Wildflower Strips Increase Aculeate Pollinator Diversity but Not Abundance in Agricultural Landscapes with Rapeseed in Crop Rotations

Abstract

The decline of pollinators in agricultural environments poses a significant threat to pollination ecosystem services. Wildflower strips are proposed as a strategy to support pollinator populations and enhance their species richness and diversity. We investigated the efficacy of flowering plant mixture in maintaining aculeate pollinator diversity (wild bees, predatory wasps, and their kleptoparasites) within intensively managed agricultural environments where rapeseed is a common rotational crop. Over four years, pollinators were counted five times per season using 250 m transect walks. Our results demonstrated that the diversity and the evenness of species abundance distribution of aculeate pollinators were higher in the sown wildflower strips, whereas mean abundance per transect was greater in the remnants of semi-natural grassland. The low diversity and evenness within the aculeate pollinator assemblage of the semi-natural habitat were attributed to the dominance of the sweat bee Lasioglossum pauxillum, which thrived on mass-flowering rapeseed and concentrated in the flowering grassland fragments after the rapeseed harvest. We conclude that wildflower strips enriched with sown flowering plant mixtures effectively enhance pollinator diversity. Furthermore, both wildflower strips and preserved patches of unmanaged or minimally managed semi-natural grassland habitats can essentially contribute to maintaining pollination ecosystem services within intensive agricultural environments.

1. Introduction

Pollinating insects provide economically valuable ecosystem services, contributing to the pollination of both crops and wild plants, and thereby impacting food production and plant biodiversity [1,2,3,4,5,6,7,8]. However, recent decades of intensive agro-ecosystem management have triggered a decline in pollinators [4,9,10,11]. Agricultural intensification, a primary driver of impacts on pollinator diversity and their services, promotes the fragmentation, destruction, and loss of semi-natural habitats, consequently reducing pollinators’ foraging and nesting opportunities [12,13,14,15,16]. This loss of semi-natural habitats directly results in a decline of native plant species, while reductions in pollinator populations may indirectly contribute to that decline [17,18,19].

To mitigate the impacts of habitat loss on biodiversity conservation in agricultural landscapes, the EU’s Common Agricultural Policy promotes the integration of semi-natural habitats within arable land and the implementation of agri-environmental schemes [20,21,22]. A widely adopted agri-environmental measure is the establishment of sown wildflower strips (WFSs), which serve to enhance biodiversity in agricultural landscapes while promoting ecosystem services, such as pollination and pest control [23,24,25,26,27]. Although the positive effects of WFSs on beneficial insect abundance and diversity are well-documented, their effectiveness remains inconsistent across studies [28,29,30,31,32,33]. Research on the effectiveness of the EU agricultural policy measures targeting biodiversity conservation, including pollinator protection, exhibits a geographic bias, with Central and Eastern European countries being underrepresented [34,35].

A wildflower strip is defined as a strip or plot of sown native, and/or non-native flowering herbaceous plants that provide pollen and nectar resources for native pollinators [31]. WFSs are typically established along field boundaries or within crops, requiring minimal space and easily integrating into farming practices. They are maintained for periods ranging from one to five years [20,36,37]. Notably, pollinator abundance within WFSs may decline over time following WFS establishment [38].

The selection of plant species for seed mixtures is often based on practical and economic criteria, such as cost, seed availability, germination success, or expert opinion, frequently with limited consideration of their ecological role in habitat restoration or plant–pollinator interactions [39,40]. However, optimal pollinator support requires flower mixture compositions with specific biological properties, necessitating careful planning and management [41,42,43,44,45]. Seed mixture design and implementation may vary depending on whether the goal is to promote pollinator diversity conservation or pollination services [46]. It has been suggested that improving the seed mixtures recommended to farmers could enhance their ability to attract a wider range of pollinator species and include plants that support multiple specialized pollinators [17,47,48,49,50,51,52,53].

