Pathogen prevalence and abundance in honey bee colonies involved in almond pollination

27 Pages·2015·1.62 MB·English

Apidologie Original article
*TheAuthor(s),2015
ThisarticleispublishedwithopenaccessatSpringerlink.com
DOI:10.1007/s13592-015-0395-5
Pathogen prevalence and abundance in honey bee colonies
involved in almond pollination
IanCAVIGLI1,KatieF.DAUGHENBAUGH1,MadisonMARTIN1,MichaelLERCH4,
KatieBANNER4,EmmaGARCIA1,LauraM.BRUTSCHER1,2,3,MichelleL.FLENNIKEN1,2
1DepartmentofPlantSciencesandPlantPathology,MontanaStateUniversity,Bozeman,MT59717,USA
2InstituteonEcosystems,MontanaStateUniversity,Bozeman,MT59717,USA
3DepartmentofMicrobiologyandImmunology,MontanaStateUniversity,Bozeman,MT59717,USA
4DepartmentofMathematicalSciences,MontanaStateUniversity,Bozeman,MT59717,USA
Received25June2015–Revised17August2015–Accepted22September2015
Abstract– Honeybeesareimportantpollinatorsofagriculturalcrops.Since2006,USbeekeepershaveexperi-
encedhighannualhoneybeecolonylosses,whichmaybeattributedtomultipleabioticandbioticfactors,including
pathogens.However,therelativeimportanceofthesefactorshasnotbeenfullyelucidated.Toidentifythemost
prevalent pathogens and investigate the relationship between colony strength and health, we assessed pathogen
occurrence,prevalence,andabundanceinWesternUShoneybeecoloniesinvolvedinalmondpollination.Themost
prevalent pathogens were Black queen cell virus (BQCV), Lake Sinai virus 2 (LSV2), Sacbrood virus (SBV),
Nosema ceranae, and trypanosomatids. Our results indicated that pathogen prevalence and abundance were
associatedwithbothsamplingdateandbeekeepingoperation,thatprevalencewashighestinhoneybeesamples
obtainedimmediatelyafteralmondpollination,andthatweakcolonieshadagreatermeanpathogenprevalencethan
strongcolonies.
honeybeecolonyhealth/honeybeeviruses/Blackqueencellvirus/LakeSinaivirus/almondpollination
1. INTRODUCTION produced(CAB2014).Honeybeesareessentialto
almondproduction,andover60%oftheUScom-
Insectspollinateagriculturalcropsandplantspe- merciallymanagedhoneybeecoloniesareinvolved
ciesthatenhancelandscapebiodiversity.Thevalue in almond pollination from late February to early
of insect pollinationworldwideis $153 billionan- March. In 2013–2014, approximately 1.6 million
nually, and honey bees (Apis mellifera) are the bee colonies participated in this single pollination
primarypollinatorsofUScropsvaluedatover$17 event.
billionperyear(Calderone2012;Gallaietal.2009). Honeybee-mediatedpollinationisimportantfor
A strikingexampleof theroleof honeybee polli- agriculturalproduction.Thus,highannualUShon-
nators in agriculture is California almond produc- eybeecolonylosses(averaging33%since2006)
tion, where ~80 % of the world’s almonds are concerngrowers,beekeepers,scientists,andpolicy
makers(Spleenetal.2013;Steinhaueretal.2014;
van Engelsdorp et al. 2007, 2008, 2009, 2012).
ElectronicsupplementarymaterialTheonlineversionof
Although the total number of US commercially
thisarticle(doi:10.1007/s13592-015-0395-5)contains
supplementarymaterial,whichisavailabletoauthorized managedcolonieshasremainedataround2.5mil-
users. lion since the late 1990s, beekeepers must split
colonies(i.e.,maketwocoloniesfromonecolony)
Correspondingauthor:M.Flenniken,
[email protected] more frequently in order to make up for higher
Manuscripteditor:DavidTarpy annual losses (USDA NASS). Typically, the
I.Caviglietal.
majorityof colonylosses occur during the winter Longitudinal monitoring of colony health and
monthsandthus coincide with the almond polli- pathogen prevalenceand abundanceis critical to
nation season, though a recent US management determining the role of pathogens in colony
survey indicated that 2014 summer losses were losses, yet there have been very few studies of
also significant (~27.4 %) (Steinhauer et al. thisnatureperformedintheUSA.
