1
Scientific RepoRts | 6:32023 | DOI: 10.1038/srep32023
www.nature.com/scientificreports
Migratory management and
environmental conditions aect
lifespan and oxidative stress in
honey bees
Michael Simone-Finstrom
1,*
, Hongmei Li-Byarlay
1,2,3,*
, Ming H. Huang
1
, Micheline K. Strand
4
,
Olav Rueppell
3
& David R. Tarpy
1,2
Most pollination in large-scale agriculture is dependent on managed colonies of a single species, the
honey bee Apis mellifera. More than 1 million hives are transported to California each year just to
pollinate the almonds, and bees are trucked across the country for various cropping systems. Concerns
have been raised about whether such “migratory management” causes bees undue stress; however to
date there have been no longer-term studies rigorously addressing whether migratory management
is detrimental to bee health. To address this issue, we conducted eld experiments comparing bees
from commercial and experimental migratory beekeeping operations to those from stationary colonies
to quantify eects on lifespan, colony health and productivity, and levels of oxidative damage for
individual bees. We detected a signicant decrease in lifespan of migratory adult bees relative to
stationary bees. We also found that migration aected oxidative stress levels in honey bees, but that
food scarcity had an even larger impact; some detrimental eects of migration may be alleviated by a
greater abundance of forage. In addition, rearing conditions aect levels of oxidative damage incurred
as adults. This is the rst comprehensive study on impacts of migratory management on the health and
oxidative stress of honey bees.
Honey bees (Apis mellifera) are the most economically important pollinators in North America and are crucial for
sustaining production in many agroecosystems
1
. Honey bee colonies are composed of tens of thousands of indi-
viduals, which allows them to pollinate crops eectively over a large geographic area, particularly with the assis-
tance of beekeepers who transport colonies for pollination services. e major economic driver of the beekeeping
industry in the U.S. is fullling pollination contracts for various growers, including almonds, berries, apples, and
cucurbits
2
. erefore, commercial beekeepers transport bee colonies on trucks both regionally and nationally
for many months of the year
3–5
. Given this paradigm, many colonies are repeatedly moved over several months
to a series of large monocultures, which potentially increases a colony’s exposure to pesticides
3,6
and pathogens
7
,
limits access to diversied pollen sources
8
, and forces the foraging bees to re-learn and re-assess their environ-
mental surroundings. e assumption, therefore, is that factors associated with migratory beekeeping operations
overwhelm bees and induce a stress response, ultimately contributing to increased colony losses and susceptibility
to disease, parasites, and syndrome-like eects such as Colony Collapse Disorder (CCD)
9
.
However, the impact of transporting colonies for pollination services has currently received minimal inves-
tigation
7,10
. Furthermore, no study has empirically examined the cellular consequences of stress to honey bee
workers under this paradigm. On a physiological level, the immediate eect of transportation (i.e., 24-hours
aer a 3-day trip) has been shown to lead to a reduction in the size of the glands that are essential for brood food
production in nurse bees
10
. However, the long-term implications of this nding are unclear. us, the goal of
this study was to better resolve the question regarding whether migratory beekeeping practices cause changes in
measurable levels of oxidative stress and possible impacts of the cells throughout a season.
1
Department of Entomology, North Carolina State University, Raleigh, NC, USA.
2
The W.M. Keck Center for
Behavioral Biology, North Carolina State University, Raleigh, NC, USA.
3
Department of Biology, University of North
Carolina at Greensboro, Greensboro, NC, USA.
4
Life Sciences Division, U.S. Army Research Oce, Research Triangle
Park, NC, USA.
*
These authors contributed equally to this work. Correspondence and requests for materials should
be addressed to M.S.-F. (email: Michael.SimoneFinstrom@ars.usda.gov)
Received: 14 January 2016
Accepted: 28 July 2016
Published: 24 August 2016
OPEN
www.nature.com/scientificreports/
2
Scientific RepoRts | 6:32023 | DOI: 10.1038/srep32023
Our ability to understand and quantify stress is critical for evaluating the impacts of abiotic and biotic factors
inuencing honey bee health and colony productivity. Oxidative stress is important in eukaryotic organisms and
can have severe negative eects. Reactive oxygen species (ROS) are the causative agents of oxidative stress, and
they are produced as a by-product of normal metabolic processes (or otherwise suer from diminished redox
homeostasis). Cells that lose their ability to remove excess ROS undergo oxidative stress, which leads to DNA
mutation
11
, irreparable damage of proteins
12
, and membrane instability
13
. Oxidative stress can lead to apoptosis
and cellular damage, which are intimately linked to aging
14,15
. Acute exposure to mild stress can extend lifespan
because stress-resistance mechanisms, like the production of antioxidants, can be activated
16
. However, severe
or chronic stressors, like prolonged sublethal pesticide exposure
17
, usually shorten lifespan
18
. In particular, some
theories argue that aging is simply a result of the accumulation of oxidative damage
19,20
. Furthermore, ROS may
be induced by exogenous sources (i.e. pesticides and environment). We hypothesized that migratory honey bees
experience oxidative stress and may have shorter life spans.
e biomarker malondialdehyde (MDA) is a common measure of oxidative stress in honey bees, other insects,
and vertebrate systems
21–24
. MDA is the main organic compound produced from lipid peroxidation of polyun-
saturated fatty acids in cellular membranes
25
. MDA levels reect the combined eects of exposure to oxidative
stress and the ability or lack thereof to resist oxidative damage through various repair mechanisms
26
.
