From diverse to simple: butterfly communities erode from extensive grasslands to intensively used farmland and urban areas

The severe biodiversity decline in European agricultural landscapes demands a specific evaluation of the various land-use practices. Many butterflies in Europe, as an important ecological indicator and pollinator taxon, require human interventions to sustain their populations in cultivated landscapes. However, land-use changes and management intensification are currently responsible for their decline. In this study, we compare butterfly communities occurring on 93 sites in seven widely distributed land-use types, viz. extensive meadows and pastures, semi-intensive meadows, vineyards, arable land, settlements and apple orchards. We recorded a high butterfly diversity in supposedly high nature-conservation value (HNV) grasslands (extensive meadows and pastures). All other land-use types showed significantly lower diversity, with decreasing diversity from semi-intensive meadows to apple orchards. Moreover, functional traits uncovered a general trend: extensive grasslands supported communities of more specialized and sedentary species whilst all other non-HNV land-use types showed communities characterized by mobile generalists. Community composition was driven by the land-use type and explained by plant-based indicator values for nutrients and light and temperature variables. Important life-history traits further correlated with site variables confirming the shift from specialists to generalists along increasing land-use intensity gradients and the effect of the thermal environment on phenological traits. We found supporting evidence for the effectiveness of regional Agri-Environmental Measures for butterfly conservation in European cultural landscapes and for the European conservation schemes to focus at least partly on the preservation of HNV grasslands with extensive management. Furthermore, we clearly show the poor ecological state of butterfly communities in more disturbed land-use types (including urban areas) and propose adopting measures to improve butterflies’ conservation in these environments.


Introduction
The current biodiversity crisis, i.e. the loss of species in general, is to a large extent driven by human land use (Oliver and Morecroft 2014). This applies to insects, the most diverse animal group, that is globally highly exposed to threats of agricultural origin (Wagner et al. 2021) and where severe declines have been recorded in some European countries (Biesmeijer et al. 2006;Hallmann et al. 2017). The loss of insect biodiversity is considered a major threat since it is expected to alter ecosystem functions (Potts et al. 2010) and trophic relations (Wagner 2020).
Butterflies, a charismatic insect group, are some of the best studied invertebrates and are considered a valuable indicator taxon for environmental quality and conservation (Settele et al. 2009b;Rákosy and Schmitt 2011). Many studies have pointed out a general decrease of taxonomic and functional diversity under land-use intensification and land-use changes on the cost of extensive grasslands (e.g., Ekroos et al. 2010;Börschig et al. 2013;Perović et al. 2015;Habel et al. 2019;Mora et al. 2021;Warren et al. 2021). In Europe, butterflies reach their highest diversity and abundance in grasslands and steppes, alpine and subalpine grasslands, and mesophile grasslands which are partly man-made habitats, thus relying on extensive but continued management (Stefanescu et al. 2004;Settele et al. 2009a;Bubová et al. 2015). Furthermore, some major threats to butterflies originate from intense land-use forms (e.g., pesticide exposure and fertilization; Braak et al. 2018;Brühl et al. 2021;Roth et al. 2021), threatening the survival of populations even beyond the specific sites of application (Goulson 2013;Tarmann 2019).
Most conservation-oriented studies on butterflies in Europe focus on grasslands (Settele et al. 2009a;Van Swaay et al. 2019;Hannappel and Fischer 2020), while other land-use types have been investigated less intensively (but see Schmitt et al. 2008 for vineyards andErnst et al. 2017 for orchard meadows). Further, studies mainly focused on single landuse types. However, the simultaneous comparison of several different management forms is required to efficiently analyse the effect of land use and land-use changes (Cappellari and Marini 2021). Although butterflies are the best studied insect group and detailed data on the species level is available, we often lack information on the communities inhabiting more disturbed land-use types (such as vineyards and intensive orchards) and on the effectiveness of nature conservation practices (such as government-funded agri-environmental schemes; but see Filz et al. 2013).
The European Alps are considered a butterfly diversity hotspot harbouring many endemic species (Menchetti et al. 2021), where in addition different types of land use are practised in close proximity (Kuttner et al. 2015). The few available studies investigating long term trends on a landscape scale within the European Alps agree that intense land-use practices and land-use changes (e.g., conversion of grasslands, intensification of orchards, urbanisation) are contributing to the erosion of butterfly communities (Lütolf et al. 2009;Gallou et al. 2017;Bonelli et al. 2021;Habel et al. 2021). In the mountainous province of South Tyrol (Italy), where this study was conducted, changes in the management of grasslands and expanding apple orchards are viewed as important causes of the dramatic local butterfly decline (Tarmann 2009(Tarmann , 2019Hilpold et al. 2018).
