Microhymenoptera in roadside verges and the potential of arthropod‐friendly mowing for their preservation

The worldwide decline of insects is one of the major challenges for humankind. One of its main drivers is intensive farming, which reduces habitats and food resources for insect populations and causes direct mortality by pesticides. In addition, mowing of grassland poses another threat to insects, especially when it is done frequently, such as in roadside verges. Roadside verges comprise large areas worldwide and can serve as habitats for animals and plants and as corridors to connect populations when they are maintained ecologically, for example when mowing is done arthropod‐friendly and with low frequency. Microhymenoptera are highly diverse, mostly show a parasitoid lifestyle and occur in large numbers in meadows and grasslands. Although they have an important ecological function in regulating their arthropod host populations, they were mostly neglected in studies on the influence of mowing on insects. Therefore, we assessed which families of microhymenoptera occur in roadside verges and meadows, examined if they are affected by the mowing process with a conventional mulching mower, and studied which groups benefit from arthropod‐friendly mowing, using the mowing head Eco 1200 plus from MULAG, and a flushing bar made of track tarpaulin. In our experimental grassland plots, we found specimens of 18 families from the six microhymenoptera superfamilies Chalcidoidea, Ceraphronoidea, Diaprioidea, Ichneumonoidea, Platygastroidea and Proctotrupoidea. Mowing with a conventional mulching mower caused a significant loss of up to 64% for parasitoid Hymenoptera. The Eco 1200 plus showed an arthropod‐friendly effect only on the number of individuals of Chalcidoidea, saving 38% of individuals compared with the conventional mower. The flushing bar showed a significant effect on total number of individuals with a reduction only on Chalcidoidea and a tendency for Ichneumonoidea with 30% and 47%, respectively. Our study demonstrates the detrimental effect of mowing with a conventional mulching mower on microhymenoptera and shows that this effect can be partly reversed for Chalcidoidea and Ichneumonoidea through the arthropod‐friendly mower and a flushing bar. Our study highlights the importance to consider microhymenoptera in the context of insect decline.

The Eco 1200 plus showed an arthropod-friendly effect only on the number of individuals of Chalcidoidea, saving 38% of individuals compared with the conventional mower.The flushing bar showed a significant effect on total number of individuals with a reduction only on Chalcidoidea and a tendency for Ichneumonoidea with 30% and 47%, respectively.Our study demonstrates the detrimental effect of mowing with a conventional mulching mower on microhymenoptera and shows that this effect can be partly reversed for Chalcidoidea and Ichneumonoidea through the arthropodfriendly mower and a flushing bar.Our study highlights the importance to consider microhymenoptera in the context of insect decline.

| INTRODUC TI ON
Together with climate change, the global decline of insects (Hallmann et al., 2017;Sánchez-Bayo & Wyckhuys, 2019;Seibold et al., 2019;Wagner et al., 2021) is one of the major threats for humankind.This is because insects provide services such as pest control, pollination and nutrient cycling and are therefore essential for the resilience of our ecosystems (IPBES, 2019).Recent studies report several causes for the decline of insects with intensive farming as main driver (Goulson, 2019;Sánchez-Bayo & Wyckhuys, 2019).In homogenous landscapes dominated by monocultures, insects suffer from the absence of nesting places and host plants, disconnected habitats (Harrison & Bruna, 1999) and intensive use of pesticides (Brühl et al., 2021;Sanchez-Bayo, 2011).In grasslands, insects are negatively affected by the deposition of nutrients due to airborne nitrogen (Peppler-Lisbach et al., 2020), the overuse of fertilizers (Niedrist et al., 2009), mulching (Moog et al., 2002) and frequent mowing (Niedrist et al., 2009) which reduce the diversity of wild plant species, that are essential food sources for insects (Fornoff et al., 2023).As a consequence, the number of specialist species decline and species composition becomes more homogenous (Gossner et al., 2016).In addition, mowing also has direct effects on insects.The cutting itself, physical damage through the cutting machinery (Humbert et al., 2009) and mechanical stress during hay removal was shown to cause considerable mortality in insects, ranging up to 88% of individuals being killed during a single mowing process (Hemmann et al., 1987;Humbert et al., 2010;Oppermann & Classen, 1998;Steidle et al., 2022;Von Berg et al., 2023).This is not only true for agricultural grassland, but for