Flower strips are often proposed as a tool to mitigate biodiversity loss in agricultural landscapes (Haaland et al., 2011), especially loss of functional biodiversity such as pollinators (Scheper et al., 2013) and natural enemies of pests (Holland et al., 2016). Supporting biodiversity-related ecosystem services has the potential to increase agricultural production while minimizing negative environmental impacts (Bommarco et al., 2013). Flower strips can be targeted for pollinators (Wood et al., 2015), natural enemies (Tschumi et al., 2015), both pollinators and natural enemies (Campbell et al., 2017; Grab et al., 2018) or serve different purposes (Vickery et al., 2002). Local factors, including plant community composition, size and shape of strips, configuration (e.g., field edge or field interior), can have important positive or negative effects on flower strips’ capacity to support functional biodiversity and ecosystem services in croplands (Haenke et al., 2009; Jönsson et al., 2015; Uyttenbroeck et al., 2015; Wood et al., 2016). However, the consequences of agri-environmental measures such as flower strips, on functional diversity may be context dependent, and related to the agricultural intensity in the surrounding landscape (see e.g., Tscharntke et al., 2005, 2016; Bianchi et al., 2006; Scheper et al., 2013; Grab et al., 2018). Thus, the re-occurring critique toward the European Union’s (EU), Common Agricultural Policy (CAP), that measures need to be targeted (e.g., Smith et al., 2010; Batáry et al., 2015; Pe’er et al., 2019) and based on a landscape perspective (see e.g., Tscharntke et al., 2005; Batary et al., 2011; Kleijn et al., 2011), applies also to flower strips.

In an attempt in the reform 2013 to make the CAP “greener,” and with the objective to safeguard and improve farmland biodiversity in Europe, for many farmers so called Ecological Focus Areas (EFA) became mandatory (through Pillar 1) in 2015 (while the budget for the agri-environmental schemes in Pillar II decreased) (EU, 2013). As one out of three “greening measures” (the others being crop diversification and maintaining permanent grassland at a national level), EFAs were supposed to be simple, generalized, non-contractual, and annual, and can consist of areas such as nitrogen fixing crops and uncropped field margins (EU, 2013). EFAs in the form of uncropped field margins has been criticized as inefficient (Nilsson et al., 2019), but also created a window of opportunity for annual flower strips and are encouraged to be used for such (e.g., by the Swedish Board of Agriculture). The question is what kind of effects that can be expected on farmland biodiversity from these annual strips.

Even though evidence do exist for the potential of flower strips and other non-crop habitat to have positive effects on farmland biodiversity and ecosystem services (Dicks et al., 2014, but see Zamorano et al., 2020), there are many factors that influence the magnitude and direction of these effects (Tscharntke et al., 2016; Karp et al., 2018). Most of the existing knowledge comes from perennial or multi-annual flower strips (see e.g., Jönsson et al., 2015; Wood et al., 2016), whereas little is known about the effects of annual flower strips. Exceptions to this are two studies in Switzerland showing local positive effects of targeted annual flower strips on natural enemies (Tschumi et al., 2015, 2016) and one study in Sweden showing local positive effects on species richness of pollinators (Rundlöf et al., 2018). In the two studies from Switzerland, the landscapes were traditional Swizz agricultural landscapes characterized by a small-scaled mosaic of crop fields. The positive effects in the Swedish study were strongest in the more heterogenous landscapes (Rundlöf et al., 2018).

Given the lack of scientific foundation on the use of annual flower strips on EFAs to benefit functional biodiversity, we in this study aimed to evaluate if they benefited natural enemies and pollinators in the surrounding landscape. Given that EFAs are mandatory mostly in intensively farmed landscapes, and that benefit of functional biodiversity for cash crops is most relevant in these landscapes (Nilsson et al., 2019), we evaluated the effect using experimental implementation of annual flower strips in existing EFAs in replicated simple landscapes. However, to determine if the amount of permanent structures in these landscapes modify responses to annual flower strips, we also analyzed effects of limited variation in the proportion of arable fields in these landscapes. In this way we addressed the following research questions: (I) Do EFAs in the form of annual flower strips have a positive effect on the abundance of natural enemies in adjacent fields? (II) Is there any resulting effect on pest abundance? (III) Do annual flower strips have a positive effect on the reproductive output of adjacently nesting solitary bees and wasps?

In this study, we demonstrate that annual flower strips can benefit the abundance of hoverfly larvae. However, we could not detect any other significant positive effects on natural enemies and there was no evidence of a difference in natural pest control as shown by lack of flower strip effects on aphid numbers and parazitation rates.

