The nectar report: quantitative review of nectar sugar concentrations offered by bee visited flowers in agricultural and non-agricultural landscapes

Introduction

Pollinators are an integral part of natural as well as agricultural ecosystems, with the majority of flowering plants relying on their ecosystem services (Ollerton, Winfree & Tarrant, 2011). Over the past decades bee pollinators have received particular attention, following the realization that some populations seem to be declining (Biesmeijer et al., 2006; Potts et al., 2010; Ollerton et al., 2014). While managed honey bee populations are declining in only a few geographic regions and over certain time periods (Moritz & Erler, 2016), the focus of concern has recently extended to wild bees (Roulston & Goodell, 2011; Goulson et al., 2015; Vaudo et al., 2015). Numerous potential drivers for this proposed dynamic have been put forward including changes in land use, agricultural intensification, habitat loss or fragmentation and emerging pathogens (Brown & Paxton, 2009; Winfree et al., 2009; Goulson et al., 2015). While all these factors likely contribute to some degree, changes in flower provided food resource for bees has emerged as prime candidate directly regulating bee populations (Roulston & Goodell, 2011). Bees and their larvae almost exclusively rely on flower derived nutrients, namely nectar as their primary source of carbohydrates and pollen for protein, lipids, and other micronutrients essential for development, health and survival (Michener, 2000; Brodschneider & Crailsheim, 2010; Roulston & Goodell, 2011). Large scale changes in land-use can alter the quality, abundance and availability of relevant flower derived resources, which in turn can result in nutritional mismatch leading to nutritional stress for bee populations with potential adverse effects (Potts et al., 2010; Roulston & Goodell, 2011; Goulson et al., 2015). For example, while bee pollinated crops might provide a plethora of flower derived resources during their flowering period, the lack of alternative food sources in monocultural dominated agricultural settings, might put a strain on bee species foraging outside the flowering period.

In order to ease nutritional stress on manages as well as wild bee populations in agricultural settings the establishment of complementary foraging habitats has been incentivized via agro-environmental management schemes in the EU and elsewhere (Phillips & Lowe, 2005; Vaughan & Skinner, 2008; Lye et al., 2009; Goulson et al., 2015; Potts et al., 2015). Such schemes were originally intended to provide bees with complementary flower resources outside the mass flowering periods of commercial crops, but have traditionally been intended to support social bees i.e., Bombus sp. (Vaudo et al., 2015). Only recently the effects of nutritional enhancement on wild bees which often have different habitats and nutritional requirements have come more into focus (Scheper et al., 2015). Besides quantity and timing the quality of floral resources, including total sugar content and sugar concentration have direct fitness consequences for social bees (Brodschneider & Crailsheim, 2010; Vaudo et al., 2015; Vaudo et al., 2016), but likely also solitary bees at least to some degree. Consequently qualitative aspects of nectar resources should be taken into consideration when developing management plans for complementary and high quality nutritional bee resources (Vaudo et al., 2015).

As a first step to facilitate the integration of flower resource quality in pollinator management we have used literature data to compile a geographically explicit database of nectar quality (measured as total sugar concentration) provided by bee visited flowers in an agricultural and natural setting. Given that nectar is the main carbohydrate source for adults as well as developing bees, sugar concentration is directly linked to the amount of sugar bees can extract from flowers and has traditionally served as a proxy for nectar quality (Roulston & Goodell, 2011; Vaudo et al., 2015).

We use the compiled database to compare the quality and quality variability of nectar resource bees can encounter in agricultural landscapes in- (crop and weeds) and off-field (wild) around the globe. In a second step we utilize a unique historical data set to analyze the influence of local habitat and water stress on nectar sugar concentrations and their variability. In a last step we identify crop genera, which provide sub-optimal nectar quality and suggest plant groups which could be used to nutritionally support bee populations in agricultural landscapes during times of need.