2015). Research to date suggests that multiple To better understand the role of pathogens on
biotic and abiotic factors contribute to colony honeybeecolonyhealthandmortality,wemoni-
healthandsurvival(e.g.,viruses,mites,microbes, toredcolonyhealth(usingcolonysizeasaproxy),
bee genetics, weather, forage quality and avail- pathogen prevalence, and pathogen abundance,
ability, management practices, and agrochemical before, during, and after the 2013–2014 almond
exposure)(Calderone2012;Cornmanetal.2012; pollination season (Delaplane and van der Steen
Ellis et al. 2010; Gallai et al. 2009; Nazzi and 2013; OSU2011).We utilizedpolymerase chain
Pennacchio2014;PettisandDelaplane2010;van reaction(PCR)totestfor16commonhoneybee
der Zee et al. 2012, 2014; van Engelsdorp et al. pathogens.Themajorityofthesepathogenswere
2009,2012).Althoughnosinglefactorisrespon- viruses including Lake Sinai virus 1 (LSV1),
sibleforcolonylossesorcolonycollapsedisorder LSV2, LSV3, LSV4, LSV5, Black queen cell
(CCD), honey bee samples from CCD-affected virus (BQCV), Deformed wing virus (DWV),
colonies had greater pathogen (e.g., viruses and Sacbroodvirus (SBV),Acutebee paralysisvirus
Nosema) prevalence andabundancecomparedto (ABPV), Chronic bee paralysis virus (CBPV),
unaffected colonies (Cornman et al. 2012; Cox- Israeli acute paralysis virus (IAPV), and
Fosteretal.2007;Johnsonetal.2009;Steinhauer Kashmirbeevirus(KBV).Inaddition,wetested
et al. 2015; van Engelsdorp et al. 2009). In forbacterialpathogens(i.e.,Paenibacilluslarvae
addition, several epidemiologic and temporal andMelissococcusplutonius)andeukaryoticpar-
monitoring studies have partially attributed colo- asites including the microsporidia Nosema spp.
ny losses to pathogens (i.e., viruses, bacteria, and trypanosomatids (i.e., Crithidia mellificae
fungi, trypanosomatids, and mites) (Chen et al. SF/Lotmaria passim). Herein, we present data
2014; Cornman et al. 2012; Cox-Foster et al. from anobservationalcohortstudyatthe colony
2007; Evans and Schwarz 2011; Genersch and level that provides current information regarding
Aubert2010;Lockeetal.2014;McMenaminand pathogen prevalence and abundance in Western
Genersch 2015; Nazzi and Pennacchio 2014; US commercially managed honey bee colonies
Nielsen et al. 2008; Ravoet et al. 2013; involved in almond pollination. Enhanced un-
Tentcheva et al. 2004; van Engelsdorp et al. derstanding of the dynamics of pathogenic in-
2009,2012). While these studies are informative, fections in bee colonies across different geog-
therolesofpathogensincolonymortalityandthe raphies and in independent beekeeping opera-
relationships between colony strength and patho- tions will inform ongoing and future epidemi-
gen prevalence and abundance have not been ologic studies aimed at investigating the poten-
fully elucidated. tial relationships between colony health and
Themajorityofhoneybee-infectingpathogens pathogen occurrence, prevalence, and abun-
are RNA viruses, including Acute bee paralysis dance on a larger scale (van Engelsdorp et al.
virus,Blackqueencellvirus,Israeliacuteparaly- 2013). Understanding the role of pathogens
sis virus, Kashmir bee virus, Deformed wing vi- and other factors on honey bee health is criti-
rus, Sacbrood virus, Chronic bee paralysis virus cal to the development of pollinator manage-
(reviewed in (Brutscher et al. 2015; Chen and ment and conservation strategies that limit an-
Siede 2007; de Miranda et al. 2013; Evans and nual bee colony mortality (Gallant et al. 2014).
Schwarz 2011; Genersch and Aubert 2010;
McMenamin and Genersch 2015), and the Lake 2. MATERIALSANDMETHODS
Sinai viruses (Cornman et al. 2012;
Daughenbaughetal.2015;Granbergetal.2013; Additionaldetailsregardingmethodsareavailablein
Ravoet et al. 2013, 2015; Runckel et al. 2011). theSupplementalInformation(OnlineResource1).
