Experimental evidence that honeybees depress wild insect densities in a flowering crop

Sandra A M Lindström; Lina Herbertsson; Maj Rundlöf; Riccardo Bommarco; Henrik G Smith

doi:10.1098/rspb.2016.1641

. 2016 Nov 30;283(1843):20161641. doi: 10.1098/rspb.2016.1641

Experimental evidence that honeybees depress wild insect densities in a flowering crop

Sandra A M Lindström ^1,^2,^3,^✉, Lina Herbertsson ⁴, Maj Rundlöf ³, Riccardo Bommarco ¹, Henrik G Smith ^3,⁴

PMCID: PMC5136583 PMID: 27881750

Abstract

While addition of managed honeybees (Apis mellifera) improves pollination of many entomophilous crops, it is unknown if it simultaneously suppresses the densities of wild insects through competition. To investigate this, we added 624 honeybee hives to 23 fields of oilseed rape (Brassica napus L.) over 2 years and made sure that the areas around 21 other fields were free from honeybee hives. We demonstrate that honeybee addition depresses the densities of wild insects (bumblebees, solitary bees, hoverflies, marchflies, other flies, and other flying and flower-visiting insects) even in a massive flower resource such as oilseed rape. The effect was independent of the complexity of the surrounding landscape, but increased with the size of the crop field, which suggests that the effect was caused by spatial displacement of wild insects. Our results have potential implications both for the pollination of crops (if displacement of wild pollinators offsets benefits achieved by adding honeybees) and for conservation of wild insects (if displacement results in negative fitness consequences).

Keywords: crop pollinators, flies, interspecific competition, oilseed rape, wild bees

1. Introduction

Insect pollination is an essential component of global food security because it supports yield in three quarters of agricultural crops [1]. These crops can be pollinated by both wild flower-visiting insects and managed honeybees. Although managed honeybees are important as pollinators of crops, wild flower-visiting insects sometimes provide an additive pollination benefit, irrespective of the density of honeybees [2]. In many parts of the world, wild bee populations have declined, mainly due to habitat degradation and loss caused by land-use conversion, pesticide use, pathogens, and loss of pollen- and nectar-producing flower resources [3], posing a threat to crops dependent on insect pollination [4]. Wild flower-visiting insects other than bees also provide crop pollination [5], but less is known about the factors affecting their densities [3]. Adding managed honeybees to mass-flowering crops is a widespread practice for bolstering pollination, sometimes used to compensate for insufficient numbers of wild pollinators [6], and to produce honey [7]. By moving hives containing large numbers of honeybees, the density of pollinators in the crop quickly increases. However, it is unknown if this affects the density of wild flower-visiting insects in the crop.

Honeybees have probably been managed by European farmers since the Neolithic age [8], and feral populations occurred in northern Europe until the introduction of Varroa mites [9]. Hence, honeybees have a long history of coexistence with wild flower-visiting insects. However, the agricultural landscape has gone through vast changes in the last 80 years, resulting in simplified landscapes with decreased floral resources [10]; because honeybees consume large amounts of floral resources [11], competition among species with similar foraging preferences may have increased [12]. Competition is hypothesized to be strong among similar species [13], and ecologists have therefore explored competition between honeybees and wild bees [14]. These studies often lack proper replication and independence from confounding factors [15,16]. Furthermore, interspecific competition has mainly been studied in artificial environments such as cages [17] or through non-experimental correlational studies [18,19], but not in crop fields. Studies on competition with other flower-visiting insects are scant [20]. Evidence for competition affecting densities of wild pollinators in crops is lacking, in particular where honeybee densities are manipulated in open field experiments.

Interspecific competition may result in spatial displacement, where the dominant competitor forces other species to select less optimal foraging areas or other floral species [20]. The impact of honeybees on wild insects might depend on the availability of alternative foraging habitats in the surrounding landscape (cf. [19]). Alternative foraging habitats can be flowering crops in the landscape. For example, Holzschuh et al. [21] found that occurrence of oilseed rape at the landscape scale diluted the density of bumblebees within oilseed rape fields. In addition, alternative flowering resources are also present in semi-natural habitats. Based on this, the impact of honeybees on wild insect densities is expected to be larger in heterogeneous agricultural landscapes that are rich in flowering habitats, such as semi-natural grasslands and flowering field borders within flight distance, compared with homogeneous intensively cropped landscapes where flower-visiting insects mainly are limited to foraging in mass-flowering crops.

We assessed whether densities of wild insects (bumblebees, solitary bees, hoverflies, marchflies, other flies, and other flying and flower-visiting insects) were affected by competition from honeybees by manipulating honeybee densities in 44 flowering winter oilseed rape fields situated in either heterogeneous or homogeneous agricultural landscape types. We added honeybee hives to 23 of the fields, while we controlled the absence of honeybee hives in the surroundings of the other 21 fields. We aimed to answer the following questions. (i) Is the overall density of flying and flower-visiting wild insects reduced in crop fields supplemented with honeybee hives compared to control fields? (ii) Does honeybee supplementation impact densities of all flying and flower-visiting wild insect groups in the crop fields? (iii) Are the impacts of honeybees on flying and flower-visiting wild insect densities in the crop dependent on landscape context?

2. Material and methods

(a). Study design

We selected 22 conventionally managed winter oilseed rape fields in 2011 and 22 new fields in 2012 in the region of Scania in southern Sweden. All fields were included in a previous study where yield effects from insect pollination were evaluated (except that yield data were not obtained from one field due to severe drought damages on the oilseed rape) [22]. The fields were located in areas dominated by agriculture with more than 50% agricultural land within 1 km radius around fields and without neighbouring orchards. Honeybee hives were added to 12 of the fields in 2011 and to 11 fields in 2012, while 10 fields in 2011 and 11 fields in 2012 were assigned as control fields and checked for the absence from honeybee hives within 3 km around the field (see below; electronic supplementary material, figure S1 and table S1). Honeybee-treated fields were selected based on availability of suitable placement of honeybee hives (e.g. access, security and shelter) at one short side. The fields were of comparable sizes between treatments and years (honeybee-treated fields: 13.2 ± 4.6 ha (mean ± s.d. (σ)), min 6.0 ha, max 20.0 ha, control fields: 14.3 ± 4.9 ha, min 7.5 ha, max 23.0 ha). Farmers had sown the fields with either hybrid cultivars (22 fields) or with open-pollinated cultivars (22 fields; as described in [22], and the cultivar types were evenly distributed between landscape types (see below) and honeybee treatments (electronic supplementary material, table S1).