all areas that are mown regularly including roadside verges.The latter comprise an area of around 6767 km 2 in Germany (equivalent to 1.9% of the total area) (Reck & Mueller, 2018) and 270,000 km 2 worldwide (Phillips et al., 2020).Apart from significant negative effects of roads on insects by causing mortality, acting as barriers and as sources of air pollutants and de-icing salt (Muñoz et al., 2015), road verges also positively contribute to the environment and have the potential to mitigate negative impacts (Phillips et al., 2020).While the part of the road verges directly adjacent to the street has to be mown two to three times per year to enable optimal visibility for the car drivers, the part on the far side of the street is not under economic or aesthetic pressure such as agricultural areas or parks, meaning that the application of pesticides and fertilizers is uncommon and it is mown only once per year or even less (Betz et al., 2022;Unterseher & Stottele, 2016).They are important corridors for insects (Holzschuh et al., 2010;Villemey et al., 2018) and plants (Tikka et al., 2001) in a homogenous landscape and serve as bridges between habitat patches (Fischer et al., 2022) and thus support gene flow (Holzhauer et al., 2006).Furthermore, they show a similar diversity and higher abundance of insects than surrounding grassland habitats (Villemey et al., 2018) and can provide ecosystem services such as pollination and pest control for surrounding agricultural areas, carbon storage, regulation of air, water, soil and associated pollution and cultural services, such as aesthetics for pedestrians and road users (reviewed by Phillips et al., 2020).Therefore, road verges receive increasing attention in conservation efforts of cities and communities as refuges for biodiversity (Fischer et al., 2022;New et al., 2021;Villemey et al., 2018).However, to comply with safety regulations such as reducing risks due to impaired sight, road verges must be mown regularly (Unterseher & Stottele, 2016), which affects residing animals.Consequently, a concept for arthropod-friendly mowing is needed that considers safety, costs and ecosystem services (Phillips et al., 2020).
Currently, disc mowers, drum mowers and mowers with conditioners are used widely for roadside management, but have shown to be harmful for birds (Tyler et al., 1998), amphibians (Classen et al., 1996), small mammals (Grendelmeier, 2011) and arthropods (Betz et al., 2022;Hemmann et al., 1987;Humbert et al., 2010;Steidle et al., 2022).Recently, the company MULAG, a manufacturer of roadside maintenance vehicles in Germany has developed an arthropod-friendly slope mowing head ('Eco 1200 plus') for road verges.It has six features, which are assumed to reduce arthropod mortality: (i) a flushing bar in front of the mowing head to chase arthropods away, (ii) a disc mowing system with reduced contact surface with the grass, (iii) an increased mowing height of >10 cm, (iv) a largely closed underside that prevents vertical airflow from below and thus sucking up of arthropods into the mowing head, (v) a reduced contact area of the wheels to reduce damage to ground-dwelling arthropods and (vi) a suction device which removes the mown grass without mulching and prevents the accumulation of nutrients on the site.A recently published study showed that the Eco 1200 plus, even when not equipped with the flushing bar significantly reduced mortality in Araneae, Heteroptera, Auchenorrhyncha, Lepidoptera and holometabolous larvae compared with a conventional mower (Betz et al., 2022;Steidle et al., 2022).However, the arthropod-friendly mowing head only had a relatively small positive effect on Hymenoptera and Diptera.The authors discussed that the latter were either killed in the mowing process or might have escaped from the approaching mower by flying away (Steidle et al., 2022).In fact, flight behaviour triggered by the approaching mowing machine was observed in birds (Tyler et al., 1998) and flushing bars in front of the mower to chase away insects are offered by many companies (e.g.Müthing GmbH & Co. KG, AS-Motor GmbH).
Generally, flight behaviour as escape strategy is common in some insect groups like flies (Card, 2012), while others jump away such as grasshoppers (Domenici et al., 2011), drop to the ground such arthropod-friendly mowing, biodiversity crisis, ecosystem services, flushing bar, grassland mowing, insect decline as aphids (Humphreys & Ruxton, 2019), feign death (Humphreys & Ruxton, 2018) or run away.This indicates that flushing bars in front of mowers could be beneficial for some arthropod groups, but not for others.However, we are only aware of one study showing that flushing bars decrease the mortality of birds (Klonglan et al., 1959), and an unpublished study indicating a reduction in mortality of arthropods (Hotz, 2013).