PathogenprevalenceandabundanceinUShoneybeecolonies
2.1. Longitudinalmonitoringandsampling eachbeekeeperidentified15–20coloniesofdifferential
of commercially managed honey bee health.Specifically,operation3initiatedthestudywith
colonies five weak, five average, and five strong colonies and
providedsamplesatthreetimepoints;operation2ini-
Three Montana-based (Broadwater, Yellowstone, tiated the study with five weak, 13 average, and two
and Treasurecounties)commercialbeekeepingopera- strongcoloniesandprovidedsamplesatfourtimepoints;
tions that transport their honey bee colonies ~1,200 andoperation1initiatedthestudywithfiveweak,four
miles to California (Merced and Stanislaus counties) average,andtenstrongcoloniesandprovidedsamplesat
eachwinterforthealmondbloomprovidedhoneybee fourtimepoints(SupplementalTableS2).Atotalof176
samples before (October–December 2013), during honeybeesampleswithcorrespondingcolonystrength
(February 2014), and after (March/April and observationswereobtainedandanalyzed,fourobserva-
June 2014) almond pollination (Figure 1 and tionsofcolonystrengthlackedcorrespondingsamples,
Supplemental Table S2). Colony health, using colony andeightoftheoriginalcoloniesdiedduringthecourse
populationsizeasaproxy,wasassessedbythenumber ofthisstudy(SupplementalTableS2).
of frames covered with honey bees (frame counts) at
each sampling event (Delaplane and van der Steen
2013; OSU 2011). Colony strength was defined as 2.2. Honeybeesamples
follows:weakcolonies(<5framescoveredwithbees),
averagecolonies(6–8framescoveredwithbees),and Five female bees from each sample were used
strong colonies (>9 frames covered with bees). Live for RNA extraction, cDNA synthesis, pathogen-
honey bee samples (~100 per sample) were obtained specific PCR, and qPCR. The objective for path-
fromthetopoftheframesinthemiddleofthecolony. ogen screening was to identify the most prevalent
Sampleswerecomposedoffemalebeesofmixedage, pathogens associated with honey bees sampled
includingnurse,worker,andforagerbees.Thesamples from individual colonies at each sampling event.
were collected on ice or dry ice, stored at −20 °C, Basedonempiricaldata,literaturevalues,andpractical
shippedondryice,andtransferredto−80°Cpriorto sample handling considerations, we assayed five bees
analysis.AttheonsetofthestudyinNovember2013, percolonypersamplingevent.Thefollowingequation
Oct March-
June
molecular pathogen diagnostics
(PCR, qPCR)
Feb
Figure1.Longitudinalmonitoringofcommerciallymanagedhoneybeecolonies,before,during,andafterthe2014
almondpollinationseason.Honeybeecolonies(n=54)fromthreeMontana-basedcommercialbeekeepingopera-
tionsweremonitoredbefore(October–November2013),during(February2014),andafter(March–April);after2
(June)almondpollinationinCalifornia.Colonystrengthwasmeasuredateachsamplingevent.PCRwasutilizedto
detectpathogensassociatedwitheachsample,andqPCRwasutilizedtodeterminetheabundanceofpathogens
associatedwithasubsetofsamples(SupplementalTableS2).
I.Caviglietal.
fromPirketal.(2013),N=ln(1−D)/ln(1−P)(N =sam- which were all RNA viruses, in select samples to
ple size, ln = natural logarithm, D = probability of investigate the relationship between virus abun-
detection,P =proportionofinfectedbees),predictsthat dance and honey bee colony health. Five hundred
with a sample size offivebees, pathogenicinfections nanograms of RNA from each of these samples
affecting 45 % or more of the individuals within a was reverse transcribed with M-MLV as described
colonywouldbedetectedwith95%probability(Pirk above. All qPCR reactions were performed in
etal.2013);thissamplesizehasbeenprovensufficient triplicate with a CFX Connect Real Time instru-
forthepathogen-specificPCRdetectionofhighlyprev- ment (BioRad); reaction conditions and equations
alent pathogens (Daughenbaugh et al. 2015; Runckel for determining the relative abundance based on
etal.2011). standard curves are provided in supplemental
methods (Online Resource 1).
2.3. RNAisolation
Beesampleswerehomogenizedinwaterusingbeads 2.6.1. StatisticalanalysisofPCR
(3mm)andaTissueLyzer(Qiagen)at30Hzfor2min.