The cultivation of rapeseed (Brassica napus) can have significant implications for wild pollinator populations in agricultural landscapes. Overall, numerous studies examining the impact of oilseed rape crops on the WFS have focused on biological pest control [54,55,56,57], with fewer studies addressing plant–pollinator interaction effects on pollinator abundance and diversity [58,59,60,61]. While rapeseed fields can provide abundant pollen and nectar resources [62], their impact on wild pollinator diversity and abundance depends on the context. Evidence indicates that insect pollination significantly enhances seed yield and quality in rapeseed crops, with honeybees being the most effective pollinators due to their foraging behavior and efficiency in pollen transfer between flowers [63]. However, wild bee abundance and diversity can be significantly higher in adjacent uncultivated areas than in conventional rapeseed fields, highlighting the potential negative effects of monoculture farming on pollinator diversity [64]. WFSs can compete with rapeseed for pollinators, drawing bees away from rapeseed flowers despite increasing the overall wild pollinator population [65]. The landscape context, including the presence of semi-natural habitats surrounding rapeseed fields, can play a crucial role in determining wild pollinator abundance and diversity [66,67]. Increased vegetation diversity in agricultural landscapes can enhance pollinator species richness, suggesting that integrating WFS or maintaining hedgerows could benefit both crop production and pollinator conservation [68,69,70,71].

The aim of our case study was to evaluate the potential of species-rich flowering plant seed mixtures, such as Lebensraum I^®, developed for German agro-ecosystems, for maintaining aculeate pollinator diversity (wild bees, predatory wasps, and their kleptoparasites) within intensively managed agricultural environments in temperate Eastern Europe, where rapeseed is a common rotational crop. We hypothesized that (1) species-rich plant mixtures in WFSs provide floral resources for a broader range of pollinator species, thereby enhancing the species richness and diversity of pollinator assemblages; (2) the establishment of WFSs leads to increasing over subsequent years abundance and/or species richness of aculeate pollinator populations due to richer floral resources; and (3) mass-flowering rapeseed fields offer abundant floral resources to certain wild pollinators, thereby positively influencing the abundance of particular species.

2. Materials and Methods

2.1. Study Sites

Field observations were conducted in agricultural environments where rapeseed was part of the crop rotation. We compared the diversity and relative abundance of Hymenoptera Aculeata in three sites in the Kupiškis district of Lithuania, in temperate Europe. These included two available WFSs at the edges of crop fields (Flower strip 1: 425 × 10 m, 55°48′20″ N, 25°13′25″ E, and Flower strip 2: 300 × 6 m, 55°48′11″ N, 25°13′13″ E) and a control plot with semi-natural grassland vegetation surrounded by intensively managed crop fields (Control: an approximately 12 m wide unmanaged grassland strip on both sides of a drainage ditch, 55° 48′ 07″ N, 25° 07′ 05″ E). The crop rotation in the arable fields surrounding the study sites included winter rapeseed, winter wheat, and, occasionally, faba beans. The nearest narrow strips of semi-natural grassland vegetation along drainage ditches, which are potential pollinator sources, were located approximately 500 m from the WFSs. Larger patches of semi-natural mowed meadows and pastures were situated at distances exceeding 1 km. The WFSs were established in the year 2020, and the diversity of bees and wasps was monitored from 2021 to 2024.

The vegetation of the WFSs was enriched using the plant seed mixture Lebensraum I (Saaten Zeller GmbH and Co. KG, https://www.saaten-zeller.de/, accessed on 3 March 2025), which originally included 50 flowering plant species. Some sown species did not successfully reach the flowering stage, likely due to competition with other plants, including local weeds.

Table 1. Lists of sown and naturally occurring entomophilous plants recorded in the wildflower strips and the control site. Non-native species are marked with an asterisk (*).

