**7. Consideration of the landscape factors in agro-environmental actions promoting flowering areas**

Some authors have questioned the effectiveness of agro-environmental schemes because their beneficial effects on target species do vary from one study to another (Kleijn et al., 2006). The same holds true for floral landscape enhancements dedicated specifically to bees (Dicks et al., 2010). Yet there is a growing body of evidence that the seemingly unpredictable effectiveness of floral schemes is actually dependent on the landscape context in which it is established (Decourtye et al., 2010). Landscape context typically refers to the degree of land use by humans. Intensive, simplified, agricultural landscapes are distinguished from complex landscapes with greater amounts of semi-natural habitats or habitat diversity. A handful of studies have specifically measured the influence of landscape context on the efficiency of experimental flower patches in attracting bees (Heard et al., 2007; Kohler et al. 2008; Steffan-Dewenter et al., 2002). However, to date, no consistent conclusion can be drawn on such floral schemes because different field protocols have been used with different study species and at different spatial scales.

Only one of these studies (Steffan-Dewenter et al., 2002) focused on honeybees and it reported a weak pattern of context-dependent floral scheme effectiveness. Authors have implanted experimental flower patches in a variety of landscape contexts, and have monitored honeybee foraging activity at those patches. Flower patches tended to attract fewer honeybees when implanted in landscapes characterized by a higher amount of seminatural habitats (field and forest margins, hedgerows, fallows and extensive grasslands) within a 3-km radius. Conversely, foraging activity at flower patches located in more intensive agricultural landscapes was greater, indicating that honeybees compensated for the lack of natural resources by making a disproportionate use of the floral schemes.

Yet, theoretical evidence suggests that honeybees would rather respond positively to the presence of semi-natural habitats. In particular, hedgerows, forest margins and other linear landscape elements may be used as visual landmarks by honeybees to direct their flight

Why Enhancement of Floral Resources in Agro-Ecosystems Benefit Honeybees and Beekeepers? 381

the feeding of sugar syrup solutions. These cannot fully replace natural nectar in terms of nutritive value. Consequently, making sure that honeybees have access to pollen and nectar at the right time in their natural environment, remains the best way to guarantee colony survival. Moreover, recent results reported in literature (Alaux et al., 2010; DeGrandi-Hoffman et al., 2010) highlight a connection between diet, protein levels and immune response and suggest that colony losses might be reduced by alleviating diet stress through

The goal of agro-environmental measures aiming to favour honeybee protection should be to temporally and spatially increase heterogeneous habitats for these insects (Decourtye et al., 2010). As a first goal, it is urgent to preserve current semi-natural habitats in farmlands (hedgerows, woodlands, ponds, ditches). In addition, regulations advocating the management of uncropped farmlands (field margins, farmland set-aside) for increased floral availability to bees could contribute to the maintenance of colony viability (at risk today, especially in intensive cereal farming areas). The management of uncropped farmlands designed to introduce floral resources is likely to benefit other pollinators (wild species) and sustain beekeeping activities (domestic species). But the economic question has to be addressed in relation with farmers as well, depending on the possibility to derive any agronomic benefit from floral schemes. Land managers have to seek a trade-off between the costs of introducing floral schemes, and the benefits they might receive from it because of enhanced pollination service. Yet, enhancement of farmland set-aside may require disproportionate financial investments in some areas – generally where farming practices are intensive (Kleijn et al., 2009; Van Buskirk & Willi, 2004). Therefore, it appears critical to gain a prior knowledge about where enhancement schemes are likely to reach maximal efficiency in a given landscape. In that respect, we should study the efficiency of flower enhancement schemes according to local landscape context. This issue has been poorly explored to date (Heard et al., 2007). Reliable modeling tools are needed to delineate the landscape areas likely to be associated with greater effectiveness of floral enhancement

In these intensive farming areas, the conflict between the conservation of flora, to increase the availability of food to bees, and the management of pesticide treatments complicates the reliability of the positive impact of such flowering schemes. The protection of plants producing nectar and pollen in the farmland areas must actually come with a harmless management of pesticides. Thus, rather than any particular farming practice causing current population decline, such as insecticide application or reduction of flora, the multivariate effects of agricultural practices strongly interact and should be considered collectively,

This work is a product of UMT PrADE, a unit which associates different groups of stakeholders towards bee protection, from researchers to apicultural engineers. It was financially supported in part by a grant from the French Ministry of Agriculture (MIRES). The research on the melliferous and polleniferous flora and pollen nutrients was supported by grants from FEAGA, L'Institut National de la Recherche Agronomique (INRA "Département Santé des Plantes et Environnement"), and a French-Romanian bilateral

landscape enhancement of floral resources in agro-ecosystems.

schemes, specifically in intensive agricultural systems.

rather than individually.

**9. Acknowledgements** 

program (PHC Brancusi).

path and locate food sources (Chittka & Geiger, 1995; Dyer, 1996; Dyer et al., 2008). As such, linear landscape elements are generally considered to promote landscape connectivity (Taylor et al., 1993), i.e. to facilitate movement of organisms among their resource patches by forming flight corridors (Townsend & Levey, 2005; Van Geert et al., 2010).

The mismatch between theoretical expectations of higher foraging activity in complex landscapes on one hand, and the opposite pattern actually observed by Steffan-Dewenter et al. (2002) on the other hand, most probably reveal an interplay of behavioral processes acting at different spatial scales. The study by Steffan-Dewenter et al. (2002) reports broadscale patterns, with landscape context characterized within 250-m to 3-km radii around experimental plots, i.e. a spatial window scaled on the foraging range of honeybee colonies. However, the facilitative effect of visual landmarks for the orientation of foragers probably acts at a very local scale. Perception and orientation in honeybees imply short-scale behavioral processes. For instance, an experimental modification of landmark arrangements within a range of only several tens of meters altered the capacity of honeybees to retrieve food sources (Chittka & Geiger, 1995).

Recent foraging surveys conducted in an intensive cereal farming system support the previous hypothesis that semi-natural and/or linear landscape elements enhance the use of floral schemes by honeybees at a local scale (Henry et al., in press). Honeybee foraging activity was monitored in 170 plots (22m) of *Phacelia tanacetifolia* Benth. scattered among 17 field margins, all positioned within the foraging range of a single experimental apiary. The presence of hedgerows and forest edges within a 160-m radius around the phacelia plots had a strong positive effect on honeybee foraging activity. Conversely, the presence of mass flowering crops within the same distance had a negative effect, probably because they exerted a concurrent attraction. This critical range of 160 m was not covered by the study of Steffan-Dewenter et al. (2002).

Altogether, the above-mentioned studies suggest that the effectiveness of floral enhancement schemes varies with landscape context in a multi-scale fashion. On a broad scale, i.e. at the landscape level envisioned by the foraging range of a honeybee colony (3 km grain), floral schemes will be more successful in intensive agricultural landscapes (with <15% of land cover occupied by semi-natural habitat remnants). At a smaller scale, within intensive landscapes, floral schemes will be more successful if they are implanted nearby (< 160 m) existing networks of hedgerows, forest edges, or other salient semi-natural elements. Priority should be given to areas farther (> 160 m) from other existing mass-flowering crops. These tentative recommendations still need further validation, depending on the type of floral schemes.