Samples were centrifuged for 12 min at 12,000×g For this study, we use Bpathogen prevalence^
at4°Ctopelletdebris,andRNAfromsupernatants to refer to the total number of pathogens detected
was extracted using TRIzol reagent (Life by PCR out of a target list of 16. Though our
Technologies) according to the manufacturer’s in- interest was in the relationship between strength
structions (Runckel et al. 2011). rating and pathogen prevalence, graphical analyses
indicated that there were likely relationships be-
2.4. Reversetranscription/cDNAsynthesis tween pathogen prevalence and sampling time as
well as between strength and sampling time. Thus,
cDNAsynthesisreactionswereperformedbyincu- we used a Poisson log-linear regression model and
bating 1,000–2,000 ng total RNA, Moloney murine accounted for an interaction between sample date
leukemia virus (M-MLV) reverse transcriptase (time period), beekeeping operation, colony
(Promega), and 500 ng random hexamer primers strength, and pathogen prevalence. Observations
(IDT)for1hat37°C,accordingtothemanufacturer’s with average strength rating were not included in
instructions(Runckeletal.2011). some analyses to simplify the inferences between
strong (S) and weak (W). The natural logarithm
2.5. Polymerasechainreaction(PCR) (ln) of the pathogen prevalence data was used in
comparisons between each beekeeping operation
PCRwasperformedaccordingtostandardmethods and time period combination (Pirk et al. 2013).
using the primers listed in Supplemental Table S1 For the model, we used beekeeping operation 1,
(Runckel et al. 2011). In brief, 1 μL cDNA template before almond pollination (time period 1), and
was combined with 10 pmol of each forward and weak colonies as the base level.
reverse primer and amplified with ChoiceTaq poly- Inall,ourmodelcanbeexpressed
merase (Denville) according to the manufacturer’s in-
structions using the following cycling conditions:
y∼PoissonðμÞ
95 °C for 5 min; 35 cycles of 95 °C for 30 s, 57 °C i i
logðμÞ¼β þβ (cid:2)operation2þ
for 30 s, and 72 °C for 30 s, followed by final i 0 1 i
β (cid:2)operation3 þβ (cid:2)ðS :period1Þþ
elongation at 72 °C for 4 min. The PCR products 2 i 3 i
β (cid:2)ðW:period2Þ þβ (cid:2)ðS:period2Þþ
were visualized by gel electrophoresis/fluorescence 4 i 5 i
β (cid:2)ðW:period3Þ þβ (cid:2)ðS:period3Þþ
imaging. 6 i 7 i
β (cid:2)ðW:period4Þ þβ (cid:2)ðS:period4Þ
8 i 9 i
2.6. QuantitativePCR(qPCR)
Quantitative PCR was used to analyze the rel- & wherey =thetotalabundance/prevalenceforthe
i
ative abundance of the most prevalent pathogens, ithobservationi=1,2,…,180.
PathogenprevalenceandabundanceinUShoneybeecolonies
& Operation2=1ifobservationi camefrombeekeep- effect for colony, but found the variance between
i
ingoperation2and0otherwise. colonies to be minimal compared to the overall
& Operation3=1ifobservationi camefrombeekeep- variance.
i
ingoperation3and0otherwise. logðyÞ ¼ β þβ (cid:2)operation2þ
& Period2=1ifobservationi wastakenduringand0 i 0 1 i
i β (cid:2)operation3 þβ (cid:2)period2þ
otherwise. 2 i 3 i
& Period3=1ifobservationi wastakenafterpolli- β4(cid:2)period3iþβ5(cid:2)period4iþ
i β (cid:2)S þβ (cid:2)A þγ þϵ
nationand0otherwise. 6 i 7 i jðiÞ i
& Period4=1ifobservationi wastakeninthesec-
i here γ is the random effect for colony. We
ond after pollination sampling time and 0 j(i)
assume γ ∼N(0,σ2 ), ϵ ∼N(0,σ2 ),
otherwise. j(i) colony i y
and γ and ϵ are independent for all
& Ai=1 if observation i was average (colony j=1,2,…j(i,6)0, i=1,2i,3,…,180. Variables are de-
strength)and0otherwise.
fined as they were in Section2.6.1 .
& S =1ifobservationi wasstrong(colonystrength)
i Coefficients of interest for this study are
and0otherwise. β , β , andβ . Estimated means and stan-
0 6 7
dard errors for these coefficients of interest
were obtained using the regression output.
2.6.2. StatisticalanalysisofqPCR
Models were fit using the software R (Core
Team 2014), the package lme4 was used for
The relationship between colony strength rating
the mixed model described in Section 2.6.1
and pathogen abundance was evaluated using a
(Bates et al. 2015).
log-normal regression to model the total abun-
dance with the predictor of interest, strength rat-
ing, while also accounting for the different bee- 3. RESULTS
keeping operations and the sampling time period.