Here, we present a study comparing butterfly communities between seven major landuse types, all commonly found throughout Europe. We chose to include also urbanised areas as a type of land-use with an increasing influence at a local and global scale (van Vliet 2019). Specifically, we aim to describe the communities inhabiting each land-use type by means of diversity measures, community composition and functional species traits expecting (I) the highest species diversity within pastures and meadows and a reduced, simplified and generalist community for the other land-use types. Furthermore, we compare and discuss on a regional scale (II) the effectiveness of regional Agri-Environmental Measures on meadows (subsidies for the extensive management of meadows, in compliance with the European Union Regulation Nr. 1305/2013) in conserving butterflies. Finally (III), we aim to identify key environmental factors (climatic and vegetation parameters) and species traits (phenological, morphological and behavioural traits) important for shaping butterfly community ecology along the land-use gradient under study.

Study area and sites
The study area of the Autonomous Province of South Tyrol in Northern Italy is located in the Central Southern Alps. On a landscape level agricultural land covers 24.1% (LAFIS 2020), while settlements occupy only 1.9% of the area (GEOFABRICK 2020). The predominant land-use types are meadows (7.9%), orchards (2.5%), pastures (excluding alpine pastures: 0.9%), vineyards (0.75%), and arable land (0.5%; LAFIS 2020). A minor part of meadows (0.4% of the total area and 6.5% of the meadow area) consists of subsidised meadows with mowing and fertilizing regimes regulated for nature conservation purposes. Subsidies derive from the second pillar of the Common Agricultural Policy (CAP; European Union Regulation Nr. 1305/2013) and are designed by the local administration to maintain meadow types included in the Habitats Directive (Annex 1). The local landscape of the study area is dominated by a mosaic of different land-use types with only grasslands (at intermediate elevations) and apple orchards (at lower elevations) forming more continuous land-use forms (Anderle et al. 2022).
All study sites and data are part of the long-term project "Biodiversity Monitoring South Tyrol" (BMS 2021). For the present study only sites below 1600 m a.s.l. were considered, since most agricultural production types and urban areas occur within this range (except for vineyards and apple orchards that locally rarely surpass 900 m a.s.l.) and a comparable community of butterflies can be expected. A total of 93 sites ( Fig. 1) were surveyed between 2019 and 2021, subdivided into 12 extensive meadows, 12 semi-intensive meadows, 12 pastures, 13 vineyards, 12 arable land, 12 apple orchards, and 20 settlement sites. The individual study sites were selected using a stratified approach with the landuse categories as strata. Within each stratum a randomised site selection took place. Only urban sites were selected differently (for details of site selection see Online Resource 2). The three grassland categories were defined based on their management. Extensive meadows corresponded to subsidised meadows for nature conservation purposes (e.g. the extensive use is a prerequisite). Semi-intensive meadows were defined as non-eligible for subsidies. Pastures were selected as grassland kept open by grazing (for more details see Online Resource 2). Both extensive meadows and pastures correspond to the definition of high nature value (HNV) farmland (Benedetti 2017;Lomba et al. 2020).

Butterfly surveys
Butterfly sampling was carried out on each site with four replicates (on six sites with three replicates because of weather constraints) between May and August by butterfly 1 3 experts. Exact survey dates were selected flexible to optimize weather conditions and to achieve three to four (but at least one) weeks between two consecutive surveys. Only adult individuals of the superfamily Papilionoidea were recorded. The surveys were carried out between 10:00 and 17:00 in mostly warm (> 13 °C), sunny and dry (> 60% sunshine at 13-16 °C), and calm windy conditions (Beaufort scale: 0-3) following Pollard and Yates (1994). An area of 1000 m 2 (50 × 20 m) was surveyed at all sites using a time-area count of 30 min per site and survey. The area was walked slowly in serpentine lines avoiding multiple counting of individuals. The advantage of a time-area count is an increased record of individuals and species (Rüdisser personal observation; Kadlec et al. 2012;Pellet et al. 2012). An exception was crop fields with high vegetation, where surveys were conducted on the field margins only to avoid damage to crops. However, such sites were surrounded either by other arable land or by semi-intensive meadows, limiting a potential edge effect caused by more natural habitat types. All butterflies were identified either in flight or caught and identified in hand before release. Single escapees identified to genus level in flight were proportionally allocated to the species recorded on the same site. The cryptic species pairs Colias hyale/alfacariensis, Pyrgus malvae/ malvoides and Leptidea sinapis/juvernica were treated as one single unit in all following analyses, because of difficult differentiation in the field. Species identification was based on keys of Stettmer et al. (2007), Paolucci (2013) and Gergely (2019) and nomenclature follows Wiemers et al. (2018).