Plants from the Seed Mixture Lebensraum I in WFS	Wild Entomophilous Plants, Naturally Invading the WFS	Wild Entomophilous Plants in Semi-Natural Grassland of the Control Site
Achillea millefolium Agrimonia eupatoria Anethum graveolens * Anthriscus sylvestris Borago officinalis * Camelina sativa * Carthamus tinctorius * Carum carvi Centaurea cyanus Cerastium holosteoides Cichorium intybus Clinopodium vulgare Cota tinctoria Crepis biennis Daucus carota Dipsacus fullonum * Echium vulgare Fagopyrum esculentum * Galium album Galium verum Helianthus annuus * Heracleum sphondylium Leucanthemum ircutianum Linum usitatissimum * Malva moschata Malva sylvestris Medicago sativa * Papaver rhoeas Pastinaca sativa Phacelia tanacetifolia * Plantago lanceolata Poterium sanguisorba Salvia pratensis Silene dioica Silene flos-cuculi Silene latifolia Silene vulgaris Tanacetum vulgare Trifolium incarnatum * Trifolium resupinatum *	Arctium lappa Barbarea vulgaris Capsella bursa-pastoris Chaerophyllum aromaticum Cirsium arvense Erodium cicutarium Erysimum cheiranthoides Galeopsis tetrahit Hypericum perforatum Knautia arvensis Lamium purpureum Medicago lupulina Papaver dubium Polygonum aviculare Sonchus arvensis Stellaria media Taraxacum officinale Thlaspi arvense Trifolium repens Tripleurospermum inodorum Veronica filiformis Vicia cracca Viola arvensis	Achillea millefolium Aegopodium podagraria Angelica sylvestris Anthriscus sylvestris Arctium minus Barbarea vulgaris Centaurea cyanus Centaurea jacea Cichorium intybus Cirsium arvense Cirsium olearceum Cirsium vulgare Daucus carota Euphrasia officinalis Filipendula ulmaria Galium verum Hypericum perforatum Iris pseudacorus Knautia arvensis Lathyrus pratensis Lycopus europaeus Lysimachia vulgaris Lythrum salicaria Medicago falcata Melilotus albus Melilotus officinalis Papaver rhoeas Pentanema salicinum Ranunculus acris Silene flos-cuculi Silene latifolia Solidago virgaurea Taraxacum officinale Thalictrum flavum Thalictrum lucidum Tripleurospermum inodorum Tussilago farfara Valeriana officinalis Veronica officinalis Vicia cracca Viola arvensis

details the successfully established sown plant species, the naturally occurring entomophilous plants observed within the WFSs, and the native wildflowers present at the control site. A total of 63 entomophilous plant species were observed flowering in the WFS sites; several non-native annual plant species flowered only during the first one or two years following the establishment of the WFSs. In the semi-natural grassland of the control site, 41 entomophilous flowering plant species were recorded.

The plant composition of both WFSs with enhanced floral diversity and the control site with native semi-natural grassland vegetation provided a continuous supply of floral resources from May to early September. The highest abundance and diversity of flowers were observed from June to August (Figure 1a–c).

2.2. Data Collection

Species richness and relative abundance of pollinators were assessed using 250 m long transect walks, with GPS receiver used for length measurement. Such transect length, shorter than the length of the studied WFSs, was selected to prevent oversampling of rarer pollinator species. During the walk, all aculeate pollinators within a 2 m wide band (1 m on each side of the observer) were identified to species where possible, and recorded using a digital voice recorder. Specimens that could not be identified in the field were collected using an entomological net and preserved in ethanol for laboratory verification. Molecular barcoding was applied for specimens of Lasioglossum and Nomada with uncertain identification. The taxonomy of apoid wasps followed the family-level classification proposed by [72], and species names for insects and plants were validated using the GBIF.org database [73].

Transect walks were performed under sunny weather conditions and lasted approximately 30 min, ranging from 20 to 45 min depending on pollinator abundance. In instances where wild pollinator count exceeded 300 specimens before reaching the 125 m midpoint of transect, the transect length was halved and species counts were doubled. Honeybee (Apis mellifera) counts were excluded from the analysis due to high variability in their abundance resulting from hive relocations by beekeepers. The walks were conducted approximately monthly from May to September (five times per flowering season) over four years (2021–2024), thus generating a dataset of 20 counts for each of the three sites.

2.3. Statistical Analysis

Parametric methods were used for species richness analysis, while non-parametric methods (Mann–Whitney and Kruskal–Wallis tests) were employed for abundance comparisons, as the distribution of the latter variable did not follow a normal distribution. Basic statistics (mean, standard error), normality of distribution (Kolmogorov–Smirnov, Lilliefors, and Shapiro–Wilk tests), ANOVA, and non-parametric tests were performed using Statistica version 8 (Statsoft, Tulsa, OK, USA). All other calculations were conducted using R 4.4.2 [74] and RStudio 2024.12.0 [75].