To evaluate if the relative abundance of the most 3.1. Honeybeecolonymonitoring
prevalent pathogens, which were all (+) sense andpathogendiagnostics
RNA viruses (i.e., BQCV, SBV, LSV1, and
LSV2), we utilized virus copy number as an indi- Commercially managed colonies from three
cation of infection severity, though the relation- Montana-basedbeekeepingoperationsweremon-
ship between virus copy number and virus asso- itored before, during, andafter the almondpolli-
ciated disease or affects on the honey bee host are nation season (i.e., October 2013 to June 2014).
largely unknown (de Miranda et al. 2013). Colonypopulationsizewasutilizedasaproxyfor
Pathogen abundance was defined as the summed colony health and monitored at each sampling
abundance of the RNA virus genome copy num- event.Attheonsetofthestudy,eachbeekeeping
bers, which were measured by qPCR. Samples operationselectedweak,average,andstronghon-
that did not test positive for the virus by PCR ey bee colonies that were monitored throughout
were not assessed by qPCR and given a value of thestudy.ThesecolonieswerelocatedinMontana
zero. In total, there were 53 observations of total before and after almond pollination and in
abundance after inputting zeros for negative PCR California during the almond pollination season
tests. A log-linear regression was used to model (February 2014) (Figure 1). Honey bee colony
the total abundance with the predictor of interest, strength, pathogen prevalence, and abundance
strength rating, while also accounting for the dif- weremonitoredthroughoutthestudy.
ferent beekeeping operations and the sampling
time period; 1 was added to each observation 3.1.1. Pathogendetection
since some observations had 0 total abundance.
Some bee colonies were measured multiple times To identify the pathogens associated with the
in the 53 observations. We accounted for these honey bee samples collected over the course of
repeated measures on colonies with a random thisstudy,weutilizedPCRtotestforasuiteof16
I.Caviglietal.
commonpathogens.Themajorityofthesepatho- occurrence (frequency of detection) varied by
gens were viruses including Lake Sinai virus 1 pathogen (Figure 2). The most readily detected
(LSV1),LSV2,LSV3,LSV4,LSV5,Blackqueen pathogens in this sample cohort were the Lake
cell virus (BQCV), DWV, SBV, ABPV, CBPV, Sinai viruses (LSV1, LSV2, LSV3, and LSV4),
IAPV,andKBV.Inaddition,wetestedforbacte- which accounted for 36 % of the total positive
rial pathogens (i.e., P. larvae and M. plutonius) testsinweakcolonies,34%inaveragecolonies,
and eukaryotic parasites including the and 33 % in strong colonies (Figure 2). Overall,
microsporidiaNosema spp.andtrypanosomatids the most frequently detected pathogens were
(i.e., C. mellificae SF / L. passim). In order to BQCV, LSV2, SBV, Nosema ceranae,
identify the most prevalent pathogens associated C. mellificae / L. passim, and LSV1 (Figure 2).
witheachsample(i.e.,affecting45%ormoreof Graphical analyses of pathogen prevalence
the individuals within a colony), five bees per vs. colony strength rating and time revealed
sample were utilized for pathogen analyses that each beekeeping operation had different
(Daughenbaugh et al. 2015; Pirk et al. 2013). pathogen prevalence levels and that the rela-
Pathogen-specificPCRwasperformedtoidentify tionship between colony strength and pathogen
the pathogens associated with each sample prevalence varied at different sampling times
(Supplemental Tables S1 and S2 and (Figure 3). For example, the mean pathogen
Supplemental Figures S1 and S2). In this work, prevalence was in general greater in samples
we use Bpathogen occurrence^ to indicate the obtained immediately after almond pollination
frequency of detecting a specific pathogen as a (after) for all beekeeping operations, but the
percentage of the total number of positive tests, differences were more striking for beekeeping
Bpathogenprevalence^toindicatethenumberof operations 1 and 3 (Figure 3). In general,
different pathogens detected in a sample, and pathogen prevalence was largest in the samples
Btotal pathogen abundance^ to describe the total obtained after almond pollination and lowest in
number of pathogen genome copies in select samples obtained during almond pollination for
samples. all operations (Figure 3, Supplemental
To identify the most common pathogens de- Table S2, and Supplemental Figures S1 and
tectedinthisstudy,wecalculatedtheoccurrence S2).