Materials and Methods

Analysis

Results

Nectar quality & variability

Overall nectar concentration in all regions were comparable around a median value of 40% sugar concentration (see Fig. 1, Table 1) and no significant differences between crop, weed or wild plant communities were found globally (KW chi² = 3.2, p = 0.2) or within the different geographic regions (all KW chi² <4.48, p > 0.11; see Fig. 1 and Table 1). In contrast to the median concentrations we find that the three plant communities differed in the variability of nectar quality (Global community; Fligner test chi² = 31.97, p < 0.001). This effect is mainly driven by an increased variability of the wild community (see Fig. 1) which differs significantly from the crop community on a global level (Bonferroni corrected pairwise Fligner test crop × wild chi² = 30.64 , p < 0.001), with a similar trend in the same direction when compared to the weed community (Bonferroni corrected pairwise Fligner test; crop × weed chi² = 5.02 , p = 0.08). In contrast we find that crop and weed species clearly do not differ in terms of their variability (Bonferroni corrected pairwise Fligner test weed × wild chi² = 1.01, p = 0.93). When comparing the variability of nectar quality on a species level we find that we had only a limited number of species with multiple nectar measurements (N > 2) recorded (crop N = 18, weed N = 6 and wild N = 18). Using this limited data set we find no indication of intrinsic difference in variability (measured as SD) of plant species belonging to the three different plant communities (KW chi² = 2.52, p = 0.28).

Regional effects on nectar quality and its variability

When reanalyzing the Simidtschiev (1988) sunflower dataset comparing nectar sugar concentrations we find that geographic location has a small (Median_Toschevo = 30.7%, Median_Plovdiv = 36.05% see Fig. 2), but significant influence on nectar concentrations (paired Wilcoxon test: V = 216, p < 0.0001 see Fig. 2) and its variation (Fligner test chi² = 6.12, p = 0.01 see Fig. 2). When looking at the effect of non-irrigation (natural rainfall) on nectar sugar concentration of the four tested sunflower varieties the original analysis of Simidtschiev (1988) concludes that in three of the four varieties watering did not significantly influence nectar sugar concentration and in the fourth cultivar (hybrid 260) it only decreased it by about 3.4% (mean irrigation = 50.3, mean natural rainfall = 53.7%).

Nectar quality on a genus level

In total we recorded multiple measurements for 12 crop and 16 non-crop genera and find that there is a significant difference in nectar sugar concentration between them (KW chi² = 149.23, p < 0.0001 Fig. 3). When comparing the nectar quality according to our categorization (see above) our results indicate that two crop genera, namely Capsicum (including paprika and chili) and Pyrus (pear), offer low quality nectar (median sugar concentration <20% (see Fig. 3). When looking at the genera offering high quality nectar we were able to identify 15 Genera which provide optimal nectar concentrations (35%–65%) for bees (see Fig. 3).

Discussion

In this study we have compiled the first comprehensive data set on nectar quality provided by bee visited plants in agricultural landscapes around the globe. Our data indicates that nectar sugar concentrations in bee visited flowers is strongly conserved across all communities and geographic regions with a median value around 40% (see Fig. 1 and Table 1). In addition, we find that wild plants exhibit stronger variation in nectar concentrations at the community level when compared to crop plants. However, this difference is not reflected on the species level. Using a comprehensive data set from the German literature (Simidtschiev, 1988) on sunflower varieties we find evidence that microhabitat (e.g., water availability) and geographic region might have a more limited effect on nectar sugar concentration and its variation than previously thought. Using the complete data set we identify two crop genera (Capsicum and Pyrus) which provide low quality nectar to bees during the flowering period and suggest 15 possible genera which provide high quality nectar as potential supplement nectar source.

When looking at the recorded nectar sugar concentrations we find strong support for the well-established idea that flowers are under strong selection pressure to provide nectar suitable for their respective pollinators (Baker, 1975; Harder, 1986; Perret et al., 2001). In the case of bees, the literature suggests optimal values ranging from 35–65% which is well supported by our data (Waller, 1972; Harder, 1986; Kim, Gilet & Bush, 2011). It has been suggested that the pollinator preferences for different sugar concentrations could be explained by different modes of nectar intake, which in case of bees favours higher viscosity and consequently sugar concentrations (Kim, Gilet & Bush, 2011). Our results therefore indicate that nectar quality between different plant communities in agricultural landscapes and their surroundings are (1) closely matched to pollinators needs (2) comparable between all regions and (3) in principal likely sufficient to maintain healthy bee pollinator populations.