Fields in both control and honeybee treatments were replicated in homogeneous and heterogeneous landscape types. We selected fields in either homogeneous or heterogeneous landscape types, defined by the proportion semi-natural grasslands of total agricultural land within 1 km from the centre transect of each field. This radius corresponds to the foraging distance of important wild pollinators such as bumblebees [23]. We aimed to base our landscape classification also on the amount of field borders, indirectly measured as mean block area (area of fields with a common non-cropped field border), but only partially succeeded, with some overlap between the categories. Homogeneous landscapes had larger mean block area (general linear model, GLM; F_1,42 = 36.09, p < 0.01) and smaller proportion semi-natural grasslands (GLM; response variable arcsine square root transformed, F_1,42 = 172.98, p < 0.01) within a radius of 1 km from the centre transect, than heterogeneous landscapes (electronic supplementary material, table S2). The landscape measures were comparable between honeybee treatments (electronic supplementary material, table S2; GLMs; semi-natural grasslands, response variable arcsine square root transformed, F_1,42 = 0.81, p = 0.37, mean block area, F_1,42 = 1.39, p = 0.25) and cultivar types (data not shown). The area of winter oilseed rape at 1 km distance around each field, excluding the experimental field, was twice as big in homogeneous landscapes compared with heterogeneous landscapes (GLM; F_1,42 = 9.00, p < 0.01), but was unrelated to honeybee treatments (GLM; F_1,42 = 1.24, p = 0.27) and cultivar types (GLM; F_1,42 = 0.18, p = 0.67). The size of the experimental fields was not significantly related to the area of other winter oilseed rape within 1 km radius around the fields (Pearson's correlation, r₄₂ = 0.21, p = 0.18).

Land-use data were acquired from the Integrated Administration and Control System, administrated by the Swedish Board of Agriculture, with data from farmers' commitments yearly updated within the Common Agricultural Policy, and analysed in QGIS 2.14.0-Essen [24].

(b). Adding or removing honeybees

In the 23 honeybee-treated fields, we added two honeybee hives per hectare to one of the short sides of the field, which is a commonly recommended honeybee hive density in oilseed rape [25]. The hives were present from the onset until the end of flowering. Four fields with added honeybee hives already had 4, 6, 7 and 20 permanent honeybee hives, respectively, that we included in the treatment. In total, each honeybee treated field had 12–43 hives during the experiment. We measured the strength of each honeybee hive by counting the number of honeybees leaving the hive during 2 min at mid-flowering. We found no differences in strength between hives in relation to the landscape types or cultivar type sown (linear mixed model; t₁₈ = 0.24, p = 0.81 and t₁₈ = 1.37, p = 0.18, respectively).

In Sweden, beekeepers are obliged to report the locality of their hives to the County Administration Board. Based on information from that register, information from local residents and beekeepers, and personal observations, we identified areas with low numbers or with no honeybee hives. Experimental fields lacked, to our knowledge, large apiaries within 3 km, except in one field where the distance between a control field and a honeybee-treated field was 2.6 km. One beekeeper removed a large apiary from a control field. A very limited number of single honeybee hives within 3 km of the control fields were allowed, but only if another flowering field (an oilseed rape field not included in this study or in one case a raspberry field) occurred between the honeybee hive and the control field. During the experiment, we found an initially undiscovered honeybee hive 600 m from one of the control fields. We consider the prevalence of feral honeybees negligible due to Varroa mite infestations (Ingemar Fries 2016, personal communication).

(c). Insect sampling

We measured the density of flying and flower-visiting insects by observation and netting along transects at 100, 200 and 300 m from the side of the field with the shortest width (the side with added honeybee hives in honeybee treated fields). Transects were never closer than 40 m from the field's long-side edge, and the 300 m transect was at least 140 m away from the opposite short side of the field. Landscape elements such as wells, marl pits, power lines and solitary trees were never closer than 25 m from a transect. Transects were 2 m wide and 100 m long (200 m²), and divided into four segments. Each segment was observed for 5 min at weekly intervals during flowering on four occasions in 2011 (9 May to 5 June) and on five occasions in 2012, except for one field which was visited four times only (2 May to 28 May). We initially limited surveys to days with more than 15°C, sunny weather and wind speeds less than 8 m s⁻¹, but due to cool weather in the beginning of the season and the fact that honeybees forage at lower temperatures (12–14°C) in early spring than later in the season [26], we changed the temperature criterion to more than 12°C.

Individuals of bee (Hymenoptera: Apoidea), hoverfly (Diptera: Syrphidae) and marchfly (Diptera: Bibionidae) families were identified to species. Other flies (Diptera: Brachycera) were determined to family level, and the remaining insects to order. We attempted to catch all insects that could not be determined to species or family in the field, and identified these in the laboratory. However, 94% of the insects were impossible to collect, and therefore identified to family only. Very small insects such as pollen beetles were not sampled.

(d). Data analysis

We first confirmed the success of the honeybee manipulation by analysing the honeybee densities against the treatments. Thereafter, we analysed the total density of all wild insects in a common model, where we kept estimates for the six groups; bumblebees, solitary bees, hoverflies, marchflies, other flies, and other flying and flower-visiting insects separate. Finally, we analysed the densities of the groups of wild insects in separate models in relation to the treatments.

