S Tribolium castaneum (Coleoptera: Tenebrionidae

STORED-PRODUCT

Tribolium castaneum (Coleoptera: Tenebrionidae) Associated WithRice Mills: Fumigation Efficacy and Population Rebound

KARRIE A. BUCKMAN,1,2 JAMES F. CAMPBELL,1 AND BHADRIRAJU SUBRAMANYAM3

J. Econ. Entomol. 106(1): 499Ð512 (2013); DOI: http://dx.doi.org/10.1603/EC12276

ABSTRACT The red ßour beetle,Triboliumcastaneum(Herbst) (Coleoptera: Tenebrionidae), is themost important stored-product insect pest infesting rice (Oryza sativa L.) mills in the United States.Due to the phasing out of methyl bromide in accordance with the 1987 Montreal Protocol, the efÞcacyof alternative fumigants in controlling ßour beetles in mill structures must be evaluated. Long-termtrapping data sets (2Ð6 yr) of T. castaneum in and around seven rice mills were analyzed to assess theefÞcacy of sulfuryl ßuoride fumigation (n� 25). Fumigation efÞcacy was evaluated as the percentagereduction in mean trap captures of adults and proportion of traps capturing at least one adult beetle.Beetle trap captures ßuctuated seasonally, with increased captures during the warmer months,JuneÐSeptember, that dropped off during the cooler months, OctoberÐMarch. Fumigations resultedin a 66 � 6% (mean � SE) reduction in mean trap captures within mills and a 52 � 6% reduction inthe proportion of traps capturing at least one adult beetle. Lengths of time for captures to reachprefumigation levels, or rebound rates, were variable, and adult capture levels inside were mostinßuenced by seasonal temperature changes. Temperatures inside mills followed those outside the millclosely, and a signiÞcant positive relationship between outside temperatures and trap captures wasobserved. Insideandoutside trapcapturesexhibiteda signiÞcant, positive relationship,but fumigationsconsistently led to reductions in beetle captures outside of mills, highlighting the interconnectednessof populations located inside and outside mill structures.

KEY WORDS red ßour beetle, fumigation, sulfuryl ßuoride, pheromone trap, monitoring

Rice (Oryza sativaL.) cultivation in the United Statesis restricted to �1.2 million ha in Arkansas, California,Louisiana, Mississippi, Missouri, and Texas (Snyderand Slaton 2001). In spite of the small area devoted torice production, it accounts for �11% of world riceexports (Childs and Burdett 2000), making rice pro-duction and processing important components of U.S.agriculture. In the United States, there are 53 rice millswith an estimated annual production of �6.1 millionmetric tons (Richardson and Outlaw 2010). The ricemilling process begins with the removal of the inediblerice hull from rough rice, resulting in brown rice thatretains the bran layers. Brown rice can be furtherprocessed by removing the bran layers, yielding clean(i.e., milled, white) rice. Depending on the facility,additional processing steps also may be involved, in-cluding parboiling and milling into ßour.

Insects can cause economic losses because of theirability to infest rough and milled rice and the milling

by-product material that accumulates in structuresand equipment where rice is processed. Mills receiverough rice and store it in large bins or warehouses ator near the milling facility. Insects infesting rough riceare internal feeders which develop inside intact ricekernels [e.g., Sitophilus oryzae (L.), rice weevil;Sitotroga cerealella (Olivier), Angoumois grainmoth; and Rhyzopertha dominica (F.), lesser grainborer] and external feeders that primarily exploit par-tially and completely milled rice, milling by-products(hull, germ, bran, and broken kernels), and damagedrice kernels (Mutters and Thompson 2009). The pri-mary insect pests of rice milling buildings are Tribo-lium confusum Jacquelin du Val, the confused ßourbeetle, and Tribolium castaneum (Herbst), the redßour beetle (Coleoptera: Tenebrionidae). Packagedproducts stored in warehouses and accumulations offood material in equipment and building structures arealso vulnerable to infestation. Insect movement fromrough rice storage may serve as a source of insectinfestations within mill and warehouse structures(Campbell 2008).T. confusum and T. castaneum can use brown rice,

processed rice, broken kernels, rice bran, dust, anddebris, resulting from the milling process, but theytend to develop more quickly and have a higher fe-cundity on brown rice and rice bran (McGaughey1970, 1974; Imura 1991; Via 1991). Thus, extent of rice

