Introduction

Declines in on-farm insect biodiversity as a result of agricultural intensification have been well documented in recent decades, with losses attributed to reduction and fragmentation of semi-natural habitats, simplification of crop rotations and high inputs of herbicides and pesticides (Pfiffner and Wyss 2004). Attempts to reverse this decline are currently at the forefront of both insect conservation and Integrated Pest Management (IPM) as researchers have demonstrated the pivotal role that agro-biodiversity can play in providing eco-system services to agriculture. Such services include enhancement of pollination, pest control, nutrient cycling, general conservation and carbon sequestration (Altieri 1999; Catovsky et al. 2002; Gurr et al. 2003), with the future maintenance of these services in an unpredictable climate being better buffered where diversity is higher (Altieri 1999).

Non-crop vegetation in agricultural landscapes can provide a means of conserving insects in farmed landscapes and optimising on-farm ecosystem services as a result. Flowering field margins can be used to harbour such vegetation, with margin seed mixes already developed to target bees (Carvell et al. 2006), butterflies (Pywell et al. 2004) and pest natural enemies (Landis et al. 2000). Research is currently seeking to develop margin seed mixes capable of conserving multiple insect groups simultaneously (George et al. 2010), as cross-benefits are not necessarily achieved with more targeted options. Non-crop elements that are designed for pollinator conservation, for example, often do not simultaneously make resources available to pest natural enemies (Olson and Wäckers 2007). This is because not all flowers are suitable for supporting pest natural enemies, despite a broad range of biological control agents depending on flowering vegetation as a source of nectar and pollen (Wäckers 2004; Lavandero et al. 2006; George et al. 2010). The flowering component of field margins therefore needs to be carefully considered if benefits to multiple functional insect groups (e.g. pollinators, pest natural enemies) and the ecosystem services they provide are to be realised (George et al. 2010). This is equally true to minimise resources provided to flower-feeding pest insects, which may benefit from nectar, pollen or alternative host plants present in field margins (Lavandero et al. 2006; Winkler et al. 2010).

The above in mind, it is important to determine which flowering species are best placed for inclusion in field margin seed mixes. Assessing the attractiveness of flowers to functional insect groups is a means of achieving this aim, though results are often limited by consideration of single flowering plants (or small patches of plants) over a short period. This is in spite of the fact that flowers may be numerically abundant in field margins, with individual species differing in the temporal availability of floral resources they provide. The current study aimed to assess floral attractiveness to a selection of functional insect groups using large stands of flowers grown on a commercial scale. Furthermore ‘total temporal attractiveness’ was assessed by considering cumulative numbers of insects attracted to flowers over their entire period of inflorescence.

Materials and methods

Twelve single-species flower crops, typical of those included in field margin seed mixes, were sampled in 2010 (see Table 1). Crops were grown by Emorsgate Seeds at their farm in Kings Lynne (Norfolk, UK) as part of the company’s normal commercial practice. When flowers began to bloom, yellow plastic water traps (27 cm diameter, Nickerson Bros. Ltd, UK) were placed at canopy height in the centre of each field to sample insects moving above the crop in the immediate vicinity of the trap whilst minimising edge effects. Traps were changed weekly during the period of inflorescence for each plant species, and the insects collected preserved in 70 % ethanol for later processing. Sampling ceased in each crop at the time of harvest. Though use of yellow water traps could have been expected to influence trap attractiveness to some target groups more than others (Hoback et al. 1999), any resulting bias in trap contents would have been equal across all treatments (i.e. flowering crops).

Table 1 Flowering and grass species sampled during the experiment with details on crop sizes and flowering periods provided
Full size table

For all insects, numbers trapped were summed over the entire sampling period to obtain the ‘total temporal attractiveness’ for each plant species. For two of the flowering species included in the study, data was missing for 1 week (see Table 1) due to trap contents being unrecoverable. These data were corrected by multiplying the numbers of insects trapped by SP/SP-7 (where SP = number of days for which data were collected). All insects recovered from traps were identified to Order, with pre-determined sets of species taken further taxonomically and assigned to functional groups according to Table 2. Functional groups used were: pest; pest natural enemy; pollinator; decomposer; conservation (for groups of either direct conservation interest—i.e. butterflies and bees–or particular benefit to such groups as food items—i.e. Hemiptera: Heteroptera as food for numerous farmland birds) and Banker (for groups of particular benefit as prey and/or hosts to Pest Natural Enemies—i.e. aphids).

Table 2 Functional groups used into which sets of insects were classified according to the taxonomic level shown
Full size table

Relationships between ‘total temporal attractiveness’ of plant species sampled (to all insects combined) and both crop size and sampling (i.e. flowering) period, were investigated by Pearson’s product-moment correlation, where data were first checked for normality using the Kolmogorov–Smirnov test. Analysis was not attempted for individual insect Orders as several were often either minimally represented or absent entirely depending upon the plant species sampled. Graphs were built to visually compare the attractiveness of plants to different functional insect groups, though statistical analysis of this data was not possible due to the lack of replication inherent in the experimental design.