Analyses were generally performed with generalized linear mixed-effects models (GLMMs, ‘lme4’ [27]). However, the confirmation model of the honeybee density showed skewed residuals, and the model including densities of all wild insects needed compensation for heterogeneous variances among insect groups (function ‘varIDent’ in the ‘nlme’-package [28]). We therefore specified linear mixed models (LMMs, ‘nlme’ [28,29]) that could handle these issues for these two models.

For these two LMMs, we summed the densities of flying and flower-visiting individuals in the four segments of the three transects per field and calculated the mean density per visit to one estimate per 200 m² transect and field (with separate estimates for each insect group in the model of total density of wild insects). Densities were log-transformed (log(x + 1)) prior to analysis to achieve normal error distribution.

In the separate models of the densities of the six wild insect groups, we summed the densities of flying and flower-visiting insects in the four segments of the three transects per field for all visits, to one estimate per 200 m² transect, field and group. We assumed Poisson distribution of the errors for the models with bees and other insects, and negative binomial distribution for the models with flies. The log-transformed number of visits per field was included as an offset. In cases of over-dispersion, we added an observation level random effect [30].

Honeybee treatment (HB), landscape type, year and cultivar type were included as explanatory factors in all models. Field size of each field and distance from field edge were included as covariates in all models. Distance from field edge was standardized by subtracting the mean and divided by two standard deviations with the ‘rescale’ function in the package ‘arm’ [31]. Occasionally, other parameters were rescaled to make the models converge. We included also the two-way interactions between HB and landscape type, HB and field size, and HB and distance from field edge in all models. In the models with the honeybee density and the model with the total density of wild insects, we also included the three-way interaction among HB, field size and distance from field edge as well as the remaining underlying two-way interactions. Field identity was included as a random factor to account for lack of independence among observations made in transects of the same field.

In the model with the total density of wild insects, we also included insect group, and the interactions between insect group and HB, and insect group and landscape type, as fixed effects. In addition to field identity, we also included transect identity as a random factor, nested within field identity, to account for lack of independence of insect groups sampled in the same transect. This model showed somewhat skewed residuals and to achieve normally distributed residuals, we aggregated our data on a higher level by summing the densities per segments and transects and calculating a mean value per field and group and log-transformed the value (log(x + 1)). We specified a model as above but without distance from field edge and its interactions with treatments.

We analysed the data in R v. 3.0.3.6 [32], using mixed models (LMM and GLMM) from the packages nlme [29] and lme4 [27]. When we found unexplained outliers, we ran the models with and without the outlier to see if the results changed and, if so, present both results. Models were simplified by sequential backward elimination of non-significant effects (p < 0.05) based on z-tests or t-tests from the summary procedure or, in the latter case when a factor had more than two levels, F-test from the ANOVA procedure. Factors were kept in the model as long as they were included in an interaction. We used likelihood ratio tests (LR) to evaluate significance of the variables after simplification of the models fitted with maximum-likelihood. Tests of significant factors are from the minimum adequate model, whereas tests of non-significant factors are from the simplest adequate model to which the factor could be added (e.g. the minimum adequate model for a non-significant factor and the minimum adequate model with additional factor included in interaction when necessary for a non-significant interaction). Model estimates were extracted using the effects package [33].

3. Results

(a). Successful honeybee treatment

The honeybee manipulation successfully affected the difference in density of flying and flower-visiting honeybees both between control fields and fields with added honeybees, and within fields with added honeybee hives (figure 1; electronic supplementary material, table S3). Honeybee densities declined with increasing distance from the field edge in fields with added honeybee hives, but not in control fields (figure 1; electronic supplementary material, table S3). In control fields, the density of honeybees was very low. The model estimated a mean density of 6 honeybees per 200 m² and 20 min (95% confidence interval (CI) 3–12) in control fields, while fields with added honeybee hives had a mean model estimated density of 1044 honeybees per 200 m² and 20 min (95% CI 564–1934). Honeybee densities were higher in 2011 than in 2012 and equal in the two landscape types and the oilseed rape cultivar types (figure 1; electronic supplementary material, table S3).

Figure 1. — Model estimated density of honeybees per transect (200 m², 20 min) in fields with added honeybees (dashed line, open circles) and control fields (solid line, black filled circles) at increasing distance from the field edge (electronic supplementary material, table S3). The standardized distance from the field edge of −0.6 corresponds to 100 m, 0 corresponds to 200 m and 0.6 corresponds to 300 m distance from the field edge. The shaded area denotes 95% CIs. (Online version in colour.)

(b). Overall density of wild insects

We observed 8649 flower-visiting insects and 12 022 flying insects. Flies (9175 individuals) and honeybees (7583 individuals) dominated, with intermediate densities of bumblebees (1061 individuals) and marchflies (1260 individuals), while hoverflies (579 individuals), solitary bees (529 individuals) and other flying and flower-visiting insects (484 individuals) were the least abundant groups. A list of the identified species of bees, hoverflies and marchflies, and families of other flies is specified in the electronic supplementary material (table S4).

The densities of wild insects in fields with added honeybees increased with distance from the field edge in small fields only, while densities in middle-sized and large fields decreased with field size rather than with distance from field edge within 300 m. However, the three-way interaction among HB, field size and standardized distance from field edge was non-significant (LR = 2.40, d.f. = 1, p = 0.13). As the residuals were somewhat skewed, we reran the model without distance from field edge and interactions, using data aggregated on a higher level as described above. The results of the common factors and interactions of the two models were similar, except for the interaction between HB and insect group, which was significant in the first model (LR = 12.43, d.f. = 5, p = 0.03), but not in the model with data aggregated on a higher level (LR = 5.63, d.f. = 5, p = 0.34).

The density of wild insects was lower in fields with added honeybees, and the effect depended on the oilseed rape field size (LR = 7.22, d.f. = 1, p < 0.01). In fields with added honeybee hives, the density of wild insects decreased with increased field size (figure 2).

Figure 2. — Model estimated density of wild insects per field in relation to oilseed rape field size in honeybee treated fields (dashed line, open circles) and control fields (solid line, black filled circles). The error bars denotes 95% CIs. (Online version in colour.)