Diversity was estimated using Hill’s ordered diversity index ^qD, as recommended by [76]:

$${}^{q}D={\left({\displaystyle \sum _{i=1}^S{p}_i^q}\right)}^{1/(1-q)},$$

where q ≥ 0 and q ≠ 1;

S is the total number of species in the assemblage; q is the order or the diversity index ^qD; and p_i is the proportion of the ith species in the assemblage;

$$\mathrm{for} q=1, {}^{1}D=\exp \left(-{\displaystyle \sum _{i=1}^S{p}_i\log p_i}\right); \mathrm{for} q \to \infty , {}^{\infty }D=1/\max (p_i)$$

The observed Hill’s index values (Hill numbers) were calculated using Microsoft Access database queries. The evenness of abundance distribution among species was assessed using the ordered relative evenness index, ^qRE = (^qD − 1)/(S − 1) [77], which represents the 3rd class of evenness measures according to [78]. The relationship between the observed sample size and species richness, assessed using rarefaction, and extrapolation, as well as estimated asymptotic species richness (function ChaoRichness) [79,80], estimated asymptotic diversities (functions ChaoShannon and ChaoSimpson), and estimated Hill diversity profiles (function estimateD for sample size 1000), were assessed using the R package iNEXT [81]. The estimated sampling completeness (function Completeness) and relative evenness (function Evenness) were calculated using the R package iNEXT.4steps [82]. In cases where the bootstrap method was used to evaluate sampling uncertainty and confidence intervals, we applied 100 bootstrap replications.

To achieve a more symmetrical visualization of Hill number-based ordered diversity measures, which exhibit contrasting sensitivity to species’ relative abundance, we propose using log₂(q) values for the abscissa instead of q. This approach, similar to that applied by [83] for other biodiversity measures, scales the range of log₂(q) from −∞ to ∞ (corresponding to q from 0 to ∞) in graphical representations of ordered sampling completeness, ordered diversity (^qD), and ordered relative evenness (^qRE). For visualizing the estimated diversity and evenness profiles, we used log₂(q) = 4 as the maximum abscissa value, since the ^qD and ^qRE values for log₂(q) → ∞ could not be calculated. The graphs were generated using the R package ggplot2 [84].

3. Results

3.1. Species Richness and Abundance per 250 m Transect

We recorded a total of 111 aculeate pollinator species, encompassing all six bee families native to the region, as along with six families of apoid wasps and four other wasp families from the superfamilies Chrysidoidea, Pompiloidea, Tiphioidea, and Vespoidea (

Table 2. Observed abundance of aculeate pollinator species per 250 m transect in three assemblages: Flower strip 1, Flower strip 2, and control (see Section 2).