As mentioned above, one group of insects that generally escapes from threats by flying and therefore could benefit from flushing bars is the insect order Hymenoptera.It is one of the most species-rich insect order in the world (Forbes et al., 2018), comprising ants, bees, sawflies and wasps (Peters et al., 2017).While their importance for pollination is well-known to the public (Klein et al., 2007;Potts et al., 2016), their function as natural control agents for pest species is rarely mentioned in the public discussion (but see Fornoff et al., 2023).Most Hymenoptera are small-sized parasitoids and develop in or on certain host species, mostly herbivorous insects, thereby providing natural control of their hosts and contributing to the natural balance and the resilience of our ecosystems (LaSalle, 1993;LaSalle & Gauld, 1991).Nevertheless, parasitoid Hymenoptera are mostly underrepresented in biodiversity studies (LaSalle & Gauld, 1991;Shaw & Hochberg, 2001) due to a phenomenon known as taxonomic impediment, that is the lack of resources invested in taxonomy and the resulting lack of experts especially in extremely species-rich groups (Engel et al., 2021).Therefore, it is unclear to which extent microhymenoptera are affected by insect decline.
To increase knowledge on microhymenoptera and how they are affected by mowing, we studied (i) which families of microhymenoptera occur in roadside grassland and might suffer from the mowing process, (ii) which microhymenoptera families benefit from arthropod-friendly mowing using the arthropod-friendly mowing head Eco 1200 plus from MULAG and (iii) which microhymenoptera families benefit from the use of a flushing bar attached to the mowing head.The first two questions were addressed by re-analysing the Hymenoptera samples from Steidle et al. (2022).To study the third question, we conducted a new field experiment with a flushing bar, which is one of the features of Eco 1200 plus.

| Experiment 1: Arthropod-friendly slope mower
The experiment by Betz et al. (2022)  For arthropod sampling, biocoenometers with lids (insect cages without bottom, 1 × 1 × 1 m, Mühlenberg, 1993) were used.Three biocoenometers where placed on each sampling strip with 1 m between each other.On the control track, the biocoenometers were set up before mowing.On the mown tracks, the biocoenometers were set up immediately, that is within a few seconds, after mowing with either the MK 1200 or the Eco 1200 plus.All arthropods inside the biocoenometers were collected using an insect vacuum cleaner (ecoVac from EcoTech GmbH) during a suction period of 3 min.The samples were transferred to plastic vials and filled with 70% denaturated ethanol for preservation.All arthropods taken from one biocoenometer were treated as one sample.A prior test showed that no specimens were left in the biocoenometers after sampling with the insect vacuum cleaner.For this test, we did a resampling of the same biocoenometer immediately after we took the first sample to see how many insects were left.In total, we collected 189 samples from the 21 sections (nine samples per section with three tracks each), but due to temporal constraints only 106 samples were randomly selected and used for the further analysis.

| Experiment 2: Insect flushing bar
The second experiment was conducted on meadow A (see first experiment) on 2 and 3 August 2022.Weather conditions were dry and sunny with 28-34°C air temperature.Due to the absence of rain for several weeks, the vegetation was very dry.
The flushing bar is made of truck tarpaulin.To test its effect on microhymenoptera without the effect of the mowing head, the flushing bar was attached via a rod system laterally to the UNIMOG vehicle without the mowing head and moved over the vegetation with the truck tarpaulin touching the ground.On the edge of the meadow, two tracks with a width of 2 m each were set up for the treatment and the control.Each track was separated into 10 sections, 7 m each (Figure 1).The sections for treatment and control were set up in alternating order.
Arthropods were sampled using biocoenometers on three plots per section as described for the first experiment.The biocoenometers were set up on the control track before the treatment.The other track was treated by moving the flushing bar over the vegetation at a driving speed of 2-4 km h −1 , which is at the lower limit of the usual operating speed.
Immediately after the treatment, the biocoenometers were set up and arthropods were collected by using the insect vacuum cleaner.In sum, we collected 60 samples from the 10 sections.

| Counting of Hymenoptera specimens
Arthropod specimens from both experiments were sorted into insect orders plus Araneae under a stereomicroscope in dishes with 70% denaturated ethanol.Furthermore, Hymenoptera were identified and counted on family level using Goulet and Huber (1993).