Site descriptors
Plant surveys were carried out in the same year as butterfly sampling, following the protocol of the Eurasian Dry Grassland Group (Dengler et al. 2016). Only in settlements, a transect method was applied (for more details see Online Resource 2). To describe the local ecological conditions, mean Ellenberg indicator values (Ellenberg 1974) were calculated for each site using presence-absence data of all vascular plants. Ellenberg values for temperature, light, soil moisture, soil nutrients, and soil reaction were taken from Pignatti et al. (2005). Ellenberg values describe the response of single plant species to a variety of ecological gradients and are based on botanist experience of plant species requirements. If averaged upon the plant species co-occurring in a community, these aggregated values are known to well describe environmental site conditions (Carpenter and Goodenough 2014).
Because seasonal weather changes can cause considerable year to year variation (Roy et al. 2001;McDermott Long et al. 2017), we extracted monthly mean temperature (T) and total precipitation (P) for each site at a resolution of 250 m from a locally scaled dataset of Crespi et al. (2021). Climatic variables were selected to match the recording year for each site and combined in winter (December-February), spring (March-May) and summer (June-August) seasons to account for the different development strategies of butterflies (McDermott Long et al. 2017). Furthermore, to best describe the elevational gradient the mean temperature over two years (the year before sampling occurred and the year of sampling) was extracted from the same dataset for each site (for more details on variables see Online Resource 2 and 3).

Species traits
Species traits were collected from literature and comprised traits concerning, phenology (start of flight period, regional voltinism), morphology (wing index), biotic interactions (phylogenetic diversity of larval host plants), behaviour (egg laying type, dispersal ability, feeding on non-nectar resources, larval feeding strata), overwintering stage, and climatic niche (temperature and precipitation requirements; Settele et al. 1999;Huemer 2004;Schweiger et al. 2014;Middleton-Welling et al. 2020; for details on individual species traits see Online Resource 2 and 3). Although conservation status (Red List) is not a trait per se, we decided to include the regional red list (Huemer 2004) to include a conservation relevant aspect of the recorded communities.

Statistics
Observed butterfly abundance was scaled to time of observation (minutes) on each site (to account for small record differences). Species richness was standardized to the observed mean abundance of 59 individuals per site (using extrapolation and rarefaction approaches) and species accumulation curves were computed to assess overall diversity (using abundance-based Hill numbers of q = 0 and q = 1; package iNEXT Hsieh et al. 2016). Nonparametric tests (Kruskal-Wallis test for overall land-use categories and Wilcoxon tests for pairwise comparisons corrected for false discovery rate after Benjamini and Hochberg; Pike 2011) were used to test land-use type effects on abundance, richness, diversity measures, and functional traits (scored as Community Weighted Means, CWMs).
To investigate community composition, species abundances were square-root transformed (to attenuate the prevalence of abundant taxa) and a virtual dummy species with lowest abundance was added to each site (to improve ordination stability; Clarke et al. 2006). Community composition between land-use types and year of sampling was tested with a permutation analysis of variance (PERMANOVA with Bray-Curtis distances and 999 permutations; package vegan Oksanen et al. 2017). Further, community composition drivers were analysed using a Canonical Analysis of Principal coordinates (CAP; function capscale based on Bray-Curtis distances), where a forward selection procedure (function ordistep) selected for significant variables (vegan package Oksanen et al. 2017). This approach selects significant variables based on the Akaike Information Criterion (AIC) and p values computed with a Monte Carlo permutation test. As random factors, we included the site-specific weather variables of winter mean temperature and spring sum precipitation. These weather variables scored superior in explaining the community composition in a previous constrained ordination just with climate variables (data not shown). All predictor variables were z transformed before implementation.