We detected a weak positive effect of flower strips on the abundance of hoverfly larvae, such that they were more abundant further into the field compared to at control sites. This result is consistent with previous studies on the effect of flower strips on hoverflies (Haenke et al., 2009) and may be caused by adult female hoverflies flying further into the fields to lay their eggs when there is a flower strip present to avoid competition from other hoverflies benefiting from local flower availability or that they were less constrained by energy demands due to higher nectar availability and thus able to fly further into the fields. For parasitic wasps, there was a similar but non-significant tendency. Both adult predators (Carabidae, Araneae, and Staphylinidae) and predatory larvae tended to be more abundant further into the fields, but there was no difference between flower strip and control sites. For trap-nesting solitary bees but not for solitary wasps, there was a non-significant tendency that the number of cells was higher closer to flower strips, but not for wasps (in contrast to previous findings; Hoffmann et al., 2018). However, landscape structure best explained the number of cells for both solitary bees and wasps.

We chose to study annual flower strips (Tschumi et al., 2015) rather than multi-annual or perennial flower strips (Jönsson et al., 2015), because our aim was to evaluate the value of using flower strips as EFAs on uncropped field margins in the current CAP policy design. We therefore selected a seed mixture with more or less alien annual plants (flowering in their first year) and adapted the placement and size of the strips in the study to the farmers’ management plans for EFAs, which we did not modify. In addition, we selected to study flower strips in the most intensively farmed landscapes in our region, because it is mostly in these landscapes that EFAs are mandatory for farmers. The weak or lack of effects in our study, can be a result of these specifics of our study design.

Compared to most other similar measures, such as perennial or multi-annual flower strips or buffer strips along water bodies, annual flower strips are agri-environmental schemes with a very short duration (Albrecht et al., 2020). They are generally sown in the springs and then plowed up in the fall the same year. The EFA “uncropped field margins,” evaluated here, only needs to be present until July 31st in the year they are present (in Sweden). The timing of establishment and flowering as well as termination of annual flower strips may thus be important factors determining their potential to impact organisms in the agricultural landscape. In particular, the flower strips may not be available when many species build up their populations early in the spring or during the previous year, when the annual spring sown EFA flower strips are not yet established and even less flowering. For the solitary bees and wasps, it is likely that our spring sown annual flower strip, flowered too late to provide resources during the period when many of the species are active and lay their eggs, which could explain the rather low numbers of cells in the traps. In the trap nests the bee cells mainly came from the families Hylaeus and Megachile, there were only a few Osmia, Chelostoma, and Heriades. For bees like Osmia bicornis, that are active early in the season (Hofmann et al., 2019) and is one of the most common solitary bee species in southern Sweden, the spring sown annual flower strips most likely flowered too late to be able to provide resources for their offspring. Most species from the families Hylaeus and Megachile are active later in the summer (Hofmann et al., 2019) which explains why these were the two most common groups in the trap nests we put out in the end of June- beginning of July. It should be noted that there can have been positive effects of the flower strips that we failed to detect due to timing and/or number of sampling efforts; since the flower strips were annual, we chose to only sample during the year they were present. Future studies should further investigate the effects on functional biodiversity also in the years after an annual flower strip has been in an area (see Blaauw and Isaacs, 2014).

Another key factor determining the effects of an agri-environmental measure such as flower strips, is the characteristics of the landscape in which it is placed (Tscharntke et al., 2016). The availability of semi-natural habitat is in general important for both pollinators and natural enemies (Shackelford et al., 2013; Rusch et al., 2016; Duarte et al., 2018; but see Karp et al., 2018) and a potential reason for why we did not find any effects of our flower strips on abundance of natural enemies and aphids is that potential source populations of natural enemies in our landscapes were too small (Bianchi et al., 2006; Chaplin-Kramer et al., 2011; Rusch et al., 2016). When a flower strip is annual and the duration of the measure is short, it is likely that the surrounding landscape plays an ever bigger role than for perennial or multi-annual strips where a temporal effect may allow populations to build up over time (Korpela et al., 2013; Jönsson et al., 2015; Ganser et al., 2019). If the flower strips in our study had been present for several years it is possible that they could have had a positive effect on, e.g., abundance of pollinators (Jonason et al., 2011; Jönsson et al., 2015). A sufficient amount of perennial habitat will be required in the surrounding landscape to provide founding populations that can take benefit of the resources provided by the annual flower strip (Tschumi et al., 2015; Tscharntke et al., 2016). In a recent study by Pollier et al. (2019), there were positive effects of wildflower strips on natural enemies in and close to the strips in the year of establishment. However, the average field sized reported in that study (0.5–3 ha) suggests that the landscapes were more heterogenous (Martin et al., 2019) than the landscapes of the present study (with an average field size of 4.6 ha). The positive effects we detected here, were among mobile organisms. Hoverflies, although also affected by availability of semi-natural habitats (Schirmel et al., 2018), have a life history (e.g., not being central placed foragers (Covich, 1976) allowing them to take advantage of newly established habitats even when the landscape is fragmented and rather simple, whereas the less mobile organisms showed no response (as they are not able to move from the few and fragmented habitats that can sustain a founder population).