Mention of trade names or commercial products in this publicationis solely for the purpose of providing speciÞc information and does notimply recommendation or endorsement by the U.S. Department ofAgriculture or Kansas State University. USDA is an equal opportunityprovider and employer.

1 USDA, Agricultural Research Service, Center for Grain and An-imal Health Research, 1515 College Ave., Manhattan, KS 66502.

2 Corresponding author, e-mail: [email protected] Department of Grain Science and Industry, Kansas State Univer-

sity, Manhattan, KS 66506.

milling, products produced, and types of by-productscreated at a mill likely affect beetle population growthand the need for ßour beetle management. Economicdamage is caused by direct loss of milling yield due tofeeding; rejection of product due to infestation; cost ofmanagement tactics such as fumigation and insecti-cide fogging; contamination by quinones secreted bybeetles in heavy infestations (Hodges et al. 1996; Mut-ters and Thompson 2009); and perhaps most signiÞ-cantly, loss of consumer trust in the company, product,or both. Managing these pests in rice mills is compli-cated by the difÞculty in assessing actual pest popu-lation density and the difÞculty in applying treatmentsdirectly to populations in cryptic refugia.

Pest management programs in rice mills rely on acombination of tactics including sanitation, residualand aerosol insecticides, and fumigants. Pheromonebaited traps are an important source of information onpest density and can be used to evaluate treatmentefÞcacy (Campbell et al. 2010a,b). However, becausepheromone traps only capture dispersing adults theyindirectly estimate pest abundance. Furthermore, therelationship between capture in traps and level ofinfestation can be affected by other pest managementtactics, especially application of residual insecti-cides, level of sanitation (Toews et al. 2005, 2009),and amount of immigration (Campbell and Arbogast2004).

Fumigation of milling structures is an importantcomponent of pest management of T. confusum and T.castaneum in many rice mills. Fumigations are infre-quently used because they are disruptive to millingoperations because they often take �24 h to perform,and adding set-up time, aeration of the building, andcleanup, mills may cease operations for up to 2Ð3 d. Byusing a calendar-based fumigation schedule, treat-ments can be planned over holidays or other timeswhen production within the mill is slowed or stopped.Considering that fumigations are so costly, both in lossof production and in actual cost of the application, itis critical to ensure that treatments are effective. In thepast, methyl bromide was the principal fumigant usedin structural fumigations (Fields and White 2002).However, methyl bromide was identiÞed as an ozone-depleting substance under the 1987 Montreal Protocol

on Substances that Deplete the Ozone Layer, leadingto an agreement among developed countries to phaseout its use (Fields and White 2002). Due to a lack ofeffective alternatives, the use of methyl bromide in theUnited States has continued in ßour and rice millingfacilities under the critical use exemptions process.Sulfuryl ßuoride is an alternative fumigant that hasbeen adopted by some rice mills. However, assessmentof the efÞcacy of sulfuryl ßuoride fumigation in foodfacilities is limited and restricted to ßour mills (Small2007; Tsai et al. 2011).

We used pheromone trapping at seven rice millslocated in California, Texas, and Louisiana to quantifypopulations ofT. confusum andT. castaneumoccurringwithin and around mill facilities and to assess theefÞcacy of commercial sulfuryl ßuoride structural fu-migations in terms of both initial reduction in capturesand rebound in captures over time after treatment.Because temperature has an important impact on pestpopulation growth and fumigation effectiveness, tem-perature was measured and we explored the relation-ship between captures and temperature both insideand outside the mills.