Species	Flower Strip 1	Flower Strip 2	Control	Species	Flower Strip 1	Flower Strip 2	Control
Apidae				Andrenidae
Anthophora furcata	0	1	0	Andrena alfkenella	2	0	1
Bombus bohemicus	1	0	3	Andrena bicolor	0	2	0
Bombus hortorum	4	4	0	Andrena cineraria	0	1	0
Bombus humilis	0	1	1	Andrena dorsata	5	3	0
Bombus jonellus	0	2	1	Andrena flavipes	2	2	0
Bombus lapidarius	3	5	6	Andrena fulvago	1	2	0
Bombus lucorum	4	3	3	Andrena haemorrhoa	1	2	0
Bombus magnus	2	0	0	Andrena helvola	0	1	0
Bombus muscorum	0	1	0	Andrena jacobi	0	0	1
Bombus norvegicus	0	0	1	Andrena minutula	4	1	2
Bombus pascuorum	16	5	5	Andrena minutuloides	56	33	6
Bombus ruderarius	2	1	5	Andrena nigroaenea	1	1	1
Bombus rupestris	1	0	0	Andrena ovatula	0	3	0
Bombus soroeensis	4	3	1	Andrena pilipes	1	1	0
Bombus sylvarum	37	9	17	Andrena praecox	0	0	1
Bombus terrestris	7	35	8	Andrena wilkella	7	3	1
Bombus vestalis	0	1	0	Melittidae
Bombus veteranus	12	2	9	Dasypoda hirtipes	1	0	0
Eucera longicornis	3	4	1	Melitta leporina	8	1	0
Nomada castellana	0	3	0	Bembicidae
Nomada flavoguttata	2	0	0	Gorytes quinquecinctus	0	0	1
Nomada lathburiana	0	1	0	Crabronidae
Nomada marshamella	0	1	0	Crabro cribrarius	2	0	12
Nomada panzeri	0	1	0	Crabro peltarius	0	0	1
Nomada rufipes	1	0	0	Crossocerus podagricus	0	2	0
Nomada succincta	0	1	0	Ectemnius continuus	0	0	11
Megachilidae				Ectemnius fossorius	0	0	4
Coelioxys conoidea	0	1	0	Ectemnius lapidarius	1	0	0
Heriades truncorum	2	3	0	Entomognatus brevis	1	0	0
Megachile centuncularis	0	2	0	Lindenius albilabris	5	3	1
Megachile versicolor	0	1	0	Oxybelus trispinosus	0	2	1
Osmia bicolor	1	0	1	Pemphredonidae
Osmia rufa	2	2	0	Diodontus luperus	0	1	0
Osmia spinulosa	4	2	0	Pemphredon inornata	0	1	0
Halictidae				Philanthidae
Halictus maculatus	3	4	1	Cerceris quinquefasciata	0	2	0
Halictus quadricinctus	1	1	2	Cerceris ruficornis	0	0	1
Halictus sexcinctus	1	0	0	Cerceris rybyensis	1	0	0
Halictus subauratus	6	10	2	Philanthus triangulum	0	1	0
Halictus tumulorum	29	13	40	Psenidae
Lasioglossum albipes	0	0	1	Mimumesa unicolor	1	0	0
Lasioglossum calceatum	4	27	14	Psenulus pallipes	2	0	0
Lasioglossum leucopus	15	5	2	Sphecidae
Lasioglossum leucozonium	1	2	8	Ammophila sabulosa	0	1	0
Lasioglossum morio	1	1	11	Chrysididae
Lasioglossum nitidiusculum	2	0	1	Pseudochrysis neglecta	0	1	0
Lasioglossum pauxillum	24	111	1039	Pompilidae
Lasioglossum quadrinotatum	1	0	0	Ceropales maculata	0	0	1
Lasioglossum sexnotatum	1	0	1	Arachnospila anceps	0	1	0
Lasioglossum zonulum	4	3	4	Tiphiidae
Sphecodes crassus	0	4	0	Tiphia femorata	0	1	0
Sphecodes ephippius	0	1	4	Vespidae
Sphecodes geoffrellus	0	2	0	Ancistrocerus nigricornis	1	0	0
Sphecodes gibbus	0	2	0	Dolichovespula saxonica	0	0	25
Sphecodes pellucidus	1	0	0	Dolichovespula sylvestris	2	0	5
Sphecodes scabricollis	0	0	2	Gymnomerus laevipes	1	0	0
Colletidae				Odynerus melanocephalus	0	2	0
Colletes daviesanus	2	15	0	Odynerus reniformis	1	2	0
Hylaeus brevicornis	0	0	1	Polistes dominula	17	35	13
Hylaeus communis	0	1	1	Polistes nimpha	0	1	8
Hylaeus confusus	2	0	1	Vespula germanica	6	1	2
Hylaeus nigritus	11	8	0	Vespula rufa	2	0	3
				Vespula vulgaris	0	1	12
				Total:	349	419	1311

). Of these species, 98 were identified in the WFS assemblages, while 55 species were found in the semi-natural grassland of the control site. Notably, 56 species were exclusive to the WFS assemblages and 12 species were exclusive to the control site.

Ordered sampling completeness, which represents the percentage of the observed species relative to the estimated total species richness, approached 100% for abundant species (positive Log₂(q) in Figure 2a). For all species (Log₂(q) = −∞), the completeness was 78% and 72% in the wildflower strip assemblages and 57% in the control site. Both this assessment and the Chao richness (

Table 3. Observed and estimated asymptotic species richness and diversity measures: estimator value ± SE (95% confidence interval).

