Test 3 (Crossref)


Test 3 (Crossref)

Analysis Methods Diffusion-Weighted Imaging


Obesity is associated with negative physical and mental health outcomes. Being overweight/obese is also associated with executive functioning impairments and structural changes in the brain. (CAP) analysis. Dynamic CAP metrics including dwell time (DT), frequency of occurrence, and transitions were computed. Multiple measurement models were compared based on model fit with indicators from the D-KEFS assigned a priori (shifting, inhibition, and fluency). Multiple structural equation models were computed with interactions between BMI and the dynamic CAP metrics predicting the three latent factors of shifting, inhibition, and fluency while controlling for age, sex, and head motion. Models were assessed for main effects of BMI and CAP metrics predicting the latent factors.

Honey bee

  • Gabriela M. Quinlan,

  • Meghan O. Milbrath,

  • Clint R. V. Otto,

  • Rufus Isaacs


Agriculturally important commercially managed pollinators including honey bees (Apis mellifera L., 1758) and bumble bees (Bombus impatiens Cresson, 1863) rely on the surrounding landscape to fulfill their dietary needs. A previous study in Europe demonstrated that managed honey bee foragers and unmanaged native bumble bee foragers are associated with different land uses. However, it is unclear how response to land use compares between managed honey bees and a managed native bumble bee species in the United States, where honey bees are an imported species. Furthermore, to our knowledge, no such direct comparisons of bee responses to land use have been made at the colony level. To better understand how two different social bees respond to variation in land use, we monitored the weights of A. mellifera and B. impatiens colonies placed in 12 apiaries across a range of land use in Michigan, United States in 2017. Bombus impatiens colonies gained more weight and produced more drones when surrounded by diverse agricultural land (i.e., non-corn/soybean cropland such as tree fruits and grapes), while honey bee colonies gained more weight when surrounded by more grassland/pasture land. These findings add to our understanding of how different bee species respond to agricultural landscapes, highlighting the need for further species-specific land use studies to inform tailored land management.


The European honey bee, Apis mellifera L., 1758, is the most economically important pollinator across the world [1], providing pollination to a wide range of food and forage crops. Recent developments in bee rearing have also made the common eastern bumble bee, Bombus impatiens Cresson, 1863, commercially available for crop pollination in eastern North America [2], and studies show B. impatiens to be more efficient at pollinating certain crops than honey bees [3, 4]. However, the health of honey bees [5] and population stability of many species of bumble bee [6, 7] are in jeopardy, due in part to insufficient access to suitable nutrition within their foraging range [8].

Bees rely on flowers to fulfill their dietary needs, and honey bees and bumble bees, as social generalist foragers, likely have broadly similar macro- and micro-nutrient dietary requirements [9, 10]. But, whether honey bees and bumble bees benefit from similar landscape composition is an area of debate. A study in the Northern Great Plains found that landscapes that support productive honey bee colonies also support more abundant and diverse native bee communities [11]. However, a study in Europe found that honey bees and unmanaged bumble bees were associated with different floral resources: honey bees were observed foraging more often on mass-blooming crops and bumble bees showed intermediate preference for both mass-blooming crops and semi-natural habitats [12]. Honey bees and bumble bees exhibit different foraging strategies [1315] as well as key differences in their nutrient intake [16, 17] which could lead to different responses to the same landscapes. Understanding how different pollinators respond to landscape composition is important to designing pollinator conservation strategies, as currently in the United States, pollinator conservation is generally not designed to target specific guilds [18, 19]. Furthermore, comparing the landscape response of a managed species of bumble bee, such as B. impatiens, to the response from managed honey bee, A. mellifera, could provide greater insights into how landscapes influence the health of key managed bee species.

Monitoring differences in foraging activity alone may not be a sufficient metric by which to compare species’ response to landscape composition. All previous studies, of which we are aware, use monitoring techniques such as sweep netting and bowl traps to compare the effects of land use and landscape composition on different groups of bees. However, these techniques only capture forager visitation, potentially masking downstream effects on colony-level productivity and fitness. One such metric of colony-level productivity is colony weight. Change in colony weight, for both honey bees and bumble bees, is closely tied to the amount of stored resources (pollen and nectar) as well as adult and developing bees, and therefore a good metric of comparison. Bumble bee colony fitness may also be assessed by the number of reproductives produced, which include gynes (virgin queens) and male drones. Bumble bee colonies have an annual cycle in which reproductives are produced seasonally [20]. Honey bee colonies, conversely, are perennial and overwinter with a single queen. A honey bee queen can live several years, and therefore there is not an equivalent measure of reproductive output for honey bee colonies [21].

To compare the productivity of two key managed, social, generalist species (A. mellifera and B. impatiens) in the same landscapes, in this study, we 1) compare colony weight changes of honey bee and bumble bees located at the same apiary locations, 2) model the association between colony weight change throughout a growing season to the area of multiple land use categories (corn/soybeans, non-corn/soy crops, forage land, forests, and developed land), and 3) investigate the relationship between bumble bee colony gyne and drone production and area of land use categories. Because these two species are each generalists, we expect both species’ colonies will show similar patterns in weight change among apiaries and we also expect that weight gain and the production of bumble bee reproductives (gynes/ drones) will be positively correlated with the area of surrounding forage land.