Using the trait matrix, we computed multivariate functional diversity measures and community weighted means for each trait (CWMs, using the dbFD function in the FD package; Laliberté et al. 2015). To determine how local and environmental factors influence butterflies' traits, we used a combined RLQ and fourth corner approach with the ade4 package (Dray and Dufour 2007). With three matrices, R (environmental variables), Q (butterfly traits) and L (species abundances), we performed a correspondence analysis (L) and principal component analysis (R, Q) and used permutations to evaluate if environmental variables influence traits distribution. We created an RLQ biplot to assess the relationships between species traits and environmental variables and determined the significance of each trait-environment relationship using the fourth corner analysis. We used Monte Carlo permutations (9999) to test for correlations between quantitative variables and used the "D2" correlation coefficient to test for associations between quantitative variables and each categorical value separately (Dray et al. 2014). Only a careful selection of traits (excluding highly correlated traits) and of variables (with high explained variance as resulted from the constrained ordination) was implemented in this analysis. All analyses and graphics were performed using R version 4.0.4 (R Foundation for Statistical Computing).

Results
We recorded a total of 5513 butterfly individuals belonging to 100 species (54% of the total known regional butterfly fauna; Huemer 2004). Species richness ranged from 1 to 32 species and abundance from 1 to 271 individuals per site. Twenty-two species were recorded from one site only and 13 by just one single individual.
The overall accumulation of species was largely saturated across all land-use types except semi-intensive meadows. The highest number of species (Hill number H0) and maximum diversity (H1) was clearly supported by the two extensive land-use types, viz. pastures and extensive meadows (Fig. 2). Furthermore, semi-intensive meadows and vineyards follow with an intermediate diversity, settlements, and arable lands with less, and apple orchards with the least species diversity of all (Fig. 2). This ranking was far more robust with Shannon's exponential diversity than observed species richness.
Scaled individual abundance, standardized taxonomic richness, taxonomic diversity (expH'), and functional richness on the site level followed a similar pattern of high 1 3 scores for extensive meadows and pastures in comparison to all other land-use types. Apple orchards, settlements, arable land, and vineyards on the other hand scored significantly lower. Semi-intensive meadows occupied an intermediate position between the more intense land-use types and the extensive grassland sites (Fig. 3). Further, plant species numbers on each site correlated with butterfly diversity (r = 0.325, p = 0.001), indicating an influence of the site vegetation.
The community composition of butterflies varied significantly between land-use types (PERMANOVA; F: 5.36, p = 0.001) and year of sampling (PERMANOVA; F: 2.14, p = 0.009). However, no significant interaction between year and land-use type could be detected (PERMANOVA; F: 1.00, p = 0.44; Online Resource 1). In the constrained ordination (Fig. 4) separation between the extensive grassland habitats (meadows extensive and pastures extensive) and the more intensive forms of land-use and settlements could be seen. Settlements showed high variability in their butterfly communities. The variables that significantly explained variation in butterfly community composition were the Ellenberg indicator values for nutrients and light (abbreviated as N and L) as well as mean site temperature (T_Mean, describing the elevational gradient).
Most species traits scored as community weighted means (CWMs) followed a similar pattern as the assemblage diversity measures, viz. a significantly different score for meadows extensive and pastures from the more intense types of land-use. Semi-intensive meadows occupied again an overall intermediate position (Fig. 5, Online Resource 1). Specifically, on extensive meadows and pastures, we found butterfly communities composed of more endangered species, more specialized in their feeding niche as larvae, with fewer generations per season, with a more limited dispersal propensity and a high prevalence of species overwintering as larvae (Fig. 5). Further, the temperature indicator showed a higher score on pastures and vineyards than on meadows. Other CWMs further confirmed the pattern of higher dispersal capacity, longer flight period duration, generalist feeding and less intense ant association of lycaenids larvae among the butterfly communities of more disturbed sites (Online Resource 1). The correlation matrix derived from the fourth corner analysis (for further details see Online Resource 1) detected seven significant correlations between environmental variables and butterfly traits (Fig. 6). Butterflies having more generations per year (p = 0.002), a broader larval feeding niche (p = 0.01) and higher dispersal potential (p < 0.01) were found on sites with an increased soil nutrient indicator (N). Since N is associated with more intense land-use ( Figure S4, Online Resource 1), we find further support of a less specialized and more mobile community occurring on these sites. Further, the overwintering stage as larvae correlated negatively with N values (p < 0.01), indicating that an expected higher disturbance, due to more frequent management, might damage this critical development stage. Finally, traits expected to shift with elevation correlated with mean annual site temperature: hibernation as larvae were less prevalent (p = 0.006), voltinism increased (p = 0.002), and the butterfly community temperature index was higher (p < 0.01) with higher site temperature.