We did not find an effect of the surrounding landscape on the abundance of natural enemies or the abundance and parasitation rate of aphids. We suggest that the variation in landscape complexity (75–95% arable land) was such that founding populations were consistently too low for annual flower strips to affect their populations (Tscharntke et al., 2007), whereas Tschumi et al. (2015) also did not find any effect of landscape complexity and suggested this was because landscapes were always very complex. The trap nests in our study were placed in landscapes with a broader range of landscape complexity (51–98% agricultural arable fields within 1,000 m) than the sampling of natural enemies was and for both solitary bees and wasps, there was a negative relationship between the number of cells and the proportion of arable fields around the trap nests. Since bumble bees utilize landscapes at larger spatial scales than solitary bees, landscape effects could theoretically have been mediated by interspecific competition, but we did not find any such competition in another study explicitly focusing on this (Johanna Yourstone, personal communication). We are also aware that the proxy we use to describe the landscape is coarse and acknowledge that there may be important aspects of landscape structure with consequences for natural enemies and pollinators that we did not capture.

An important characteristic of our study was that we used EFAs as planned by farmers independent of our study, but then made farmers experimentally implement flower strips on them. However, by strictly adapting our design to the farmers choices of placement of EFAs, resulted in only six of the EFAs (i.e., flower strips) being located on a field adjacent to a cereal field, reducing the statistical power of the study on natural enemies and pests. In addition, because farmers chose to create the flower strips on sugar beet, oil seed rape and onion fields, the flower strips and the focal fields were separated by a small road, which might have been enough to hamper spill-over from the flower strips, since such effects often are restricted to only a few meters from the flower strips (Pollier et al., 2019). Furthermore, while studying potential effects of the flower strips on natural enemies and aphids we used existing field edges as controls, which differed in appearance and structure, from broad with ample vegetation to narrow with less vegetation, potentially creating a noise hiding any effect of flower strips in the paired design. In addition, except for a small area around transects, pesticides were freely applied on fields, which may have reduced populations able to take advantage from flower strips. In fact, pesticides may even drift from the field into the transects (Ricci et al., 2019). However, regardless of these limitations, we found little evidence that EFAs with annual flower strips as planned by farmers enhance populations or ecosystem services provided by pollinators or natural enemies in intensive agricultural landscapes.

Annual flower strips have become increasingly common in agricultural landscapes. They are agri-environmental measures with potential to increase farmland biodiversity and ecosystem services (Tschumi et al., 2015), as well as being flexible and relatively easy to implement for farmers (Nilsson et al., 2019). Nevertheless, the results from our study highlights that their potential for promoting functional biodiversity such as pollinators and natural enemies, especially in simple landscapes, is limited but also needs to be studied more carefully. For overwintering arthropods annual flower strips, plowed up in the fall, may even act as ecological traps (Ganser et al., 2019). As other studies before, we stress, the importance of basing conservation and agri-environmental measures on scientific evidence (see e.g., Dicks et al., 2014). In intensively managed and ecologically simple agricultural landscapes, where perennial habitat is scarce, the annual flower strips would most likely have a better effect if combined with more permanent agri-environmental measures (Krimmer et al., 2019). This type of management arrangement is something that future research needs to study more specifically. We chose to study annual flower strips because of the annual character of the EFAs and their potential to be more appealing to farmers than permanent measures, though, for long-term positive effects on biodiversity in farmland, networks of perennial or at least multi-annual flower strips (Korpela et al., 2013; Jönsson et al., 2015; Ganser et al., 2019; Albrecht et al., 2020) and semi natural habitats such as semi-natural grasslands (Ekroos et al., 2013) have considerably greater potential.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

HS received the funding. LN, BK, and HS designed the study. LN and BK did fieldwork and data collection. LN wrote the manuscript with input from HS and BK. All authors conducted the statistical analysis.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.