Results and discussion

Results demonstrated that there was no significant correlation between ‘total temporal attractiveness’ of flowers to insects and sampling (i.e. flowering) period (P = 0.191, r = 0.387) (Fig. 1). A general trend of increasing insect load with sampling period seemed evident, but with notable exceptions that likely yielded a lack of any relationship between these variables. Despite being sampled for a period of 43 days, for example, L. corniculatus attracted fewer insects overall then several species sampled for less than half this period. C. segetum also appeared notably unattractive in relation to its relatively long sampling period. Conversely, P. veris, which was only sampled for 1 week, attracted more insects overall than several species sampled over much longer periods. As sampling of different plant species was initiated at different times, it is possible that some of this variation could be explained by natural temporal fluctuations in insect populations, though an innate ability/inability to attract insects likely exerted a stronger effect in the majority of cases. Variation in the overall areas of the different flower crops sampled could have potentially effected ‘total temporal attractiveness’, though this was not supported statistically (P = 0.930, r = 0.027).

Fig. 1
figure 1

Total temporal attractiveness of different flowering species to different orders of insects. Plants are displayed in ascending order of sampling period (see Table 1), with sampling periods shown

Full size image

The results also suggest that certain plant species may differentially attract certain beneficial insect Orders during flowering, with some species (e.g. K. arvensis and A. millefolium) appearing to be more ‘generalist’ in their attractiveness than others. This was supported by further classifying sets of insects into functional groups, with K. arvensis displaying notably high attractiveness to parasitoid wasps, aphids, bees and Lepidoptera and A. millefolium traps recording high levels of predatory beetles (primarily soldier beetles), all Hemiptera (i.e. aphids and other bugs), hoverflies and Lepidoptera (Fig. 2). L. vulgare also appeared to display high generality in attractiveness to beneficial insects, with relatively high catches of parasitoids, hoverflies and Lepidoptera (Fig. 2). Work elsewhere has found K. arvensis flowers to be preferred by bees, butterflies and burnets (Franzén and Nilsson 2008). Similarly, A. millefolium has been independently found to benefit pest natural enemies, for example by increasing longevity of hoverflies (Langoya and van Rijn 2008). This was also found to be the case for hoverflies feeding on C. cyanus (Langoya and van Rijn 2008), with this species also attracting hoverflies in the current study (Fig. 2).

Fig. 2
figure 2

Total temporal attractiveness of different flowering species to different functional groups of insects (see Table 2). Plants are displayed in ascending order of sampling period (see Table 1), with functional groups shown a predatory Coleoptera, b pest Coleoptera, c parasitoid Hymenoptera, d Apidae, e Syrphidae, f Sarcophagidae, g Lepidoptera, h Hemiptera

Full size image

Though relatively little attention is paid to insect decomposer groups in the literature (as compared to pollinators and pest natural enemies), nutrient cycling is a key ecosystem service in agriculture (Altieri 1999). Along with hoverflies, C. cyanus also attracted large numbers of necrophagous flesh flies (along with V. sativa and to a lesser extent, L. vulgare) (Fig. 2). These results combined suggest that certain flowering plants could be used in field margins to benefit multiple functional insect groups, thus optimising the combined ecosystem services they provide, from pest control and pollination, to conservation and decomposition.

As both K. arvensis and A. millefolium attracted high numbers of aphids, it is possible that aphid-feeding pest natural enemies (e.g. hoverflies, parasitoid wasps and predatory beetles) were responding to cues from their aphid hosts as well as those of flowers. With both flowers and aphids potentially attracting (and supporting) these natural enemies, such species may hold strong potential as a banker plants in field margins, supporting in-crop pest control as a result. Indeed, A. millefolium is already being considered for this role in field margin seed mixes (George et al. 2010).

As well as attracting beneficial insects, the results suggest that some flowering plants may be more likely to attract proportionally high levels of pests, which may make such species less suited to agricultural field margins in some settings (Lavandero et al. 2006; George et al. 2010; Winkler et al. 2010). Both C. cyanus and L. vulgare attracted high proportions of Thysanoptera (many of which are pests), for example, with several species proving to be attractive to pollen beetles (including K. arvensis and A. millefolium, both previously noted as displaying good general attractiveness to beneficial insects) (Fig. 2). The extent of pest attraction to certain flowering species may have been underestimated in the current study as it was generally assumed that the functional groups Lepidoptera, Homoptera and Hemiptera contained insects of either direct or indirect benefit to ecosystem services. This needn’t necessarily be the case, however, with certain flowering plants being associated with pest aphids or providing floral food for pest Lepidoptera (Fitter and Peat 1994; Winkler et al. 2010). In addition, members of the Parasitica may have represented species at the 4th trophic level (i.e. hyper-parasitoids) as well as the 3rd (primary parasitoids), with the former potentially having a negative impact on pest control (Lavandero et al. 2006).

This study has shown that certain flowers, throughout their entire inflorescence, are more likely to attract large numbers of pest and beneficial insects than others, though differences between species are not dependant on flowering period alone. Thus, whilst field margins should seek to provide prolonged floral resources to conserve target beneficial groups per se, achieving this simply through the use of plants with longer flowering periods is unlikely to optimise benefits to functional groups of insects and the ecosystem services they provide. As the study was conducted over a single season without replication, further research is required to confirm the essentially preliminary results obtained, for example through use of a fully replicated randomised block design that retains the large crop areas used. As attraction to a floral resource does not necessarily imply resource suitability (Wäckers 2004), studies to confirm actual flower use would also be beneficial to support the data presented on floral attraction.