We determined how the movement of managed honey bee colonies across dierent agricultural landscapes
inuenced colony health and productivity, adult lifespan, and levels of oxidative stress, measured as MDA. Our
study is the rst to examine the long-term eects of migratory colony management on stress accumulation in
honey bees. is study was conducted in three parts where we: 1) determined the eect of migratory manage-
ment on honey bee lifespan; 2) investigated the eects of migratory management on colony health, productivity,
lifespan, and oxidative stress on either stationary or migratory bees; and 3) investigated the eects of intensive,
short-term migratory movement on levels of oxidative stress in honey bees.
Methods
A scheme of the general experimental designs for Experiments 1, 2, and 3 is shown in Fig.1.
Experiment 1: Eects of commercial migratory operation on bee lifespan. To remove the con-
founding eects of the hive environment and energetic costs related to foraging behavior, the lifespan of worker
bees from stationary and migratory colonies was determined under controlled conditions following standard
Figure 1. e experimental schemes of Experiment 1, 2, and 3. e rst panel depicts where the bees were
transported for the dierent experiments, the second shows the procedures used (lifespan analysis in incubator
cages or oxidative stress analyses), and the third depicts the paired design aspect where colonies were matched,
and newly emerged bees were paint-marked and swapped across these pairs. is gure was created in
Microso Powerpoint. Pictures were taken by and all items designed or modied by M. Simone-Finstrom. Maps
were modied from gures available through Wikimedia Commons. e original US map is available at
https://en.wikipedia.org/wiki/File:USA_Counties.svg, and the original NC map can be found at https://commons.
wikimedia.org/wiki/File:Bluenc.png; both are in the public domain.
www.nature.com/scientificreports/
3
Scientific RepoRts | 6:32023 | DOI: 10.1038/srep32023
protocols
27
. Frames of emerging workers were collected from 8 colonies from a large apiary in Henderson, NC
shortly aer completing pollination services in California (~4,500 km distance). An additional 8 colonies were
chosen of comparable population maintained at the Lake Wheeler Farms Bee Research Facility at North Carolina
State University, Raleigh, NC to represent stationary hives. Following established methods
27
, ~45 newly emerged
workers from each of the 16 colonies were color-marked using Testors enamel paint and divided evenly among
12 plastic cups, so that a total of ~60 bees were maintained in each cup. Mortality of color-marked bees was
recorded daily. Since small cage populations tend to decline more rapidly, populations were replenished with
newly emerged worker bees from a non-experimental source to maintain ≥ 30 bees per cage at all times. Bees
were fed 50% sucrose solution ad libitum.
Lifespan analyses were conducted twice during the experimental period (Fig.1). Aer the bees had been
collected for the rst trial in May 2012 (early season), the colonies were transported to Maine for subsequent
pollination of lowbush blueberry. Colonies returned to North Carolina and frames containing emerging bees
were collected again for a subsequent trial in late June 2012 (late season). Two colonies could not be used for the
second trial, as one became queenless and one died, so two other colonies were used. Dierences in adult worker
lifespan were compared using the parametric survival analysis JMP Pro 10 with colony treatment (i.e., stationary
versus migratory) and season (early versus late) as factors in a general linear model.
Experiment 2: Eects of experimental migratory management on colony demography, bee
lifespan, and oxidative stress. Colony treatments. Colonies were established from queenless divisions
of overwintered hives and initiated in a single bee yard at the Lake Wheeler Farms Bee Research Facility at North
Carolina State University, Raleigh, NC in April 2012. Each split was made to contain three brood frames (two
sealed, one open), one honey frame, one pollen and ve foundation frames (wire with wax). Sister queens were
reared from a single Italian queen, placed as virgins within these colonies and allowed to mate naturally. In this
way, all colonies had similar genetic backgrounds.
Each colony was maintained in a single Langstroth hive body containing nine frames of comb for the dura-
tion of the study. Each colony was given 1–2 extra boxes (“honey supers”) above a queen excluder when needed
to ensure it maintained only one box with brood but had enough space to grow to deter swarming. Otherwise,
standard beekeeping practices were followed as in the rst experiment.