Discussion
The comparison of all seven major land-use types in South Tyrol regarding their butterfly diversity, community composition and functional traits revealed a consistent pattern, dividing the habitat types into two main groups: extensively managed grasslands (both meadows Fig. 5 Community weighted means of selected butterfly species traits showing different life history strategies between land-use types. Pairwise comparisons based on Wilcoxon tests (different letters denote significant pairwise differences between land-use categories of p < 0.05). The dashed line is the overall mean of each metric over the complete dataset. A Red List score corresponding to IUCN classes (0 = Regionally extinct -5 = Least concern), note the y axis is inverted: lower values denoting a higher proportion of threatened species. B Larval host plant phylogenetic diversity (Faith's PD). C Dispersion score following Settele et al. (1999) (1 = sedentary -9 = migratory species). D Voltinism: number of generations per year. E Binomial trait for overwintering as larvae. F European-wide mean annual temperature requirements of each species (Schweiger et al. 2014) 1 3 and pastures considered high nature value grasslands; Benedetti 2017) and a second group with more intense land-use types (such as apple orchards, arable land, and vineyards) and settlements. This dichotomy of extensive grasslands versus other, more intense land-use forms is generally well-known and leads to communities comprised of more specialist, sedentary species of high conservation value vs. assemblages dominated by mobile generalists (e.g., Börschig et al. 2013). Our results corroborate both the value of extensive grassland and the common negative effect of all other more intense land-use types in supporting butterfly communities.
The high diversity and increased presence of threatened and specialist species (such as Phengaris arion, Chazara briseis and Parnassius apollo) occurring in extensive grassland were reported also for other taxa in South Tyrol (Hilpold et al. 2018) and Central Europe (e.g., Batáry et al. 2010;Weiss et al. 2013). High nature value grasslands are therefore of great importance in supporting biodiversity and the responsibility of local stakeholders should be acknowledged in both valuing and protecting these land-use types. Our study indicates that the current local implementation of the European Union policy (GAP, European Union Regulation Nr. 1305/2013) to subsidize and regulate the management of extensive meadows appears successful in supporting an abundant, species-rich and diverse butterfly community with a high proportion of threatened species compared to non-subsidized semi-intensive meadows.
In South Tyrol, the butterfly communities of pastures and extensive meadows were strikingly similar. The only notable difference was a higher thermophilic character in pastures, explainable since locally, extensive grasslands on south-facing slopes are often drier and often used as pastures. Nevertheless, we underline the high importance of extensive pastures and the great responsibility to conserve them in the future, as already reported by other studies (e.g., Balmer and Erhardt 2000;Dolek and Geyer 2002;Settele et al. 2009a;Habel et al. 2021). Moreover, since mowing can be detrimental to some species (e.g., Fig. 6 Results of the fourth-corner tests of butterfly traits in relation to environmental variables. Coloured squares represent significant associations [red = positive (+); blue = negative (−)]. Details on single axis scores and traits in Online Resource 1, 2 and 3. ns not significant; "." p < 0.1; "**" p < 0.01 Bruppacher et al. 2016), a combination of practices on extensive grasslands is considered advantageous for butterfly conservation (Bubová et al. 2015;Fiedler et al. 2017). Hence, conservation efforts should include and increase subsidies to preserve the management of extensive pastures wherever this land-use type is still regionally available. In our study region pastures, especially at lower elevations, are increasingly abandoned due to a lack of grazing animals and rationalization of dairy farming, a trend that contributes to the erosion of species richness at lower elevations in the Alps (Habel et al. 2021).