Fragments of specimens that could not be assigned to a family were excluded from the analysis.From the first experiment, only 104 out of 189 were randomly selected and sorted due to time constraints, resulting in 86 samples from meadow A and 18 samples from meadow B.

| Statistical analysis
Statistical analysis was performed in 'R', version 4.0.4(R Core Team, 2021).For the analysis, we used the total number of indi- tests (Hothorn et al., 2008).Data from experiment 2 were analysed using paired Wilcoxon Rank Sum tests, as they were not normally distributed.Here, the total individual number included 19 microhymenoptera specimens that could not be assigned to a superfamily due to their poor condition.We did not use a mowing head in this experiment, therefore poor condition of specimens must have been caused by the sampling.

| Experiment 1
In experiment 1, a total of 7137 specimens of microhymenoptera was caught that could be assigned to 18 families.Mymaridae was the family with the highest number of individuals for all samples  the highest number of individuals, followed by Ceraphronoidea, Ichneumonoidea, Platygastroidea, Diaprioidea and Proctotrupoidea (Table 1).
Statistical analysis showed a significant influence of mowing on the total number of individuals and the number of individuals for the superfamilies Chalcidoidea, Platygastroidea, Ceraphronoidea, and Ichneumonoidea (Table 2).We conducted single comparisons between treatments for all tested groups and found a significant decrease of 60% for the total number of individuals (Figure 2a), and 64% for the number of Chalcidoidea (Figure 2g), 52% for Platygastroidea (Figure 2i), 56% for Ceraphronoidea (Figure 2e) and 59% for Ichneumonoidea (Figure 2c) for plots mown with the MK 1200 standard mower compared to unmown control plots.
Likewise, for all tested groups, the number of individuals was significantly lower in plots mown with the Eco 1200 plus compared to unmown plots, ranging from 43% to 51%.A reduction in the loss of insects by mowing with the Eco 1200 plus as compared to the MK 1200, that is an arthropod-friendly effect of the Eco 1200 plus, was found only for the total number of individuals and the number of Chalcidoidea, with a reduction of 29% and 38% respectively.

| DISCUSS ION
To our knowledge, this is one of the few studies to assess the abundance of Hymenoptera families on meadows and the first in roadside verges.In the first experiment, microhymenoptera are represented by a high abundance with up to 220 individuals per m 2 exhibiting a high diversity with up to 18 families and 6 superfamilies.This agrees with earlier studies which also report that microhymenoptera represent a large proportion of insects and species of meadows and grassland (Boness, 1953;Schmitt, 2004).Our data also demonstrate that the diversity of microhymenoptera is large, as we caught members from 18 families and six superfamilies.Anderson et al. (2011) showed that parasitoid Hymenoptera family richness is significantly related to arthropod taxon richness in agricultural grassland.They conclude that parasitoid Hymenoptera are valuable indicators for arthropod diversity.Therefore, the high number of microhymenoptera families that we found in our experiments might indicate a high downstream diversity, that is a large diversity of host taxa and associated food plants.In addition to their high diversity and individual numbers, parasitoid Hymenoptera are of high economic importance because they are natural antagonists of herbivorous insects (LaSalle, 1993).Nevertheless, they are generally understudied in monitoring programmes due to the lack of knowledge on their biology (Shaw & Hochberg, 2001).Our results strongly emphasize the need to consider this insect group in insect conservation efforts including arthropod-friendly mowing concepts for roadside verges.
Just as extensively managed meadows, extensively mown roadsides have the potential to provide important reservoirs for beneficial insects in our cultural landscape.shows for the first time that roadside mowing with a conventional flail mulching mower also has severe effects on the abundance of small-sized parasitoid Hymenoptera, even after one mowing event.
In analogy to grasshoppers (Chisté et al., 2016), it can be assumed that repeated mowing, for example up to three times per year (Unterseher & Stottele, 2016), will cause even higher mortality in microhymenoptera, as they are probably less mobile than grasshoppers.It remains to be studied, how this reduction of all microhymenoptera taxa is affecting the food web of meadows and other trophic levels.