Homogeneous landscapes had fewer wild insects, and the effect depended on insect group (LR_. = 12.33, d.f. = 5, p = 0.03).

(c). Density per group of wild insects

When the six insect groups were analysed separately, their densities were in most cases negatively affected by honeybee addition (table 1). Wild bee (bumblebee and solitary bee) densities decreased with increased field size when honeybees were added, but slightly increased or were unaffected by field size in control fields (figure 3a,b). However, when removing three data points with very high values from a large control field, the analysis showed that honeybee-treated fields had lower densities of solitary bees than control fields, irrespective of field area. Hoverfly densities showed a similar, but non-significant, tendency. Bumblebee densities increased with the distance from honeybee hives, but were approximately similar at all measured distances from the field edge in control fields (figure 4a). Other flies' and other flying and flower-visiting insects' densities were lower in honeybee-treated fields, but the effect depended on the distance from the honeybee hives (figure 4b,c).

Table 1.

Effects on density per transect of six different groups of wild insects, of honeybee treatment (HB), landscape type, year, cultivar type, distance from field edge (DIST) and field size with interactions in 44 oilseed rape fields. Analyses were performed with likelihood ratio tests using generalized linear mixed-effects models. Italicized numbers show significant factors. When a factor or interaction was included in a higher-order interaction, no values are reported.

variable	bumblebees		solitary bees		hoverflies^a		marchflies		other flies^b		other insects
variable		p		p		p		p		p		p
HB
landscape type	20.13₁	<0.01	0.06₁	0.80	4.05₁	0.04	2.19₁	0.14	9.80₁	<0.01	3.06₁	0.08
year	23.85₁	<0.01	0.39₁	0.53	2.23₁	0.14	0.72₁	0.40	78.51₁	<0.01	2.79₁	0.09
cultivar type	4.81₁	0.03	0.05₁	0.83	0.19₁	0.66	1.00₁	0.32	6.24₁	0.01	0.19₁	0.67
DIST			7.10₁	<0.01	4.27₁	0.04
field size									7.20₁	0.01
DIST × field size	0.98₁	0.32	1.58₁	0.21	0.06₁	0.81			0.68₁	0.41	3.78₁	0.05
HB × DIST	6.91₁	<0.01	0.13₁	0.72	1.05₁	0.31			6.77₁	0.01	4.10₁	0.04
HB × landscape type	0.76₁	0.38	0.07₁	0.79	0.67₁	0.41	0.12₁	0.73	0.13₁	0.72	0.32₁	0.57
HB × field size	5.83₁	0.02	5.52₁	0.02	3.71₁	0.05			0.14₁	0.71	1.54₁	0.21
HB × field size × DIST	0.15₁	0.70	1.54₁	0.21	2.73₁	0.10	3.79₁	0.05	0.20₁	0.66	<0.01₁	0.99

Open in a new tab

^aOne marginally non-significant value (p = 0.054) were kept in the simplest model.

^bDistance from field edge was standardized in this model.

Figure 3. — Model estimated densities of (a) bumblebees and (b) solitary bees flies per transect and field (200 m², 20 min) in fields with added honeybees (dashed line, open circles) and in control fields (solid line, black filled circles) in relation to field size (table 1). The error bars denotes 95% CIs. (Online version in colour.)

Figure 4. — Model estimated densities of (a) bumblebees, (b) other flies and (c) other flying and flower-visiting wild insects per transect (200 m², 20 min) in fields with added honeybees (dashed line, open circles) and in control fields (solid line, black filled circles) in relation to the distance from field edge (table 1). The error bars denotes 95% CIs. Note the different scales. (Online version in colour.)

In small fields, there was a non-significant tendency that marchfly densities increased with distance from honeybee hives, but slightly increased with distance from field edge in control fields (HB × field size × DIST; table 1).

Bumblebees and hoverflies were fewer in homogeneous landscapes than in heterogeneous landscapes (table 1). Other flies were more abundant in homogeneous landscapes than in heterogeneous landscapes (table 1).

4. Discussion

Adding honeybee hives adjacent to flowering oilseed rape fields reduced the overall density of wild insects in the crop compared with control fields without honeybee hives. Previous studies have observed effects on wild bees from honeybee addition [15,18–20,34–36], but we found negative effects on all wild insect densities when analysed separately; bumblebees, solitary bees, marchflies, hoverflies, other flies, and other flying and flower-visiting insects. The demonstrated effects were independent of heterogeneity of the surrounding landscape.

Honeybee effects on wild insects can either be caused by population changes of the wild insects or by their displacement [15]. As honeybee hives were added at the onset of flowering (only 37 of the 624 honeybee hives in the experiment were permanently positioned at their locations) and because the study period during the flowering of the oilseed rape was rather short, population effects are unlikely to be the cause of the observed effects. The densities of bumblebees and flies other than hoverflies and marchflies decreased with increasing distance from the field edge in control fields, while the densities increased with distance from the honeybee hives, indicating that displacement was occurring when honeybees were present. A similar, but non-significant, tendency was found for the density of all wild insects in small fields. However, we could not unequivocally confirm this tendency because the model showed skewed residuals. Adding honeybees decreased the over-all densities of wild insects more in large fields than in small fields. This is an indication of displacement, where wild insects avoid areas with honeybees due to exploitative competition [19,35] or interference [37], possibly depending on the larger alternative foraging area within larger fields. Larger fields did not have higher densities of honeybees within 300 m from the honeybee hives, even if they had higher numbers of honeybee hives at the field edges than smaller fields. Honeybees are capable of foraging at long distances and have been suggested to be less sensitive to field size than other bee species are [16]. Still, displacement is the most probable explanation for the observed effects. The displacement effect on bumblebees and solitary bees were more pronounced in large fields than in small fields when analysed separately. These bees are central place foragers and bound to their nesting sites, and larger fields provide more alternative forage within flight distance from the nest compared with small fields. The other groups, which are non-central place foragers and not bound to a particular field by its nest, were not affected by field size.