Semi-intensive meadows occupied an overall intermediate position, with higher diversity than other intense land-use forms, but significantly lower than extensive meadows (as reported by others; Settele et al. 2009a;Bubová et al. 2015;Perović et al. 2015). However, considering that this land-use type is widely applied in the region of investigation, it is still of major importance to support butterfly populations at the landscape level. Moreover, the ongoing intensification of land-use (e.g., increased silage production) will likely transform these semi-intensive meadows into intensive ones and might further threaten butterfly populations (Hannappel and Fischer 2020). Hence, incentives to reduce land-use intensity at these semi-intensive sites, e.g., through subsidies, could contribute to the preservation of threatened grassland biota.
The butterfly communities observed on the other more intense land-use types were strikingly similar to each other from many different perspectives. Vineyards, characterized by more open and drier climatic conditions, supported overall a more diverse butterfly assemblage, however, the differences to the other intense land-use types were negligible on the site level. In South Tyrol, arable land scored similarly to settlements regarding overall butterfly diversity, with a community composition resembling those on semi-intensive meadows. The lowest position in terms of supported diversity was occupied by apple orchards that nowadays dominate many low elevation areas in South Tyrol, confirming the general expectation that these intensely managed orchards represent a very unsuitable habitat for butterflies. In apple orchards, the generalist pest species Pieris rapae makes up 61.2% of the recorded individual butterflies and is the only regularly occurring species (Online Resource 1). This low conservation value contrasts starkly with classical orchard meadows that have been shown to support diverse insect assemblages in Central Europe (Plieninger et al. 2015). The reasons that make these intensive apple orchards such an unfavourable land-use type for butterflies is likely a combination of factors, from the frequent application of additional chemicals (e.g., herbicides, insecticides, fungicides), low plant species diversity (as larval host and adult nectar resources), and frequent disturbance due to constant management. Finally, butterfly assemblages in settlements showed high variability in the species composition, suggesting a certain potential to support a diverse community. Concerning butterfly diversity and CWMs, settlements appear to follow the general trend of more intense land-use types, so that the local urban areas limit butterflies similarly to other intense land-use types. This leads us to suggest a less intensive management of road margins, city parks and to set aside temporary fallows or sow flower stripes to improve habitat quality for butterflies in urbanized areas (as already shown effective by other studies: Aguilera et al. 2019;Dylewski et al. 2019). Finally, in settlements, we recorded the introduced species Cacyreus mashalli (Quacchia et al. 2008) and the recently spreading subspecies Pieris mannii alpigena (also recorded in vineyards; Neu et al. 2021), posing potential threats to natural communities.
Our study intentionally focused on the site level effect of land-use types. However, the landscape around focal habitats is known to be an important feature influencing butterfly communities (Kral et al. 2018;Le Provost et al. 2021). To best account also for stochastic effects in mobile insects such as butterflies (e.g., spillover effects from 1 3 neighbouring habitats) one should consider the accumulation curves based on Hill N1, which downweighs the effect of rare species and delivering a clearer picture of the supported diversity within each land-use type. Furthermore, a test for spatial autocorrelation was not significant for all diversity measures considered (Online Resource 1). Nevertheless, some of the outcomes of our study are most probably influenced by the surrounding landscape such as the similarity between the butterfly communities of arable land and semi-intensive meadows (the predominant land-use type in the region), and a similar effect can be expected for vineyards and settlements site diversity and community composition.
Concluding, the locally (in South Tyrol) applied measures of nature conservation in HNV grasslands (meadows) appeared to conserve efficiently butterflies in extensive meadows and we propose to apply a dedicated subsidies program for extensive pastures at lower elevations, where a diverse and threatened community was recorded. Moreover, we also suggest a regulation of semi-intensive meadows (for example, delaying the mowing term or leaving meadow portions standing; Bruppacher et al. 2016), which would greatly improve overall butterfly conservation in the region, since this land-use is currently predominant in the regional agricultural landscape. Furthermore, a radical change in the management of apple orchards would likely improve the overall regional conservation of butterflies at lower elevations (Tarmann 2019). Suggestions for improvement include a reduced management intensity, e.g. reducing application of agrochemicals, reducing mowing or mulching frequency, and implementing flower strips and hedges. Finally, different management of urban green spaces (for example by sowing flower strips, leaving unmanaged fallow areas and reducing mowing frequency; Dylewski et al. 2019) would greatly benefit butterflies and also other pollinators (e.g., Hofmann and Renner 2020) in urban areas.

Declarations
Competing interests The authors have no relevant financial or non-financial interests to disclose.
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/.