Within the recent years, several companies have developed arthropod-friendly mowing machines to mitigate the detrimental effect of mowing on arthropods (Von Berg et al., 2023).In a recent study, it has been demonstrated that one of these mowing machines, the Eco 1200 plus provided by the company MULAG, is able to reduce the mortality of spiders, planthoppers, bugs and holometabolous larvae (Betz et al., 2022;Steidle et al., 2022).Generally, flushing bars are a simple and cheap method (Hotz, 2013) that can easily be attached to mowing machines.Klonglan et al. (1959) reported on a device designed to reduce the mortality of nesting birds that was made of a bar with chains hanging with some distance to each other that were drawn over a meadow in front of the mowing machine.With respect to insects, we are not aware of any published study that has tested the effect of flushing bars.We used a flushing bar made from truck tarpaulin mounted on a rod system that was attached in front of the mowing head.Thereby, we found a significant chasing effect only for Chalcidoidea and an almost significant effect for Ichneumonoidea, but no effect for the other superfamilies of Hymenoptera which we have studied.The reason for that could not be explained by our data, as the amount of apterous and brachypterous individuals was comparatively low for all superfamilies.This indicates that flushing bars made of truck tarpaulin have a limited effect on insects and are by themselves not sufficient to protect insects.Additional features are required to qualify a mowing machine as arthropod-friendly.It is unlikely, that other types of flushing bars, for example vertically attached metal chains (Klonglan et al., 1959) or simple bars (Hotz, 2013) are more efficient, as they have a much smaller surface area to chase away small-sized insects.However, more studies are required to answer these questions.
In conclusion, our study demonstrates that meadows on roadsides can harbour a huge diversity of microhymenoptera, a group of insects which is largely understudied but is ecologically important due to its high position in food webs.Roadside mowing causes considerable mortality within this group.The use of an arthropod-friendly mowing head might have a positive effect on Chalcidoidea, but not on the other superfamilies.Chalcidoidea and Ichneumonoidea might also benefit from a flushing bar made of track tarpaulin, mounted in front of the mowing head to chase away the insects.In addition, more factors need to be considered for a sustainable mowing regime, such as, for example low mowing speed to increase the escape interval of animals (Klonglan et al., 1959), leaving unmown plots as refuges and hibernation sites, and removal of cut grass (Unterweger et al., 2018;Van de Poel & Zehm, 2014).As roadside mowing has safety purposes and cannot be avoided, it is important to establish mowing concepts that reduce insect mortality.
and Steidle et al. (2022) was conducted on the edges of two meadows (Meadow A: 48°26′18.4″N8°10′21.0″E, land parcel 213/25, on August 20-212,020 and 24-272,020; Meadow B: 48°26′17.2″N8°10′18.0″E, land parcel 213/25, on 2 September 2020) North of the MULAG company site in Löcherberg near Bad Peterstal-Griesbach (Germany, Baden-Württemberg) between 10 AM and 7 PM.Both meadows were alternately wet to moist nutrient-rich valley meadows, which are mown two to three times per year which is similar to the intensively maintained part of the roadside verges next to the street.The mown tracks were no ditches but merged seamlessly into the meadow.Weather conditions were dry and sunny.A conventional mulching mower (MK 1200 flail mulching head with MS blades) and an Eco 1200 plus green maintenance head without flushing bar were used for the experiment.The rear mowing arm of an UNIMOG vehicle was equipped with the Eco 1200 plus without flushing bar and a suction device to remove the cut grass.The front mowing arm of the same vehicle was equipped with the conventional MK 1200 but without a suction device.Mowing speed was 2-4 km h −1 and mowing height was 10 cm.To compare the effect of the Eco 1200 plus on arthropods with the conventional MK 1200, three tracks were set up on the edge of both meadows with a width of 2 m each.The tracks were separated into 7 m sections for the different treatments which resulted in a strip with 126 m length on meadow A (18 sections), and a strip with 21 m length on meadow B (three sections).For a visualized experimental setup seeSteidle et al. (2022).For the three different treatments the tracks were either mown with the Eco 1200 plus, the conventional MK 1200 or not mown for the control.The mowing treatments were assigned at random to the tracks for each 7 m section.The mown tracks had a distance of 0.8 m (mowing heads width: 1.2 m) in between to avoid that the different treatments did affect each other.