We cannot conclude whether or not insect groups were affected differently by honeybee addition, because the results of the interaction between HB and insect group differed between models with different aggregation levels. However, all groups of wild insects were negatively affected by honeybee addition, when analysed separately (i.e. bumblebees, solitary bees, marchflies, hoverflies, other flies, and other flying and flower-visiting insects). The effect from honeybee addition depended on field size and/or distance from field edge for all groups, irrespective if the insects were central place foragers or not.

Oilseed rape being a major food resource for pollen- and nectar-feeding insects early in the season in altered agricultural landscapes [38,39], and the rest of the landscape playing a minor role at this time, might explain why the effect by honeybees was not, as expected, altered by landscape type.

Displacement of wild insects caused by honeybee addition, leading to reduced densities of wild flower-visiting insects in flowering crops, might have negative impacts on the pollination service they provide, particularly in crops dependent on native pollinators. We have previously reported moderate effects on yield with supplemented honeybee pollination from the same study system as reported here. Displacement of wild pollinators might have reduced the effect of the supplementation, but because the wild insects were few compared with the honeybees, we assume the effect to be minor in winter oilseed rape. Crop pollination by wild insects has been shown to increase seed set [2,40]. Hence, honeybee addition to insect pollinated crops may, at least partly, offset the benefit for pollination.

Negative effects of interspecific competition early in the season might have consequences for pollinator populations lasting throughout the season, because it is mainly queens that forage in oilseed rape. This might be an explanation for the findings of a recent study by Herbertsson et al. [34] using part of the study system presented here, where decreased densities of bumblebees were seen in field borders after the flowering period of the oilseed rape at sites with added honeybees. Further studies are required to determine whether honeybee addition leads to fitness consequences of wild insects caused by displacement. If so, attempting to mitigate a loss of pollination services due to decreased populations of wild pollinators by adding honeybees can potentially increase the negative pressures on wild insects. This is highly important in the light of the reduction in biodiversity in agricultural landscapes.

Supplementary Material

Additional information about study design, landscape characteristics, honeybee statistics and a species list

rspb20161641supp1.pdf^{(162.9KB, pdf)}

Acknowledgements

We thank farmers and beekeepers for letting us work with their fields and honeybee hives. We also thank Martin Stjernman and SAPES for providing landscape data.

Data accessibility

Datasets associated with this article are available at Dryad: http://dx.doi.org/10.5061/dryad.rt166 [41].

Authors' contributions

S.A.M.L., M.R., H.G.S. and R.B. designed the experiments. S.A.M.L. performed the experiments with help from L.H. in setting up the honeybee treatment in 2012. S.A.M.L. analysed the data with help from H.G.S. S.L. wrote the manuscript with comments from all other authors. All authors gave final approval for publication.

Competing interests

We have no competing interests.

Funding

Funding was provided by the Swedish Farmers' Foundation for Agricultural Research (V1133010), the Swedish Board of Agriculture (28-13610-10) to S.A.M.L. and R.B., and by the Swedish research council FORMAS to R.B. (220-2012-1044) and H.G.S. (210-2009-1680).