of each pathogen as a percentage of the total To examine the relationship between colony
numberofpositivetestsforeachcolonystrength strength and pathogen prevalence, a Poisson
category (Figure 2). Throughout this study, 176 log-linear regression was used to compare the
samples with corresponding colony strength mean number of pathogens present in strong
observations were obtained (Supplemental vs. weak colonies (Table I). This model
Table S2). Colony strength observations were accounted for potential differences in each bee-
unevenly distributed between weak, average, keeping operation, as well as the potential
and strong colony ratings (i.e., weak colony interaction between time period and colony
strength was observed at 41 sampling events, strength. The analyses presented herein are fo-
average colony strength was observed at 54 cused on comparisons between weak and
sampling events, and strong colony strength strong colonies. The natural logarithm (ln) of
was observed at 81 sampling events). In addi- the pathogen prevalence data was used in com-
tion, the total number of pathogen-specific PCR parison between the mean pathogen prevalence
tests varied with each category. Specifically, in weak and strong colonies for each beekeep-
weak colonies had 122 positive tests, average ing operation at each sampling event (Pirk
colonies had 178 positive tests, and strong colo- et al. 2013). A statistical model was used to
nies had 292 positive tests. Therefore, on aver- calculate the estimated differences between the
ageacrossallsamplingdates,weakcolonieshad pathogen prevalence in weak and strong honey
2.98pathogenspersample,averagestrengthcol- bee colonies and associated confidence inter-
onieshad3.30pathogenspersample,andstrong vals, based on the standard errors of the dif-
colonies had 3.60 pathogens per sample. The ferences, given the covariates (Table I,
PathogenprevalenceandabundanceinUShoneybeecolonies
KBV C.m. / L.p. N12o%s. B19Q%C V L3CS%V.m4 6. %/ L .p. 1N2o%s. B2Q4C%V P. l1a%rvae C.m.9 /% L.p. N1o3s%. B2Q4C%V
1% 11% DWV LSV3 LSV4
8% 4% 2% DWV
L3S%V4 D5W%V LSV3 4%
L3S%V3 LSV2 S13B%V L2S0V%2 S1B9%V 5% LSV2 S1B6%V
22% L8S%V1 LSV1 19% LS7%V1
7%
weak average strong
Figure2.Distributionofhoneybeepathogensdetectedinweak,average,andstrongcolonies.Honeybeesamples
wereobtainedfrommonitorcoloniesfromOctober2013toJune2014.PCRwasusedtotestfor16honeybee-
infectingpathogensincludingviruses(ABPV,BQCV,CBPV,DWV,IAPV,KBV,SBV,LSV1,LSV2,LSV3,LSV4,
andLSV5,microsporidia(N.ceranae),bacteria(P.larvae andM.plutonius),andtrypanosomatids(C.m./L.p.).
Thepathogenoccurrenceinweak(<5frames;n=41),average(sixtoeightframes),orstrong(>9frames;n=81)
honey bee colonies is shown as a percentage of the total number of pathogens detected by PCR. The percent
pathogenoccurrenceforeachcolonystrengthratingisasfollows:weak(BQCV19%,DWV8%,SBV13%,LSV1
8%,LSV222%,LSV33%,LSV43%,KBV1%,C.m./L.p. 11%,N.ceranae 12%),average(BQCV24%,
DWV5%,SBV19%,LSV17%,LSV220%,LSV34%,LSV43%,C.m./L.p. 6%,N.ceranae 12%),andstrong
(BQCV24%,DWV4%,SBV16%,LSV17%,LSV219%,LSV35%,LSV42%,P.larvae 1%,C.m./L.p. 9%,
N.ceranae 13%).
Section 2). For the majority of the sampling a useful overall representation of the data,
dates, the relationship between pathogen prev- but since the results for each pathogen seem
alence and colony strength indicated that weak to depend solely upon the sampling date, we
colonies had slightly greater pathogen preva- could not make statistical claims regarding
lence than strong colonies (i.e., multiplicative the apparent association of any pathogen
differences >1), whereas this trend was re- with colony strength.
versed in the last sampling period, (i.e., after To determine which pathogens were most
two had a multiplicative difference <1) common at each time point, we calculated the
(Table I). However, these observed trends percent occurrence of each pathogen (Figure 4).
could not be differentiated from the expected The total number of pathogen-specific PCR
variation. The largest, though non-significant, positive tests varied for each time period.
difference between the mean number of patho- Specifically, there were 155 positive tests in
gens in weak colonies, as compared to strong bee samples collected before almond pollina-
colonies, after accounting for beekeeper, oc- tion, 79 positive tests during almond pollina-
curred immediately after almond pollination tion, 223 positive tests immediately after al-
(Table I). Samples obtained from weak colo- mond pollination, and 130 positive tests from
nies directly after almond pollination (March– 20 colonies that were sampled in June. This
April) had an estimated 31 % larger mean representation of the data indicates that BQCV,
pathogen number than strong colonies, with LSV2, and SBV were more common than other
an associated 95 % confidence interval of 11 honey bee pathogens including DWV,
% less to 93 % more after accounting for N. ceranae, LSV3, C. mellificae/L. passim,
beekeeping operation (Table I). In addition P. larvae, and KBV. The relative percentages
to mean pathogen prevalence, we investigat- of each pathogen vary with the time of sam-
ed if particular pathogens were present at a pling (e.g., LSV2 reached peak detection levels
higher proportion in colonies of different during almond pollination); however, pathogen
strengths (i.e., weak, strong, and average) occurrence also varied for each beekeeping op-
(Supplemental Figure S2). This summary is eration (Supplemental Figure S1).