On May 1, 2012 (early season) all colonies were matched for size based on the numbers of frames of bees
and brood before the start of the experiment and divided into two treatments. Stationary colonies (N = 9) were
maintained in a single apiary bordering a forested park (Yates Mill) and agricultural elds containing mainly corn
and wheat. Migratory colonies (N = 10) were initially moved to the Central Crops Research Station in Clayton,
NC. Every 21 days (equivalent to one honey bee brood developmental cycle), the migratory colonies were moved
among the North Carolina Department of Agriculture Research Station in Goldsboro, NC or Rocky Mount, NC,
so that the colonies were moved 35–60 miles on each of ve trips (see Supplemental Methods, Tables S1 and S2,
for more information on the locations and primary crops). is continued from May 4, 2012 until the nal move
back to Clayton, NC on July 27, 2012, where the hives remained until the conclusion of the study.
To assess colony growth and general health, queen status, amount of stored pollen, some adult worker bees,
and amount of capped brood were measured in May, June, July, and August 2012 following standard protocols
28
.
Infestation of the ectoparasite Varroa destructor on adult bees was determined at the end of the study period by
sampling 300 bees per colony into 95% ethanol to dislodge the mites from the bees, following standard methods
29
.
Five colonies (two stationary and three migratory) died during the experiment and were excluded from all colony
status analyses, resulting in 7 colonies per treatment. A repeated-measures ANOVA was used to examine dier-
ences in colony status between stationary and migratory colonies during the study period.
Lifespan analysis. To remove the confounding eects of the hive environment and energetic costs related to
foraging behavior, the lifespan of workers from the stationary or migratory colonies was determined under con-
trolled conditions as described for Experiment 1. Lifespan analysis was conducted once in July towards the end
of the experimental period and aer the colonies had been transported 4 times. Dierences in lifespan between
workers from stationary and migratory colonies were analyzed as described for Experiment 1.
Analysis of oxidative stress (lipid peroxidation via MDA). Newly emerged worker bees were collected from each
colony and paint-marked in May 2012 (before the 2
nd
move, early season) and in July 2012 (before the 5
th
move,
late season). A paired-colony cross-fostering design, where a portion of these newly emerged, painted adult bees
was swapped between a paired colony from the opposite treatment, enabled a separate assessment of the eects
of rearing (larval) environment and adult environment (Fig.1). Paint-marked bees were sampled at 14 and 28
days old (‘age’) either from a brood frame as hive bees or outside the colony entrance as active foragers (‘collection
type’). Bees were stored at − 80 °C until analyses for oxidative stress.
ROS-mediated oxidative damage was quantified by measuring MDA level in individual worker heads.
The assay was conducted using the OXItek
™
Thiobarbituric Acid Reactive Substances (TBARS) Assay Kit
(ZeptoMetrix Corp). e head of each bee was ash-frozen in liquid nitrogen and immediately pulverized in a
1.5 mL micro-centrifuge tube with a sterile plastic pestle. e tissue was then mixed with 280 µ L of PBS, vortexed
to homogenize the cellular suspension, and centrifuged briey to precipitate pieces of tissue and cuticle. Aliquots
of the supernatant were used subsequently in the quantication assays. Both the TBARS and the Pierce
™
BCA
™
Protein Assay kits (ermo Scientic) were used according to the manufacturers’ recommendations. Total sol-
uble protein was determined by BCA Protein Assay and used to normalize the corresponding TBARS amounts.
Oxidative damage as measured by normalized MDA levels was examined in three biological replicates (colony)
in either early season (May) or late season (July). Each condition (rearing and adult environment) included ≥ 6
www.nature.com/scientificreports/
4
Scientific RepoRts | 6:32023 | DOI: 10.1038/srep32023
individual bees, and a total of 282 bees were tested. A greater number of hive bees (N = 210) were assayed than
foragers (N = 72) because of the low availability of foragers at the end of the experiments.
Data were analyzed using a three-way ANOVA to examine if dierences in oxidative stress (level of MDA)
were due to eects of rearing environment, adult environment, and season (May versus July). Colony eect was
considered as a random factor. Tukey-Kramer post-hoc tests were used to make pair-wise comparisons of dier-
ent experimental groups. Dierences were considered signicant at α = 0.05.
Experiment 3: Eect of intensive migratory management on oxidative stress. Colony treatment.
An independent set of 3 colony pairs (6 hives total) was selected from the Lake Wheeler Farms Bee Research
Facility at North Carolina State University, Raleigh, NC in August, 2012. Each colony was maintained in a single
brood box (as described for Experiment 2) and trucked 3 hours daily for 6 consecutive days to novel locations
within North Carolina. e colonies were moved each night, then opened upon arrival so that they could forage
in the new location during the day before they were moved again. e trips were, on average, 218 miles with a
range of 205–232 miles (see Supplemental Methods, Table S3, for more information on the locations).