Materials and Methods

Rice Mills. Populations of T. confusum and T. cas-taneum were monitored in seven rice mills between2005 and 2011. The mills are located in two U.S. re-gions:millsCA1,CA2, andCA3are located innorthernCalifornia (CA), and mills TX1, TX2, TX3, and LA1 inTexas (TX) or Louisiana (LA). The mills weregrouped this way because each region representeddiscrete rice growing areas with different climaticconditions that may have caused differences in fumi-gationefÞcacyandpopulation rebound.Mills varied inconstruction material (i.e., concrete, timber, metal, ora combination of these materials), size, and number ofßoors (Table 1).

Although fumigations were the primary focus of ouranalysis, it is important to note that other practicessuch as sanitation, insecticide fogging, and perimeterinsecticide sprays were used concurrently as part of anintegrated pest management program. The type, fre-quency, and efÞcacy of these additional tactics varied

Table 1. Monitoring information for rice mills

Mill Volume, m3 Monitoring datesNo. trapping

periodsDays per trapping period,

mean � SD (range)No. traps

(insidea/outsideb)

CA1 11,900 12 May 2005Ð26 Dec. 2007 52 18 � 8 (3Ð42) 19/0c, 30/5, 31/5d

CA2 76,700 22 July 2005Ð18 Feb. 2011 108 19 � 6 (11Ð42) 30/5 or 31/5e

CA3 64,400 5 May 2005Ð15 Feb. 2011 113 19 � 6 (11Ð42) 20/2f, 28/7, 30/6e

TX1 152,900 19 Oct. 2005Ð31 Jan. 2007 18 26 � 11 (14Ð56) 30/5TX2 113,300 19 Oct. 2005Ð31 Jan. 2007 18 26 � 11 (14Ð56) 30/5LA1 5,100 6 June 2007Ð15 Sept. 2010 47 25 � 9 (11Ð51) 18/11TX3 6,000 6 Jan. 2009Ð27 Jan. 2011 43 17 � 4 (13Ð28) 12/8

a Located inside the mill structure.b Located outside, near mill structure.c Trap number and location for Þrst trapping period, 12Ð28 May 2005.d Trap number and location after 13 April 2007.e Trap number and location after 6 April 2007.f Trap number and location for Þrst trapping period, 5Ð20 May 2005.

500 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 106, no. 1

among mills. Although it was not expected that thesetactics signiÞcantly affected immediate reduction inpests after fumigation, they could affect rebound rates.Monitoring T. confusum and T. castaneum. Person-

nel at each mill serviced traps independently, solengths of trapping periods varied (Table 1). We setthe standard trappingperiodat14dandcorrecteddatato reßect this standard trapping period for ease ofcomparison. Between 19 and 36 traps were placed atvarious locations inside and outside each mill (Table1). A change in trap number and location occurred atmills CA1, CA2, and CA3 due to a change in shippingmethod, allowing an expanded number of traps.

Traps (Storgard The DOME Trap, Trece Incorpo-rated, Adair, OK) consisted of two interlocking plasticpieces: a top, in which up to three pheromone lurescould be attached, and a bottom, a circular ramp witha central pitfall trap containing a 3.5-cm-diameter Þl-ter paper saturated with kairomone, an oil-based foodattractant (Storgard oil, Trece Incorporated). Beetleswere attracted to the trap by the Tribolium spp. ag-gregation pheromone (Storgard Cap, Trece Incorpo-rated, Adair, OK) and the food attractant. Traps intheir original conÞguration (unmodiÞed traps) wereused throughout the monitoring period at mills TX1and TX2 and until March 2007 at mills CA1, CA2, andCA3. Entire traps were sent to the mills in advance ofthe trap change, and fresh food attractant and, ifneeded, new pheromone lures were added to cleantraps that replaced traps removed from the mill. Entiretraps were placed individually in plastic bags andshipped back to the laboratory at Kansas State Uni-versity for processing.