I.Caviglietal.
9 weak
Operation 3
8 average
strong C3 C8
7
C2 C4 C6C7 C10 C12 C18
6
C7 C9
5 C6 C8C10 C12 C5 C9 C11 C16 C17
C2 C4
4 C1 C3 C5 C11 C13 C7
3 C14 C5 C6 C9 C12 C1
C15 C10
2
C2 C4
C1 C3 C8 C11
1
0
9
Operation 2
8
s
n B5
e 7
g
ho 6 B19 B6 B6
at B15 B3
p B4 B9 B14 B16 B21 B1 B4 B14
of 5 B20 B8B10B12 B24
er 4 B11 B18 B21 B3B5 B8B10 B17 B7 B9B22 B15B17B25
mb B1B3B5B6B7BB89B1B01B21B31B41B516 B3 B6B8 B20 B21 B1B4 B7 B13 B13 B23 B21
nu 23 B17 B1B4B5 B9B10B11B14B15B16 B19 B12 B16
B7 B12 B17
1
B19
B13 B11 B18 B20
0
9
Operation 1
8
7
A14
A12 A15 A19
6
A11A13 A16A18 A25 A9
5
A7A9 A17 A20 A7 A11 A19A20A23
4
A13 A10 A15
3
A9A11
A8A10A12 A18 A19 A23 A7A9A11 A16 A20 A10A12 A17
2
A15
A7 A14 A17 A24 A25 A8A10A12 A17
1
A21
A13 A16 A20 A22 A14A15A18A19 A23 A25
0
before during after after 2
Oct.-Dec. February March-April June
sample time period
PathogenprevalenceandabundanceinUShoneybeecolonies
ƒ
Figure3.Colonyhealthandpathogenprevalencebefore, 3.1.3. Colonymortality
during,andafteralmondpollinationforeachbeekeeping
operation.Honeybeesampleswereobtainedfrommonitor
The overall colony mortality in this study
coloniesfromOctober2013toJune2014.PCRwasusedto
was 14.8 %. Colony loss varied with the time
testfor16honeybee-infectingpathogens.Pathogenprev-
alence(y-axis)referstothenumberofdifferentpathogens of sampling and beekeeping operation
detectedinasample.Colonystrengthwasmonitoredasa (Table III). Colony mortality was greatest dur-
proxyforcolonyhealth;weak(<5frames),averagecolo- ing almond pollination, during which 13 % of
nies(6–8frames),orstrong(>9frames).Eachcolonywas
the colonies died. Colony mortality varied by
assignedauniqueidentifier(e.g.,“C1”forcolonynumber1
beekeeping operation; specifically, operation 2
inbeekeepingoperation3);iconsizeandshapewereused
tographicallyillustratecolonystrengthoverthecourseof experienced only 5.0 % loss, operation 1 had
thestudy(x-axis). 15.8 % loss, and operation 3 experienced 20 %
loss during almond pollination. Colony losses
in this sample cohort were less than average
3.1.2. Pathogenabundance US colony losses, which have been ~33 %
since 2006 (Steinhauer et al. 2015).
To examine if total pathogen abundance
was associated with colony strength, we per-
formed qPCR for the most commonly detect- 4. DISCUSSION
ed pathogens in our sample cohort, which
were all RNA viruses (i.e., BQCV, LSV2, Honey bees play animportant roleas pollina-
LSV1, and SBV) (Supplemental Table S2). torsofnumerousagriculturalcrops.Themajority
In total, there were 53 observations of total of commercially managedhoney bee colonies in
abundance with values that ranged from 0 to theUSaretransportedthroughouttheyeartomeet
1.19×1011; total pathogen abundance values the pollination demands of crops including
were log natural (ln) transformed after adding California almond production. Commercial bee-
1 (Figure 5). We used log-linear regression to keepersintheUSandsomepartsofEuropehave
model the total abundance with the predictor experienced high annual average losses, though
of interest, strength rating, while also ac- the factors most responsible for these losses are
counting for the different beekeeping opera- notwellunderstood.Theobjectivesofthisstudy
tions and the sampling time period and de- weretoidentifythe pathogens currentlycirculat-
termined that the variance among colonies ing in Western US-based commercial honey bee
was minimal compared to the overall vari- colonies involved in almond pollination and to
ance (estimates and their associated confi- examinetherelationshipbetweencolonystrength,
dence intervals provided (Table II)). pathogen prevalence, and abundance. Pathogen
Table I. Estimatedmultiplicative differenceinpathogen prevalencefromstrong toweakcolonies indesignated
samplingperiod.