References

1.Klein A-M, Vaissière BE, Cane JH, Steffan-Dewenter I, Cunningham SA, Kremen C, Tscharntke T. 2007. Importance of pollinators in changing landscapes for world crops. Proc. R. Soc. B. 274, 303–313. ( 10.1098/rspb.2006.3721) [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Button L, Elle E. 2014. Wild bumble bees reduce pollination deficits in a crop mostly visited by managed honey bees. Agric. Ecosyst. Environ. 197, 255–263. ( 10.1016/j.agee.2014.08.004) [DOI] [Google Scholar]
3.Goulson D, Nicholls E, Botías C, Rotheray EL. 2015. Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science 347, 1255957 ( 10.1126/science.1255957) [DOI] [PubMed] [Google Scholar]
4.Garibaldi LA, Aizen MA, Klein AM, Cunningham SA, Harder LD. 2011. Global growth and stability of agricultural yield decrease with pollinator dependence. Proc. Natl Acad. Sci. USA 108, 5909–5914. ( 10.1073/pnas.1012431108) [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Rader R, et al. 2015. Non-bee insects are important contributors to global crop pollination. Proc. Natl Acad. Sci. USA 113, 201517092. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Marini L, Quaranta M, Fontana P, Biesmeijer JC, Bommarco R. 2012. Landscape context and elevation affect pollinator communities in intensive apple orchards. Basic Appl. Ecol. 13, 681–689. ( 10.1016/j.baae.2012.09.003) [DOI] [Google Scholar]
7.Breeze TD, Bailey AP, Balcombe KG, Potts SG. 2011. Pollination services in the UK: how important are honeybees? Agric. Ecosyst. Environ. 142, 137–143. ( 10.1016/j.agee.2011.03.020) [DOI] [Google Scholar]
8.Roffet-Salque M, et al. 2015. Widespread exploitation of the honeybee by early Neolithic farmers. Nature 527, 226–230. ( 10.1038/nature15757) [DOI] [PubMed] [Google Scholar]
9.Jaffé R, et al. 2010. Estimating the density of honeybee colonies across their natural range to fill the gap in pollinator decline censuses. Conserv. Biol. 24, 583–593. ( 10.1111/j.1523-1739.2009.01331.x) [DOI] [PubMed] [Google Scholar]
10.Senapathi D, et al. 2015. The impact of over 80 years of land cover changes on bee and wasp pollinator communities in England. Proc. R. Soc. B 282, 20150294 ( 10.1098/rspb.2015.0294) [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Cane JH, Tepedino VJ. In press. Gauging the effect of honey bee pollen collection on native bee communities. Conserv. Lett. ( 10.1111/conl.12263) [DOI] [Google Scholar]
12.Fontaine C, Collin CL, Dajoz I. 2008. Generalist foraging of pollinators: diet expansion at high density. J Ecol. 96, 1002–1010. ( 10.1111/j.1365-2745.2008.01405.x) [DOI] [Google Scholar]
13.Pianka ER. 1981. Competition and niche theory. In Theoretical ecology: principles and applications (ed. May RM.), 2nd edn, pp. 167–197. Blackwell. [Google Scholar]
14.Paini D, Roberts J. 2005. Commercial honey bees (Apis mellifera) reduce the fecundity of an Australian native bee (Hylaeus alcyoneus). Biol. Conserv. 123, 103–112. ( 10.1016/j.biocon.2004.11.001) [DOI] [Google Scholar]
15.Paini DR. 2004. Impact of the introduced honey bee (Apis mellifera) (Hymenoptera: Apidae) on native bees: a review. Austr. Ecol. 29, 399–407. ( 10.1111/j.1442-9993.2004.01376.x) [DOI] [Google Scholar]
16.Artz DR, Hsu CL, Nault BA. 2011. Influence of honey bee, Apis mellifera, hives and field size on foraging activity of native bee species in pumpkin fields. Environ. Entomol. 40, 1144–1158. ( 10.1603/EN10218) [DOI] [PubMed] [Google Scholar]
17.Hudewenz A, Klein A-M. 2015. Red mason bees cannot compete with honey bees for floral resources in a cage experiment. Ecol. Evol. 5, 5049–5056. ( 10.1002/ece3.1762) [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Dupont YL, Hansen DM, Valido A, Olesen JM. 2004. Impact of introduced honey bees on native pollination interactions of the endemic Echium wildpretii (Boraginaceae) on Tenerife, Canary Islands. Biol. Conserv. 118, 301–311. ( 10.1016/j.biocon.2003.09.010) [DOI] [Google Scholar]
19.Torné-Noguera A, Rodrigo A, Osorio S, Bosch J. 2015. Collateral effects of beekeeping: impacts on pollen-nectar resources and wild bee communities. Basic Appl. Ecol. 17, 199–209. ( 10.1016/j.baae.2015.11.004) [DOI] [Google Scholar]
20.Goulson D. 2003. Effects of introduced bees on native ecosystems. Annu. Rev. Ecol. Evol. Syst. 34, 1–26. ( 10.1146/annurev.ecolsys.34.011802.132355) [DOI] [Google Scholar]
21.Holzschuh A, Dormann CF, Tscharntke T, Steffan-Dewenter I. 2011. Expansion of mass-flowering crops leads to transient pollinator dilution and reduced wild plant pollination. Proc. R. Soc. B 278, 3444–3451. ( 10.1098/rspb.2011.0268) [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Lindström SAM, Herbertsson L, Rundlöf M, Smith HG, Bommarco R. 2016. Large-scale pollination experiment demonstrates the importance of insect pollination in winter oilseed rape. Oecologia 180, 759–769. ( 10.1007/s00442-015-3517-x) [DOI] [PubMed] [Google Scholar]
23.Walther-Hellwig K, Frankl R. 2000. Foraging habitats and foraging distances of bumblebees, Bombus spp. (Hym., Apidae), in an agricultural landscape. J. Appl. Entomol. 124, 299–306. ( 10.1046/j.1439-0418.2000.00484.x) [DOI] [Google Scholar]
24.Quantum GIS Development Team. 2016. QGIS geographic information system. See http://qgis.osgeo.org.
25.Breeze TD, et al. 2014. Agricultural policies exacerbate honeybee pollination service supply-demand mismatches across Europe. PLoS ONE 9, e82996 ( 10.1371/journal.pone.0082996) [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Winston ML. 1991. The biology of the honey bee. Cambridge, MA: Harvard University Press. [Google Scholar]
27.Bates D, Maechler M, Bolker B, Walker S.2014. lme4: Linear mixed-effects models using Eigen and S4. See http://CRAN.R-project.org/package=lme4 .
28.Pinheiro JC, Bates DM. 2000. Mixed-effects models in S and S-PLUS. Berlin, Germany: Springer. [Google Scholar]
29.Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team. 2016. nlme: linear and nonlinear mixed effects models. See http://cran.r-project.org/package=nlme. [Google Scholar]
30.Harrison XA. 2014. Using observation-level random effects to model overdispersion in count data in ecology and evolution. PeerJ 2, e616 ( 10.7717/peerj.616) [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Gelman A, Su YS.2013. arm: data analysis using regression and multilevel/hierarchical models. See http://cran.r-project.org/package=arm .
32.R Core Team. 2013. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]
33.Fox J. 2003. Effect displays in R for generalised linear models. J. Stat. Softw. 8, 1–27. [Google Scholar]
34.Herbertsson L, Lindström SAM, Rundlöf M, Bommarco R, Smith HG. 2016. Competition between managed honeybees and wild bumblebees depends on landscape context. Basic Appl. Ecol. 17, 609–616 ( 10.1016/j.baae.2016.05.001) [DOI] [Google Scholar]
35.Balfour NJ, Gandy S, Ratnieks FLW. 2015. Exploitative competition alters bee foraging and flower choice. Behav. Ecol. Sociobiol. 69, 1731–1738. ( 10.1007/s00265-015-1985-y) [DOI] [Google Scholar]
36.Thomson D. 2004. Competitive interactions between the invasive European honey bee and native bumble bees. Ecology 85, 458–470. ( 10.1890/02-0626) [DOI] [Google Scholar]
37.Reitz SR, Trumble JT. 2002. Competitive displacement among insects and arachnids. Annu. Rev. Entomol. 47, 435–465. ( 10.1146/annurev.ento.47.091201.145227) [DOI] [PubMed] [Google Scholar]
38.Westphal C, Steffan-Dewenter I, Tscharntke T. 2009. Mass flowering oilseed rape improves early colony growth but not sexual reproduction of bumblebees. J. Appl. Ecol. 46, 187–193. ( 10.1111/j.1365-2664.2008.01580.x) [DOI] [Google Scholar]
39.Stanley DA, Stout JC. 2014. Pollinator sharing between mass-flowering oilseed rape and co-flowering wild plants: implications for wild plant pollination. Plant Ecol. 215, 315–325. ( 10.1007/s11258-014-0301-7) [DOI] [Google Scholar]
40.Garibaldi LA, et al. 2013. Wild pollinators enhance fruit set of crops regardless of honey bee abundance. Science 339, 1608–1611. ( 10.1126/science.1230200) [DOI] [PubMed] [Google Scholar]
41.Lindström S, Herbertsson L, Rundlöf M, Bommarco R, Smith H. 2016. Data from: Experimental evidence that honeybees depress wild insect densities in a flowering crop. Dryad Digital Repository ( 10.5061/dryad.rt166) [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Additional information about study design, landscape characteristics, honeybee statistics and a species list