ModiÞed traps were used at mills CA1, CA2, andCA3, after March 2007, and for the entire monitoringperiod at mills LA1 and TX3. ModiÞed traps weremade by gluing a rivet to the bottom of the pitfallportion of the trap and attaching the pheromone lureto the top of the rivet. A hole was punched in thecenter of Þlter papers to accommodate the rivet. Thisplaced the pheromone in roughly the same location,suspended above the kairomones-treated Þlter paper,as an unmodiÞed trap but enabled trap bottoms fullyloaded with pheromone lures and food attractant to beshipped to and from the mills. This reduced the laborand time required to service traps at the mills.

To limit trap loss, modiÞed traps were attached to ametal base designed to keep traps in place. The metalbase consisted of a perforated metal plate (15.5 by 15.5cm) onto which two 1.5-cm-diameter washers and arotating metal clip were attached. The trap lid edgeswere slid under the washers and the clip rotated toeither lock the trap in place on the metal plate or allowit to be removed for replacement. At the conclusion ofa trapping period, traps were removed from the metalbase and bottoms of the traps switched with second setof trap bottoms with fresh food attractant and, if nec-essary, new pheromone lures. Bottoms were shippedback to the laboratory at the Center for Grain andAnimal Health Research for processing. Bottoms wereplaced in a wooden frame designed to hold the trapsin place and prevent sample loss during shipment.

For both trap types, pheromone lures were replacedapproximately every 8 wk. For modiÞed traps, two setsof pheromone lures were rotated between shipmentsandwere stored ina freezerbetweenplacements,withonly total time in the mills, not time in freezer, usedfor calculating replacement time. Once in the labora-tory, insects were removed from the trap and all T.confusum and T. castaneum adults were enumerated.Fumigations.Mills were fumigated two to six times

during the monitoring period; except mill TX3 thatwas not fumigated (Table 2). All fumigations werewith sulfuryl ßuoride (SF) (ProFume, Dow Agro-Sciences, Indianapolis, IN) with the exception of theÞnal fumigation at mill, LA1, with methyl bromide(MB) (Meth-O-Gas 100, Great Lakes Chemical Cor-poration, West Lafayette, IN). For each SF fumiga-tion, the fumigation service providers used a softwareprogram (ProFume Fumiguide, Dow AgroSciences)that calculated the proper dosage based on the size ofthe space to be fumigated, type of fumigation (i.e., astructure or a commodity), length of the fumigation,pest species and life stages being targeted, half-losstime (HLT), and temperature. SpeciÞc treatment in-formation for all fumigations included in the analysiswas not available. Fumigations for which informationwas available (two SF fumigations at each CA mill)had target accumulated SF dosage of 340Ð500 g-h/m3

and observed accumulated dosages of 350Ð674 g-h/m3. Observed HLTs ranged from 11 to 37 h. The MBfumigation at mill LA1 was performed with a targetdosage of 5.7 kg-h (0.68 kg/28 m3) for a total of 612 kgof MB. Treatment dates, interior temperatures if avail-able, andwindspeedduring fumigationsarepresentedin Table 2.Monitoring Temperature. Temperatures inside

mills CA2, CA3, LA1, and TX3 were monitored be-ginning in December 2008 for mills CA2 and CA3,November 2008 for mill LA1, and March 2009 for millTX3. Small data loggers (SmartButton Data Logger,ACR Systems Inc., Surrey, BC, Canada) were attachedto the bottom of three traps sent to each mill. In millCA2, data loggers were placed on the Þrst and Þfthßoors of the mill and in a warehouse. In mill CA3, dataloggers were placed in the basement and on the sec-ond and seventh ßoors of the mill. In mill LA1, dataloggers were placed on the Þrst and Þfth ßoors of themill and in a processed rice storage warehouse. In millTX3, the data loggers were placed in the mill, dryerelevator tunnel, and elevator tunnel. When traps wereshipped back to the laboratory, data loggers wereremoved, temperature data downloaded, and loggersreturned to the trap. Superßuous temperature datalogged before traps being placed in the mill and aftertraps were removed from the mill were excluded fromthe data set based on shipping and receiving dates.Only temperature data collected from loggers placedinside the mill are included. Outside temperatures andwind speeds for each mill were obtained from localweather stations near each mill (www.wunderground.com). For both inside and outside temperatures, thedaily min/mean/max are presented.