viduals (including three unidentified microhymenoptera due to poor condition) and number of individuals on superfamiliy level due to low individual numbers on family level: Chalcidoidea (Encyrtidae, , Eupelmidae, Eurytomidae, Signiphoridae), Platygastroidea (Scelionidae, Platygastridae), Ceraphronoidea (Ceraphronidae, Megaspilidae) and Ichneumonoidea (Ichneumonidae, Braconidae).Diapriidae and Proctotrupidae were not included in the analysis due to low individual numbers.Data were checked for normal distribution using Shapiro-Wilk normality test.Residual plots were optically checked for homogeneity of variances.To analyse data from experiment 1, models were calculated with the number of individuals as dependent variable, treatment (unmown, MK 1200, Eco 1200 plus) as a factor and site (meadow A, meadow B) as random factor.As data were not normally distributed and/or showed inhomogeneous variances, we analysed data using general models (GLMs) and generalized mixed models (GLMMs), family negative binomial (Bates et al., 2014) for the total parasitoid number and the four superfamilies.Afterwards, we performed ANOVA and Tukey

F
I G U R E 1 Experimental design of experiment 2 showing the two tracks in each 7-m section and the plots with the biocoenometers within each track on the meadow.[Colour figure can be viewed at wileyonlinelibrary.com] Rank Sum test revealed that the flushing bar significantly reduced the total number of microhymenoptera on the treated plots compared to untreated control plots by 30% (W = 343.5,p = 0.023, Figure 2b).A significant reduction in the number of individuals by the flushing bar on the level of superfamilies was found for Chalcidoidea (W = 332.5,p = 0.013, Figure 2h) and a tendency for Ichneumonoidea (W = 217.5,p = 0.054, Figure 2d), with 30% and 47% respectively.No influence of the flushing bar was found for Platygastroidea (W = 148, p = 0.110, Figure 2j) and Ceraphronoidea (W = 147, p = 0.514, Figure 2f).
For Hymenoptera, this study has revealed either about 15% (based on median:Steidle et al., 2022) or 22% (based on mean:Betz et al., 2022) more individuals on plots mown with the supposedly arthropod-friendly mowing head as opposed to the conventional mowing head.The more detailed analysis of the samples provided in the present paper shows similar results for microhymenoptera, that is 29% more individuals for the arthropod-friendly mowing head, based on mean.This difference is almost exclusively based on the superfamily Chalcidoidea, which constitutes 59% of all individuals of the microhymenoptera caught in this experiment, whereas no significant differences were found for the other superfamilies.However, we cannot exclude that this could be due to the low number of individuals caught from the other superfamilies.Interestingly, the reduction of Chalcidoidea in the plots mown with the arthropod-friendly mowing head and the standard mowing head in experiment 1 was much higher than in the plots treated with the flushing bar in experiment 2 (arthropod-friendly mowing head: −50%; standard mowing head: −61%; flushing bar: −26%; all based on the median).Assuming that mowing heads and flushing bar have a similar chasing effect, this would mean that a considerable part, that is 24%-34% of the Chalcidoidea in experiment 1 were killed by mowing.In contrast, the numbers are similar for Ichneumonoidea (arthropod-friendly mowing head: −41%; standard mowing head: −41%; flushing bar: −50%; all based on the median) supporting the idea that Ichneumonoidea might have been completely chased away and escaped by flying.However, more experiments are required to support these ideas.
Total number of individuals collected of different superfamilies and families of Hymenoptera in experiments 1 and 2.
TA B L E 1 Test statistics of experiment 1 for ANOVA based on general or generalized mixed models with treatment as factor and study site as random factor.Effects of a conventional mulching mower (MK 1200 flail mulching head with MS blades), an arthropod-friendly mower (Eco 1200 plus, experiment 1), and a flushing bar (experiment 2) on the total number of individuals of (a, b) all microhymenoptera as well as (c-j) on different superfamilies.Shown are box-and-whisker plots of the individuals caught per m 2 in untreated control plots (Control) and plots mown with the MK 1200 (MK) or the Eco 1200 plus (Eco) mowing head or treated with a flushing bar.Different letters (experiment 1), and p-values with asterisks (experiment 2) above the bars indicate significant differences between the treatments.[Colour figure can be viewed at wileyonlinelibrary.com] TA B L E 2 F I G U R E 2