Materials and methods

Site selection and land use quantification

In the summer of 2017, we assessed weights of honey bee and bumble bee colonies at 12 apiary sites across southern Michigan. Apiary locations were selected from our collaborating beekeeper’s existing apiary locations to be spatially independent within a 2 km radius and to capture a range of land use available in the region (Fig 1A). Because these apiaries were on private land and the land owner provided us permission to access the colonies, field work permits were not required. Land use was determined using the 2017 30 m2 resolution Cropland Data Layer (CDL) [22]. In R Studio version 3.6.3 [23], the raster [24] and rgeos [25] packages were used to calculate the area of each land cover within 2 km of each focal apiary. These CDL land use classifications were then binned into six broad land use categories of interest: corn/soybeans, non-corn/soy crops (sugar beets, dry beans, potatoes, watermelons, cucumbers, peas, cherries, peaches, apples, grapes, asparagus, peppers, squash, blueberries, cabbage, celery), forage (sunflower, alfalfa, clover/wildflowers, other hay–non-alfalfa, fallow/idle cropland, grassland/pasture, woody wetlands, herbaceous wetlands, shrubland), forests (deciduous forest, mixed forest, evergreen forest), and developed land (developed open space, developed low intensity, developed medium intensity, developed high intensity) (S1 Table). An “other” (not applicable) category included grain crops (rye, oats, barley, spring wheat, etc.), other non-flowering crops (sod grass seed, Christmas trees), and undefined categories (other crop, barren).

Figure 1: Location of 12 apiaries in Michigan, United States (A) and land use within 2 km (B) where honey bee and bumble bee colonies were sampled. Apiary locations are indicated with points, labeled with apiary number, and surrounded by a 2 km buffer. Land use is binned into six broad categories–forage, non-corn/soy crops, forests, developed, corn/soybeans, and all other land uses.

Honey bee colonies

Each apiary contained on average 39 ± 2 (mean ± S. E.) commercial, migratory honey bee colonies. In each apiary, two honey bee colonies were placed on hive scales (SolutionBee, Raleigh, NC) that log colony weights every 15 minutes. Colonies were first inspected to ensure they were strong and contained a laying queen. Rather than use raw weights, which may be affected by the beekeeper adding or removing equipment, we instead calculated the cumulative change in weight over time for each colony. Sudden changes in colony weight (>3 kg/15 min) were removed, resulting in a smooth, continuous weight change curve [27]. We also filtered out data points in which the raw colony weight was less than 10 kg (approximate weight of an empty colony), which would suggest that the scale was malfunctioning or that the hive’s weight was not correctly distributed on the scale. After this data processing, honey bee colony weights (i.e., cumulative weight change) were collected on a similar schedule as the bumble bee colonies: June 30, July 24, and August 10. To obtain the most accurate weight, scales were read at midnight when all foragers should be inside the hive and when maintenance by the beekeeper is unlikely to occur.

Statistical analysis

All statistical analysis was carried out in R Studio version 3.6.3 [23]. We assessed changes in weight gain over two time periods to accommodate potential differences in durations of colony growth between species [20, 21]: changes in colony weight were calculated for each colony between the first round (June 28/30) and third/final round (August 10), as well as between the first round (June 28/30) and second round (July 24). Within each apiary we averaged the change in weight of the two honey bee colonies and the change in weight and number of gynes and drones produced by the three bumble bee colonies. We also averaged honey bee and bumble bee weights within apiary and monthly weigh-in round. Due to hive scale malfunctioning, weight change values could not be calculated for four apiaries over the shorter time period (n = 8 remaining) and five apiaries over the longer time period (n = 7 remaining). Pearson’s product moment correlation was used to determine the correlation between the two species’ average colony weights and weight changes. Additive generalized linear models (GLM) were used to regress the change in weights and number of gynes and drones produced in each yard with the area of surrounding non-corn/soy crops, forage land, forests, and developed land, each scaled and centered. The area of corn/soybeans was strongly correlated (r = 0.68) with both forests and developed land and caused multicollinearity issues when included in the GLM. We chose to exclude the corn/soybean variable from our GLM because corn and soybeans are also not traditionally considered bee-supportive forage [28]. Once corn/ soybeans were excluded, none of the models had multicollinearity issues, based upon variance inflation factors (VIF < 3) [29]. The r2glmm package was used to calculate partial R2 values [30]. Because honey bees and bumble bees respond to landscapes at different spatial scales due to differences in their foraging ranges [31], we assessed each species’s response to land use over 1 km, 3 km, 4 km, and 6 km, in addition to our original range of 2 km (S2 Table). The area of each land use category was correlated (r > 0.40) across spatial scales, suggesting that landscape composition around our apiaries was relatively conserved across these distances (S3 Table).


Forests (primarily deciduous forests) were the dominant land use surrounding our apiaries, ranging from 15% - 52% of surrounding land area and averaging 30% of land within the 2 km surrounding area (Fig 1B). The area of non-corn/soy cropland (primarily tree fruits and grapes) ranged from 0% - 18% of the surrounding area and made up on average 4% of the surrounding area. Forage land ranged from 9% - 39% and averaged at 22% across sites and was made up, in large part, by woody wetlands and grassland/pasture. Our study area also included a range of corn/soy cropland (2% - 47%; 22% (min-max; mean)) and developed land (4% - 40%; 12%) (mostly open space and low intensity developed land) (Fig 2, S1 Table).