February 2013 BUCKMAN ET AL.: FUMIGATION EFFICACY AND POPULATION REBOUND IN RICE MILLS 501

Tab

le2

.Su

lfur

ylflu

orid

efu

mig

atio

nin

form

atio

nan

dre

duct

ion

inm

ean

no.

ofT

.ca

stan

eum

adul

tsca

ptur

edan

dpr

opor

tion

oftr

aps

wit

hca

ptur

e

Mill

Fum

igat

ion

dat

e

Insi

de

tem

p(�

C)

(min

/mean

/max

)

Outs

ide

tem

p(�

C)

(min

/mean

/max

)

Win

dsp

eed

(km

/h)

(max

/mean

)

Mean

captu

rebefo

refu

mig

atio

n

Mean

captu

reaf

ter

fum

igat

ion

Reduct

ion

inca

ptu

resa

(%)

Pro

port

ion

of

trap

sw

ith

captu

res

befo

refu

mig

atio

n

Pro

port

ion

of

trap

sw

ith

captu

res

afte

rfu

mig

atio

n

Reduct

ion

inpro

port

iona

(%)

CA

128

May

2005

13/1

9/26

34/2

41.

8�

1.0

0.1

�0.

192

0.28

0.10

642

Sept.

2005

13/2

3/33

23/1

69.

8�

5.1

6.7

�2.

732

0.53

0.67

026

May

2006

11/1

7/24

29/1

40.

4�

0.2

0.9

�0.

50

0.23

0.36

01

Sept.

2006

14/2

4/36

16/6

3.7

�1.

10.

2�

0.1

940.

650.

1479

24M

ay20

07N

A/2

7/N

A13

/22/

3219

/11

0.2

�0.

10.

0�

0.0

100

0.13

0.00

100

31A

ug.20

0723

/30/

3926

/13

0.6

�0.

30.

0�

0.0

940.

290.

0389

CA

25

Aug.20

0517

/28/

3916

/64.

3�

1.3

0.3

�0.

192

0.72

0.28

6118

Aug.20

0614

/23/

3219

/10

9.5

�2.

62.

8�

1.3

710.

850.

806

21Ju

ly20

07N

A/2

9/N

A16

/24/

3329

/13

1.2

�0.

20.

2�

0.2

800.

630.

1084

25Ju

ly20

0814

/25/

3619

/54.

3�

1.2

0.3

�0.

194

0.85

0.23

7313

Aug.20

0922

/27/

3118

/27/

3627

/63.

2�

0.7

0.5

�0.

185

0.72

0.35

5120

Aug.20

1020

/24/

2813

/23/

3224

/10

0.5

�0.

20.

2�

0.1

680.

310.

1358

CA

321

May

2005

10/1

8/24

23/8

2.0

�0.

70.

4�

0.2

810.

600.

2656

3A

ug.20

0614

/24/

3426

/11

13.7

�3.

00.

6�

0.2

960.

820.

3064

14Ju

ly20

07N

A/2

7/N

A14

/24/

3326

/13

1.3

�0.

40.

2�

0.1

830.

500.

2060

8A

ug.20

0813

/21/

2927

/16

2.1

�1.

11.

2�

1.1

650.

370.

1755

7A

ug.20

0921

/23/

2513

/22/

2923

/62.

3�

1.3

1.0

�0.

759

0.46

0.30

3512

Aug.20

1018

/23/

2912

/22/

3224

/61.

5�

0.5

0.6

�0.