Timeperiod Estimatedmultiplicative Lower Upper
difference 95% 95%
Before(October–December) 1.05 0.66 1.66
During(February) 1.17 0.68 2.03
After(March–April) 1.31 0.89 1.93
After2(June) 0.65 0.34 1.23
I.Caviglietal.
ƒ
Nos. Figure4.Distributionofhoneybeepathogensdetectedin
4%
LSV3 C.m. / L.p. coloniesbefore,during,andafteralmondpollination.Hon-
1% 10%
BQCV eybeesampleswereobtainedfrommonitorcoloniesfrom
LSV2 30% Oct.-Dec. October2013toJune2014.PCRwasusedtotestfor16
15% honeybee-infectingpathogensincluding:viruses(ABPV,
before
BQCV, CBPV, DWV, IAPV, KBV, SBV, LSV1, LSV2,
L1S0V%1 D7W%V LSV3,LSV4,andLSV5,microsporidia(N.ceranae),bac-
SBV teria (P. larvae and M. plutonius), and trypanosomatids
23%
(C.m./L.p.).Pathogenoccurrence,thefrequencyofdetect-
ingaspecificpathogenasapercentageofthetotalnumber
of positive tests, was assessed at different time periods
throughoutthestudy.Therewere155positivetestsinbee
Nos.
C.m. / L.p. 15% BQCV samplescollectedbeforealmondpollination(October–De-
4% 24% cember2013),79positivetestsduringalmondpollination
P. larvae
1% (February), 223 positive tests immediately after almond
L2S%V 3 SBV February pollination(March–May),and130positivetestsfrom20
8% coloniesthatweresampledinJune2014.
during
LSV2 LSV1
33% 13%
greater loads of these pathogens. Surprisingly,
therehavebeenfewhoneybeecolonymonitoring
studies in the US; therefore, the data presented
Nos.
13% BQCV hereinprovideanimportantexaminationofcolo-
18%
nyhealthandpathogenprevalenceandabundance
C.m14. %/ L .p. D6W%V March-April inWesternUS-basedhoneybeecoloniesinvolved
P. larvae after
1% SBV inalmondpollination.
LSV4 LSV3 15%
1% 8% There are multiple biotic and abiotic factors
L1S7V%2 L7S%V 1 that contribute to honey bee colony health and
survival. The focus for this study was pathogen
prevalence and abundance, since several studies
indicate that pathogens affect colony health.
Nos. BQCV Specifically, one US-based study that compared
KBV 19% 22%
1% multiple variables in CCD and non-CCD-
DWV
LSV4 3% June affected colonies determined that weak or dead
9%
SBV after2 colonies were more likely to neighbor other
15%
weak or dead colonies and that CCD-affected
LSV3 LSV2
5% 23% samples had a higher pathogen incidence than
LSV1 controls, whereas pesticide levels were
3%
„
Figure5.Colonyhealthandpathogenabundancebefore,
incidence and abundance have been associated during,andafteralmondpollinationforeachbeekeeping
withCCD-affectedcoloniesandcolonylossesin operation.Honeybeesampleswereobtainedfrommonitor
the US (Chen et al. 2014; Cornman et al. 2012; colonies from October 2013 to June 2014. Quantitative
PCR (qPCR) was used to determine the abundance of
Cox-Foster et al. 2007; Daughenbaugh et al.
BQCV,LSV2,LSV1,andSBV;totalpathogenabundance
2015; Li et al. 2013; van Engelsdorp et al.
is the sum of these values (y-axis = log natural of total
2009),Canada(vanderZeeetal.2012),Austria pathogenabundance).Intotal,therewere53observations
(Berényi et al. 2006), Denmark (Nielson et al. of total abundance in a subset of samples analyzed by
2008), Luxembourg (Clermont et al. 2014) and qPCR. Each sample is represented by its unique colony
identifier (e.g., “C4” for colony number 4 inbeekeeping
Belgium (Ravoet et al. 2013). Therefore, we hy-
operation3);iconsizeandshapewereusedtographically
pothesized that weak colonies would harbor a
illustrate colony strength over the course of the study,
greater number of pathogens, as well as have October2013–June2014(x-axis).