Estimator	Flower Strip 1	Flower Strip 2	Control
Observed richness	62	74	55
Estimated Chao richness	79.2 ± 9.6 (68.2–109.8)	102.2 ± 13.3 (85.7–142.0)	96.1 ± 23.8 (69.4–172.7)
Observed Shannon	28.5	23.3	3.30
Estimated Chao Shannon	32.4 ± 2.4 (28. 5–37.1)	26.7 ± 2.4 (23.3–31.4)	3.43 ± 0.21 (3.30–3.83)
Observed Simpson	16.2	9.92	1.57
Estimated Chao Simpson	17.0 ± 1.4 (16.2–19.7)	10.1 ± 1.0 (9.9–12.0)	1.58 ± 0.04 (1.57–1.65)

) suggested that approximately 17, 28, and 41 additional scarce species remained unrecorded in Flower strip 1, Flower strip 2, and the control site, respectively. The estimated relationship between the number of specimens and the number of species, assessed using rarefaction and extrapolation methods, revealed significantly higher expected richness in the WFS assemblages compared to the control site (Figure 2b).

The mean abundance of aculeate pollinators per 250 m transect was markedly higher in the control site (73 ± 51) compared to the WFSs (21.2 ± 3.4) (Figure 3a). However, this difference was not significant (Mann–Whitney U test: p = 0.19) due to high variability of this abundance in the control site (coefficient of variation: 296.9%). Species richness per 250 m transect was slightly higher in the flower strips (8.1 ± 0.7) than in the control site (7.1 ± 1.1) (Figure 3b), but this difference was also not significant (ANOVA: F = 0.63, p = 0.43). Pairwise comparisons of transect observations on the same day revealed no significant differences between the WFSs and the control site (paired t-tests: t = 0.39, p = 0.70 for Flower strip 1 vs. control; t = 1.39, p = 0.18 for Flower strip 2 vs. control).

3.2. Changes During Summer Season

Neither the relative abundance nor the species richness of aculeate pollinators demonstrated significant variation during the summer season (Figure 4a,b). However, a remarkable fluctuation in abundance occurred at the control site in August, reaching 289 ± 214 specimens per 250 m transect compared to 16.6 ± 3.9 specimens at the WFS sites (Figure 4a). Although the difference was substantial, it was not statistically significant (ANOVA: F = 1.2, p = 0.32). This peak of abundance was primarily due to a high concentration of Lasioglossum pauxillum males on still flowering plants within a narrow strip of minimally managed semi-natural grassland vegetation along a drainage ditch, following the winter rapeseed harvest when the surrounding crop fields lacked blooming plants.

Specifically, abundance of this sweat bee species at the control site in August reached 256 ± 206 specimens (primarily males) per 250 m transect, while the combined abundance of all other aculeate pollinator species was 33.3 ± 8.9 specimens per transect. L. pauxillum was also present in the WFS sites but was more commonly recorded in May and June (6.8 ± 3.2 females per transect) than in August (4.6 ± 1.7, primarily males, per transect).

3.3. Changes over Years

At Flower strip 1, we observed a significant increase in aculeate pollinator mean abundance per 250 m transect over the years, rising from 7.5 ± 3.1 observed specimens in 2021 to 38.4 ± 11.5 in 2024 (R² = 0.34, regression slope B = 10.4 ± 3.6, p = 0.01). Species richness also increased from 5.8 ± 1.9 to 9.8 ± 2.8; however, this change was not significant (R² = 0.13, regression slope B = 1.37 ± 0.89, p = 0.14). Similarly, changes in pollinator abundance and richness at Flower strip 2 and the control site were not significant.

3.4. Diversity

Estimated asymptotic diversity measures (Chao Shannon and Chao Simpson estimates) were higher in the assemblages of the WFS sites compared to the control site (Table 3). The observed ordered diversity (Figure 5a), estimated ordered diversity for a sample of 1000 specimens (Figure 5b), and the relative evenness (Figure 5c) were all markedly higher in aculeate pollinator assemblages from the WFSs compared to the control site. This difference was consistent across the entire range of Hill numbers, reflecting all relative abundance classes from total species richness (log₂(q) = −∞) and those encompassing rare species (negative log₂(q)) to common and dominant species (positive log₂(q)).