458

0.50

0.30

40T

X1

11N

ov.20

0519

/22/

2616

/60.

9�

0.3

0.7

�0.

725

0.40

0.08

8015

Apri

l20

0621

/24/

2940

/19

0.3

�0.

10.

3�

0.1

110.

440.

3324

4Ju

ly20

0624

/27/

2923

/60.

8�

0.2

0.5

�0.

135

0.57

0.52

8T

X2

11N

ov.20

0519

/22/

2616

/65.

2�

1.6

0.2

�0.

196

0.72

0.19

7415

Apri

l20

0621

/24/

2940

/19

0.7

�0.

30.

3�

0.2

600.

460.

2056

4Ju

ly20

0624

/27/

2923

/61.

8�

0.5

0.8

�0.

257

0.79

0.43

46L

A1

13Ju

ne

2008

NA

/33/

NA

23/2

8/33

26/8

1.0

�0.

30.

5�

0.2

460.

710.

5030

24Ju

ly20

09b

27/2

9/33

23/2

8/33

11/2

1.3

�0.

60.

6�

0.3

540.

530.

2848

Mean

�SE

Mc

2.9

�0.

70.

8�

0.3

66�

60.

54�

0.04

0.28

�0.

0452

�6

NA

,n

ot

avai

lable

.a

Wh

en

there

was

anin

creas

ein

beetl

eca

ptu

reor

pro

port

ion

of

trap

sw

ith

captu

res

afte

ra

fum

igat

ion

,perc

en

tage

reduct

ion

was

zero

.b

Fum

igat

ion

wit

hm

eth

yl

bro

mid

e.

cM

ean

sdo

not

incl

ude

meth

yl

bro

mid

efu

mig

atio

nat

mill

LA

1.


Statistical Analysis. T. confusum and T. castaneumpopulations within the mills were quantiÞed in twoways: beetle capture per trap and number of traps withone or more beetles. Beetle capture data were ex-pressed as the mean number of T. confusum or T.castaneum per trap inside or outside the mill (here-after mean inside or outside trap captures) correctedto reßect the standard trapping period. Traps captur-ing one or more beetles was expressed as the propor-tion of inside or outside traps capturing at least onebeetle over the sampling period (hereafter the pro-portion of inside or outside traps with captures).

EfÞcacy of SF fumigations was measured as per-centage reduction in mean inside trap captures andthe percentage reduction in the proportion of insidetraps with captures. The efÞcacy of the MB fumigationwas not analyzed due to the lack of replication. Thesemeasures of efÞcacy were compared among mills us-ing analysis of variance (ANOVA) (PROC GLM, SAS9.2, SAS Institute, Cary, NC). The same measures offumigation efÞcacy were compared between regions,CA and TX/LA, by using a StudentÕs t-test (PROCTTEST, SAS 9.2, SAS Institute). To fulÞll the normalityassumption, the arcsine square-root transformationwas applied to percentage reduction in mean beetlesper inside trap. PearsonÕs correlation coefÞcients werecalculated to explore the relationship of fumigationefÞcacy and rate of rebound with maximum, mean,and minimum temperatures and maximum and meanwind speed during fumigations (PROC CORR, SAS9.2, SAS Institute). PearsonÕs correlation coefÞcientsalso were calculated to explore the relationship be-tween fumigation efÞcacy and mean captures per in-side trap during the sampling period immediately pre-ceding treatment (PROC CORR, SAS 9.2, SASInstitute).