In contrast to the median sugar concentrations we find elevated nectar quality variability of the wild plant community compared to the crop and likely weed plants communities (see Fig. 1). However, based on the results of our limited species level data set we have no indication that this pattern is reflected on the species level where plants of the different communities exhibited comparable variability. In particular the species belonging to the weed community are interesting in this regard as they are wild species which grow under (invade) standardized agricultural conditions. A priori we could expect that the more standardized growing conditions in agricultural could reduce nectar quality variability when compared to natural habitats. As we do not find differences between these three groups on the species level our results support the view that growing conditions might have a more limited influence on the variability of nectar sugar concentration and supports a plant species specific nectar concentration of at least some species. In turn this suggest that the observed variation on the community level likely reflects the elevated phylogenetic diversity in wild plants compared to crop and weed species (Meyer, DuVal & Jensen, 2012). Indeed, when looking at the community level, we find the wild community containing more than twice the number of wild genera (N = 63) compared to the crop (N = 29), as well as the weed community (N = 18), which seems the most likely explanation for the observed pattern.

The comprehensive study of Simidtschiev (1988) offers the unique opportunity to study the effects of geography and water availability on nectar sugar concentration. These results support our initial findings that geography, in this case only has a limited absolute effect on nectar sugar concentration and its variation (See Fig. 2). In particular the fact that plants “defended” their nectar sugar concentrations against variation in water availability suggest that at least sunflowers have plant species specific nectar sugar concentrations. In his original analysis Simidtschiev (1988) suggests that instead of changes in nectar concentration, nectar volume responds to reduced water availability, which in turn reduce the caloric value and consequently resource quality for bees. It would be interesting to analyses what parameters best explain the observed variation (including temperature, rainfall soil types ect.) because such factor undoubtedly play a role in shaping nectar concentrations to some degree (Corbet et al., 1979). Unfortunately, these parameters were not recorded by Simidtschiev (1988) for the study duration. This highlights the importance of multiple measurements to adequately characterize resource quality for bees. In a next step it would be very important to include such measurements into the data base to provide a more detailed picture of nectar quality to serve as a robust basis to improve the resource quality offered to bee populations in the future.

In our quality analysis we have identified two genera of crop plants, which provide low quality nectar sugar concentrations. While it is well-known that pears (Pyrus) are not considered attractive to social bees due to their low nectar (Fig. 3) quality (Maurizio & Grafl, 1980) the even lower levels in Capsicum (e.g., paprika and chilies) suggest that these crops will likely not be able to sustain managed bee populations on their own. Such information, coupled with detailed information regarding geographic abundance of these crops, could be used to identify potential targets for a resource quality intervention such as agro-environmental scheme. We suggest 15 plant genera which provide adequate nectar quality for bees and could be used as feeding supplement. For example, pear orchards might be a good option for such interventions as the supplemental nectar resources could be located close by, which in turn could attract and retain social as well as solitary bees which use pear flowers as a source of pollen (Westrich, 1989).

Our study is a first step towards the integration of resource quality in bee conservation practices. The obvious next step could be to include additional quality markers for nectar including nectar volume, sugar composition and non-sugar components, which clearly play an important role in determining nectar quality for bees (Vaudo et al., 2015). In particular nectar volume and flowering season are promising next steps to get a more complete picture of nectar quality because together concentration and volume define the caloric value of individual flowers, entire plants and ultimate vegetation types. The caloric value of flowers is likely important for most bee species, however most studies directly linking nectar quality to bee fitness stem from social bees such as honey and bumblebees. Currently it is less clear how important these factors are for solitary bees’ fitness considering their deviating ecology. Such information would be useful to specifically address the needs of solitary bees. A second important step would be to compile a similar data set for the second important flower resource pollen and its quality markers such as crude protein content, amino acid composition or lipid content (Roulston, Cane & Buchmann, 2000). Such information should be combined with detailed geographic and information on exact flowering periods in order to estimate the resource availability and quality during the season in a given location and plan and implement targeted interventions to support bee populations.

Such a tool could support farmers, scientists, regulators NGOs and the industry when designing optimized alternative flower resources for bees in agricultural landscapes. We hope that this data set will serve as a starting point to help facilitate a data informed discussion about pollinator conservation in agricultural landscapes between all relevant stakeholders and will ultimately help to reduce nutritional stress for bee populations in modern agricultural landscapes.

Conclusion

In this study we conducted the first systematic review of nectar sugar quality bees can encounter in agricultural landscapes in- and off-field around the globe. We report that nectar sugar concentrations do not vary between regions or habitats with a median of around 40% nectar total sugar concentration. We have identified several crop genera providing nectar with sub optimal sugar concentrations for bees which could potentially benefit from alternative nectar sources during their mass flowering period. This dataset is only a first step toward integrating nectar sugar concentrations into bee management practices and we hope that this data resource will facilitate communication between all relevant stakeholders and ultimately help to reduce nutritional stress for bee populations in modern agricultural landscapes.