2015 declines towards the end of the growing period. Therefore, aphid-hunting wasps might still be nesting at the end of the growing season, while bees and other wasps have finished nesting earlier or possibly before the first mulching took place. We also expected the caterpillar-and beetle larvae-hunting wasp nest numbers to be negatively affected by mulching twice or in September, as insect larvae are negatively influenced by mulching (Humbert 2010). But this was not the case. As, insect larvae are also very abundant at the end of the growing season (Holmes et al. 1979) the larvae-hunting wasps on our examined forest meadows may also find their larval food in forests or due to their phenology earlier or even before the first mulching took place. Future studies on the responses of individual herbivore-hunting wasp species and their food resources to mulching over several years and taking seasonal activity patterns of species into account may confirm our observations and show the impact that mulching has across trophic levels.
Mulching in June, as we recommend for aphid-hunting wasps agrees with the current mulching recommendations which recommend mulching either in June/July or twice in June/July and August/September with the conservation of the overall vegetation community in mind . The reason behind this, is that late mulching leads to biomass accumulation and thereby inhibits the growth of light dependent plant species . With regards to mulching, the site conditions are important, e.g., for meadows which were historically kept open by grazing, machine mulching is no substitute for preserving the plant species community (Römermann et al. 2009). Additionally, late flowering plants are often adversely affected by early mulching (Römermann et al. 2009). Also, certain soils and climatic conditions are unfavourable for mulching (e.g., wet soils and cold air temperatures) ). An alternative to mulching the whole grassland area in June, is mulching, or even mowing in strips. This might prove to be the best management strategy to conserve the overall plant and insect community (Hoste-Danyłow et al. 2010;Humbert et al. 2009).
If mulching is applied, the cuttings remain in the meadow . Therefore, it is often claimed that mulching leads to a biomass and nutrient accumulation ). However, this was not the case in the 45 year long management experiment "Offenhaltungsversuche Baden-Württem-berg" (Brauckmann 2013;Schreiber et al. 2013). Our mulching experiment was too short in duration to answer this question for forest meadows. It took Doležal et al. (2011) five to six years to show the changes that occurred in the vegetation communities induced by mulching. Furthermore, the intra-annual variation in the number of bee nests, beetle-larvae hunting wasp nests and barklice-hunting wasp nests shows that there are potentially other factors such as climatic parameters that may influence populations even more than do food resources. To fully understand the effect of mulching of isolated forest meadows on bee and wasp populations long-term studies are required.
Our study focused on meadows completely surrounded by forests. Therefore, the management of the meadows alters the complete non-forest habitat. This scenario also likely occurs in patches of seminatural habitats, meadows, flower strips, field margins in agricultural landscapes surrounded by monotone habitats of low natural value. Therefore, our results might also apply to isolated habitats in agricultural landscapes but further research is needed for confirmation.

Conclusion
Mulching affects the studied functional groups of cavitynesting insects on forest meadows differently. The high mobility of cavity-nesting insects may allow at least some groups, for example, bees to find resources outside the managed grassland that compensates for the temporary lack of resources due to mulching. However, aphid-hunting wasps were negatively affected. To which extent resource type and resource recovery after mulching contributes to cavitynesting insect population dynamics and if strip-mulching could be an alternative strategy, needs to be investigated in further studies. For management implications, our data suggest that mulching of forest meadows in June produces the least severe effect for aphid-hunting wasps nest numbers. We therefore recommend mulching in June rather than in September or twice a year for protecting aphid-hunting wasps.

Conflicts of interest No.
Ethical approval No.

Consent for publication
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.