Seven days prior to the rst move, newly emerged bees were paint-marked and returned to their hive in addi-
tion to a paired stationary hive from Experiment 2. e day aer the nal move, 14d-old bees were collected from
each hive and stored at − 80 °C for subsequent measures of oxidative stress as described above.
Due to logistics of driving these colonies around daily, we were only able to t six colonies on our truck. So we
included the maximum number of colonies that we possibly could given the available equipment. For each colony,
we have sucient samples (individuals per colony) for our analyses.
Measure of oxidative stress (lipid peroxidation via MDA). e procedure and statistical analyses were the same
as described for Experiment 2. ree colony pairs were sampled, and at least six bees from each treatment group
were analyzed (rearing environment, adult environment, age, and collection type), resulting in a total of 185 bees.
Results
Experiment 1: Eects of commercial migratory operation. Overall, lifespan was greater for bees
reared in stationary colonies as compared to migratory colonies and for the trial conducted in June versus
May, with the overall dierence between treatments being approximately 1 day. Colony treatment (X
2
= 11.39,
p = 0.0007) and trial (X
2
= 21.11, p < 0.0001) signicantly impacted worker lifespan, but there was no interaction
between the two eects (p > 0.5). In May, aer the colonies returned from California, the lifespan (mean a num-
ber of days ± s.e.) of individuals from stationary colonies was 19.45 ± 0.32 (N = 327) and 18.01 ± 0.32 (N = 338)
for bees from migratory colonies (Fig.2). Aer the bees returned from conducting pollination services in Maine
in June, the mean was 20.49 ± 0.35 days (N = 382) for individuals from stationary colonies and 19.89 ± 0.35 days
(N = 378) from migratory colonies (Fig.2).
Experiment 2: Effects of experimental migratory management. Colony health and status.
Demographic data was normally distributed based on the goodness of t test (Shapiro-Wilk W test; JMP Pro
10). For the number of adult bees and amount of sealed brood, there was an eect of time alone (p = 0.008 and
p = 0.003, respectively). However, the amount of stored pollen was dierent with respect to the amount of pollen
collected across the two colony treatments over the study period, with migratory colonies having more stored
pollen than stationary colonies at the nal time point (F
2,11
= 0.83, p = 0.03, Fig.3c). ere were no statistical dif-
ferences in the numbers of adult bees (F1,11 = 0.08, p = 0.6; Fig.3a), and no interaction between colony type and
Figure 2. In Experiment 1, bees reared in a commercial, migratory beekeeping operation (dashed line) had
reduced lifespan compared to bees reared in a stationary operation (solid lines; X
2
= 11.39, p = 0.0007) aer
returning to North Carolina following almond pollination (black) and again aer blueberry pollination
(gray).
www.nature.com/scientificreports/
5
Scientific RepoRts | 6:32023 | DOI: 10.1038/srep32023
time of measurement. ere were no statistical dierences of sealed brood either (F
1,11
= 0.28, p = 0.26, Fig.3b),
and no interaction with treatment over time.
ere was no dierence in number of mites at the end of the experiment, though there was a non-signicant
trend for percentage of mite infestation (p = 0.17) to be higher in migratory (17.8%) versus stationary (10.3%)
colonies, with the variance being slightly higher in migratory colonies (Levene test for unequal variance: p = 0.06,
Figure S1).
Lifespan. e mean lifespan (number of days ± s.e.) of workers from stationary colonies (22.19 ± 0.32; N = 291)
was about 1 day greater than that of workers from migratory colonies (21.34 ± 0.32; N = 207; X
2
= 6.48, p = 0.011,
Fig.4). is nding was consistent with the results of Experiment 1.
Oxidative stress (lipid peroxidation). Log transformation was applied to the data in order to have a normal dis-
tribution and positive values. Although some signicant group dierences were observed (Fig.5), no signicant
overall eects were detected for age (F
1,272
= 1.30, p = 0.25), adult environment (F
1,272
= 1.08, p = 0.30), rearing
environment (F
1,272
= 0.18, p = 0.67), season (F
1,272
= 0.42, p = 0.52), or collection type (F
1,272
= 0.27, p = 0.60).
However, significant interactions were detected for “adult environment x season” (F
1,272
= 8.19, p < 0.005),
Figure 3. Colony health and productivity. Box plots showing the minimum, rst quartile, median, third
quartile, and a maximum of the following types of data from Experiment 2: (a) Number of adult bees per
colony; (b) Amount of sealed pupal cells, and (c) Number of cells containing stored pollen. N = 7 colonies per
treatment. For each measure there was a signicant eect of time, but not treatment, except for pollen (c) where
migratory colonies had more pollen signicantly in August, as indicated by the star (treatment*time interaction:
F
2,11
= 0.83, p = 0.03).