rspb20161641supp1.pdf^{(162.9KB, pdf)}

Data Availability Statement

Datasets associated with this article are available at Dryad: http://dx.doi.org/10.5061/dryad.rt166 [41].

[RSPB20161641C1] 1.Klein A-M, Vaissière BE, Cane JH, Steffan-Dewenter I, Cunningham SA, Kremen C, Tscharntke T. 2007. Importance of pollinators in changing landscapes for world crops. Proc. R. Soc. B. 274, 303–313. ( 10.1098/rspb.2006.3721) [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C2] 2.Button L, Elle E. 2014. Wild bumble bees reduce pollination deficits in a crop mostly visited by managed honey bees. Agric. Ecosyst. Environ. 197, 255–263. ( 10.1016/j.agee.2014.08.004) [DOI] [Google Scholar]

[RSPB20161641C3] 3.Goulson D, Nicholls E, Botías C, Rotheray EL. 2015. Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science 347, 1255957 ( 10.1126/science.1255957) [DOI] [PubMed] [Google Scholar]

[RSPB20161641C4] 4.Garibaldi LA, Aizen MA, Klein AM, Cunningham SA, Harder LD. 2011. Global growth and stability of agricultural yield decrease with pollinator dependence. Proc. Natl Acad. Sci. USA 108, 5909–5914. ( 10.1073/pnas.1012431108) [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C5] 5.Rader R, et al. 2015. Non-bee insects are important contributors to global crop pollination. Proc. Natl Acad. Sci. USA 113, 201517092. [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C6] 6.Marini L, Quaranta M, Fontana P, Biesmeijer JC, Bommarco R. 2012. Landscape context and elevation affect pollinator communities in intensive apple orchards. Basic Appl. Ecol. 13, 681–689. ( 10.1016/j.baae.2012.09.003) [DOI] [Google Scholar]

[RSPB20161641C7] 7.Breeze TD, Bailey AP, Balcombe KG, Potts SG. 2011. Pollination services in the UK: how important are honeybees? Agric. Ecosyst. Environ. 142, 137–143. ( 10.1016/j.agee.2011.03.020) [DOI] [Google Scholar]

[RSPB20161641C8] 8.Roffet-Salque M, et al. 2015. Widespread exploitation of the honeybee by early Neolithic farmers. Nature 527, 226–230. ( 10.1038/nature15757) [DOI] [PubMed] [Google Scholar]

[RSPB20161641C9] 9.Jaffé R, et al. 2010. Estimating the density of honeybee colonies across their natural range to fill the gap in pollinator decline censuses. Conserv. Biol. 24, 583–593. ( 10.1111/j.1523-1739.2009.01331.x) [DOI] [PubMed] [Google Scholar]

[RSPB20161641C10] 10.Senapathi D, et al. 2015. The impact of over 80 years of land cover changes on bee and wasp pollinator communities in England. Proc. R. Soc. B 282, 20150294 ( 10.1098/rspb.2015.0294) [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C11] 11.Cane JH, Tepedino VJ. In press. Gauging the effect of honey bee pollen collection on native bee communities. Conserv. Lett. ( 10.1111/conl.12263) [DOI] [Google Scholar]

[RSPB20161641C12] 12.Fontaine C, Collin CL, Dajoz I. 2008. Generalist foraging of pollinators: diet expansion at high density. J Ecol. 96, 1002–1010. ( 10.1111/j.1365-2745.2008.01405.x) [DOI] [Google Scholar]

[RSPB20161641C13] 13.Pianka ER. 1981. Competition and niche theory. In Theoretical ecology: principles and applications (ed. May RM.), 2nd edn, pp. 167–197. Blackwell. [Google Scholar]

[RSPB20161641C14] 14.Paini D, Roberts J. 2005. Commercial honey bees (Apis mellifera) reduce the fecundity of an Australian native bee (Hylaeus alcyoneus). Biol. Conserv. 123, 103–112. ( 10.1016/j.biocon.2004.11.001) [DOI] [Google Scholar]

[RSPB20161641C15] 15.Paini DR. 2004. Impact of the introduced honey bee (Apis mellifera) (Hymenoptera: Apidae) on native bees: a review. Austr. Ecol. 29, 399–407. ( 10.1111/j.1442-9993.2004.01376.x) [DOI] [Google Scholar]

[RSPB20161641C16] 16.Artz DR, Hsu CL, Nault BA. 2011. Influence of honey bee, Apis mellifera, hives and field size on foraging activity of native bee species in pumpkin fields. Environ. Entomol. 40, 1144–1158. ( 10.1603/EN10218) [DOI] [PubMed] [Google Scholar]

[RSPB20161641C17] 17.Hudewenz A, Klein A-M. 2015. Red mason bees cannot compete with honey bees for floral resources in a cage experiment. Ecol. Evol. 5, 5049–5056. ( 10.1002/ece3.1762) [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C18] 18.Dupont YL, Hansen DM, Valido A, Olesen JM. 2004. Impact of introduced honey bees on native pollination interactions of the endemic Echium wildpretii (Boraginaceae) on Tenerife, Canary Islands. Biol. Conserv. 118, 301–311. ( 10.1016/j.biocon.2003.09.010) [DOI] [Google Scholar]