4. Discussion

This small-scale case study demonstrated that aculeate pollinator species richness did not exhibit a considerable increase in the WFS site assemblages compared to the control site with semi-natural grassland vegetation (Figure 3b). However, both observed and estimated diversity, along with species distribution evenness, were significantly higher in the WFS assemblages compared to the control site (Figure 5a–c). Therefore, our first hypothesis is partially supported: the WFSs enhanced the diversity of aculeate pollinator assemblages, even though the species richness did not significantly increase. The ordered diversity and relative evenness indices showed that the WFS pollinator assemblages were more diverse across all relative abundance classes, from rare to common and dominant species. These results highlight the positive impact of WFS on the diversity of wild bees and other pollinators, which is a finding consistent with other studies [41,42,60].

An increase in aculeate pollinator abundance over the four years following its establishment was recorded at one of the two WFS sampling sites, which is a trend also observed in previous studies [33,48,60,85]. However, this pattern was not evident at the second WFS site; therefore, our second hypothesis is not strongly supported. We also did not observe a decline in pollinator abundance or diversity within the WFSs over time due to reduction in floral diversity, as reported in another study [38]. Since the impact of sown flower mixtures on pollination services increases with flower diversity and the time since establishment [31], our findings underscore the role of WFSs in enhancing the surviving pollinator diversity, thus likely supporting the pollination ecosystem services in agricultural environments. We suggest that aculeate pollinator populations, already present in small patches of semi-natural grasslands in the agricultural landscapes surrounding the study site, are able to quickly colonize the established WFSs and exploit new floral resources. As a result, a measurable increase in pollinator abundance was observed at one of the study sites.

The remarkably high abundance of Lasioglossum pauxillum recorded during late-summer transect counts at the control site, following the rapeseed harvest in surrounding more or less distant fields, suggests that particular species may benefit from flowering crops, relying on semi-natural vegetation patches as a refuge during periods of floral scarcity in agricultural environments. Therefore, we consider these observations to support our third hypothesis. This finding aligns with prior research [86], which demonstrated a positive correlation between bumblebee densities and the availability of mass-flowering oilseed rape. The comparatively lower abundance of L. pauxillum per transect count in the WFSs compared to the control site, despite the proximity of rapeseed fields, could be attributed to the larger area and greater floral abundance within the WFS, likely leading to a dilution effect and thus a lower concentration of L. pauxillum specimens.

Furthermore, our results suggest that semi-natural or minimally managed vegetation patches along roads or drainage ditches in intensively farmed landscapes can provide critical resources for pollinator populations during periods when flowering crops, such as rapeseed, are no longer in bloom. Semi-natural habitats have been identified as key predictors of pollinator species richness [87]. Although the remnants of such habitats in agricultural environments may support a narrower range of pollinator species compared to WFSs, as evidenced by our control site, they still play a vital role in conserving bee diversity [88]. Our findings corroborate existing evidence that allowing entomophilous weeds to persist within agricultural environments can provide benefits for flower-visiting insects comparable to those of WFSs [89,90,91]. Furthermore, they do not contradict the assertion that preserved semi-natural habitat patches may sustain greater pollinator abundance than WFSs or mass-flowering monocultures, such as sunflower [92].

5. Conclusions

In intensive agricultural environments, wildflower strips enhanced with species-rich flowering plant seed mixtures (in our case, 50 plant species) can provide floral resources for diverse pollinator assemblages. These assemblages exhibit higher species diversity and relative evenness of abundance distribution compared to pollinator assemblages in patches of semi-natural vegetation.

In this study, the abundance and species richness of aculeate pollinator assemblages remained relatively stable over four years of observations following the establishment of the WFSs. In one case, abundance increased, although the growth in species richness was not significant. These findings suggest that pollinator populations are present in agricultural landscapes, exploiting floral resources from mass-flowering crops, such as rapeseed, and remnants of semi-natural vegetation. These populations can immediately colonize the WFSs and benefit from their establishment.

Mass-flowering crops may serve as suitable resources for certain wild pollinator species, driving their population growth. During periods when mass-flowering crops are not in bloom, these pollinators are able to persist within the semi-natural grassland vegetation patches; therefore, together with WFSs, these patches can play a critical role in sustaining pollinator populations. Consequently, both the establishment of WFSs and the conservation of semi-natural, minimally managed habitats are essential for maintaining the ecosystem service of pollination in intensively managed agricultural systems.