[RSPB20161641C19] 19.Torné-Noguera A, Rodrigo A, Osorio S, Bosch J. 2015. Collateral effects of beekeeping: impacts on pollen-nectar resources and wild bee communities. Basic Appl. Ecol. 17, 199–209. ( 10.1016/j.baae.2015.11.004) [DOI] [Google Scholar]

[RSPB20161641C20] 20.Goulson D. 2003. Effects of introduced bees on native ecosystems. Annu. Rev. Ecol. Evol. Syst. 34, 1–26. ( 10.1146/annurev.ecolsys.34.011802.132355) [DOI] [Google Scholar]

[RSPB20161641C21] 21.Holzschuh A, Dormann CF, Tscharntke T, Steffan-Dewenter I. 2011. Expansion of mass-flowering crops leads to transient pollinator dilution and reduced wild plant pollination. Proc. R. Soc. B 278, 3444–3451. ( 10.1098/rspb.2011.0268) [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C22] 22.Lindström SAM, Herbertsson L, Rundlöf M, Smith HG, Bommarco R. 2016. Large-scale pollination experiment demonstrates the importance of insect pollination in winter oilseed rape. Oecologia 180, 759–769. ( 10.1007/s00442-015-3517-x) [DOI] [PubMed] [Google Scholar]

[RSPB20161641C23] 23.Walther-Hellwig K, Frankl R. 2000. Foraging habitats and foraging distances of bumblebees, Bombus spp. (Hym., Apidae), in an agricultural landscape. J. Appl. Entomol. 124, 299–306. ( 10.1046/j.1439-0418.2000.00484.x) [DOI] [Google Scholar]

[RSPB20161641C24] 24.Quantum GIS Development Team. 2016. QGIS geographic information system. See http://qgis.osgeo.org.

[RSPB20161641C25] 25.Breeze TD, et al. 2014. Agricultural policies exacerbate honeybee pollination service supply-demand mismatches across Europe. PLoS ONE 9, e82996 ( 10.1371/journal.pone.0082996) [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C26] 26.Winston ML. 1991. The biology of the honey bee. Cambridge, MA: Harvard University Press. [Google Scholar]

[RSPB20161641C27] 27.Bates D, Maechler M, Bolker B, Walker S.2014. lme4: Linear mixed-effects models using Eigen and S4. See http://CRAN.R-project.org/package=lme4 .

[RSPB20161641C28] 28.Pinheiro JC, Bates DM. 2000. Mixed-effects models in S and S-PLUS. Berlin, Germany: Springer. [Google Scholar]

[RSPB20161641C29] 29.Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team. 2016. nlme: linear and nonlinear mixed effects models. See http://cran.r-project.org/package=nlme. [Google Scholar]

[RSPB20161641C30] 30.Harrison XA. 2014. Using observation-level random effects to model overdispersion in count data in ecology and evolution. PeerJ 2, e616 ( 10.7717/peerj.616) [DOI] [PMC free article] [PubMed] [Google Scholar]

[RSPB20161641C31] 31.Gelman A, Su YS.2013. arm: data analysis using regression and multilevel/hierarchical models. See http://cran.r-project.org/package=arm .

[RSPB20161641C32] 32.R Core Team. 2013. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]

[RSPB20161641C33] 33.Fox J. 2003. Effect displays in R for generalised linear models. J. Stat. Softw. 8, 1–27. [Google Scholar]

[RSPB20161641C34] 34.Herbertsson L, Lindström SAM, Rundlöf M, Bommarco R, Smith HG. 2016. Competition between managed honeybees and wild bumblebees depends on landscape context. Basic Appl. Ecol. 17, 609–616 ( 10.1016/j.baae.2016.05.001) [DOI] [Google Scholar]

[RSPB20161641C35] 35.Balfour NJ, Gandy S, Ratnieks FLW. 2015. Exploitative competition alters bee foraging and flower choice. Behav. Ecol. Sociobiol. 69, 1731–1738. ( 10.1007/s00265-015-1985-y) [DOI] [Google Scholar]

[RSPB20161641C36] 36.Thomson D. 2004. Competitive interactions between the invasive European honey bee and native bumble bees. Ecology 85, 458–470. ( 10.1890/02-0626) [DOI] [Google Scholar]

[RSPB20161641C37] 37.Reitz SR, Trumble JT. 2002. Competitive displacement among insects and arachnids. Annu. Rev. Entomol. 47, 435–465. ( 10.1146/annurev.ento.47.091201.145227) [DOI] [PubMed] [Google Scholar]

[RSPB20161641C38] 38.Westphal C, Steffan-Dewenter I, Tscharntke T. 2009. Mass flowering oilseed rape improves early colony growth but not sexual reproduction of bumblebees. J. Appl. Ecol. 46, 187–193. ( 10.1111/j.1365-2664.2008.01580.x) [DOI] [Google Scholar]

[RSPB20161641C39] 39.Stanley DA, Stout JC. 2014. Pollinator sharing between mass-flowering oilseed rape and co-flowering wild plants: implications for wild plant pollination. Plant Ecol. 215, 315–325. ( 10.1007/s11258-014-0301-7) [DOI] [Google Scholar]

[RSPB20161641C40] 40.Garibaldi LA, et al. 2013. Wild pollinators enhance fruit set of crops regardless of honey bee abundance. Science 339, 1608–1611. ( 10.1126/science.1230200) [DOI] [PubMed] [Google Scholar]

[RSPB20161641C41] 41.Lindström S, Herbertsson L, Rundlöf M, Bommarco R, Smith H. 2016. Data from: Experimental evidence that honeybees depress wild insect densities in a flowering crop. Dryad Digital Repository ( 10.5061/dryad.rt166) [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Experimental evidence that honeybees depress wild insect densities in a flowering crop

Sandra A M Lindström

Lina Herbertsson

Maj Rundlöf

Riccardo Bommarco

Henrik G Smith

Abstract

1. Introduction