Time-to-event analysis was used to assess the rate ofrebound of beetle populations after fumigation. Weused the thresholds established by Campbell et al.(2010b) for Tribolium in ßour mills of 2.5 beetles pertrap per 14 d and 50% of traps capturing at least onebeetle. These thresholds were the median beetles perinside trap and proportion of traps with captures dur-ing sampling periods immediately before fumigationsin two midwestern ßour mills (Campbell et al. 2010b).Because the median beetles per inside trap and pro-portion of traps with captures in rice mills was 1.6 and53%, respectively, time-to-event analyses also wereperformed using these thresholds. Number of dayspostfumigation until each of these thresholds was metor exceeded was determined and used in the time-to-event analysis (KaplanÐMeier single group survivalanalysis, SigmaPlot 12.2, Systat Software, Chicago, IL).To determine whether mill region affected reboundrate, we also performed a time-to-event analysis com-paring rebound rates between the CA and TX/LAregions (KaplanÐMeier log-rank survival analysis,SigmaPlot 12.2, Systat Software). Cases where thethreshold was not reached before another fumigationor before the conclusion of monitoring were censored.PearsonÕs correlation coefÞcients were calculated todetermine whether mean inside trap captures imme-

diately before fumigation was related to the rate ofrebound (PROC CORR, SAS 9.2, SAS Institute).

We used the mean daily inside and outside temper-atures to evaluate the relationship between beetlecaptures and temperature. We performed linear re-gressions to determine whether increased beetle cap-tures outside the mill led to increased captures insidethe mill, and whether changes in outside temperatureled to changes in mean inside beetle capture or inproportion of traps with captures (PROC GLM, SAS9.2, SAS Institute). Linear regression also was used todetermine the extent to which outside temperatureswere related to inside temperatures (mills CA2, CA3,LA1, and TX3). Mill was included as a factor in theregressions, so that if a signiÞcant relationship be-tween thevariableswas found, thenatureandstrengthof the relationship could be compared among mills bycomparing slope estimates among mills (CONTRASTstatement, PROC GLM, SAS 9.2, SAS Institute). Tocompensate for multiple comparisons, we used a Bon-ferroni correction, reducing the critical P value from0.05 to 0.0024 for the regressions of mean inside cap-tures on mean outside captures, mean inside captureson outside temperature, and proportion of traps withcaptures on outside temperature and to 0.0083 for theregression of inside temperature on outside temper-ature. Data are presented as untransformed means �SEM, unless otherwise noted.

Results

T. castaneum was captured in traps at all the mills;however, T. confusumwas captured at only four of themills: CA1 (0.40% T. confusum), CA2 (0.07% T. con-fusum), CA3 (1.08% T. confusum), and LA1 (0.03% T.confusum). Most of the T. confusum individuals werecaptured at mill CA3, 100 total individuals, whereas atmills CA1, CA2, and LA1 only 13 individuals in totalwere captured. Because so few T. confusumwere cap-tured, we focused our analysis on trap captures of T.castaneum of which 26,600 total individuals were cap-tured.

Mean inside trap captures followed a distinctly sea-sonal pattern in most of the mills, with highest cap-tures recorded in the warm months, and dropped offas outdoor temperatures cooled (Fig. 1). This trendwas most apparent in mills monitored for several years,i.e., mills CA1, CA2, CA3, LA1, and TX3, but it was lessobvious in mills TX1 and TX2 that were monitored forjust over 1 yr (Fig. 1). Proportion of inside traps withcaptures followed a similar trend (Fig. 2). The meanchange in trap capture between two consecutive sam-pling periods was close to zero, �0.0 � 0.1 beetles pertrap, indicating that no clear trends or patterns wereobserved. Similarly, the mean change in proportion oftraps capturing at least one beetle was close to zero,�0.00 � 0.01.

In total, 25 SF fumigations were performed duringthe monitoring periods at the six treated mills. Fumi-gations resulted in a 66 � 6% reduction in beetlescaptured per trap, from 2.9 � 0.7 per trap before SFfumigation, to 0.8 � 0.3 per trap in the sampling period


after fumigation (Table 2). Fumigation efÞcacy, asmeasured by the percentage reduction in mean insidetrap captures, did not signiÞcantly vary by mill (F �1.63; df � 5, 19; P� 0.2012). The impact of region onreduction in capture was marginally nonsigniÞcant(t � 1.94, df � 23, P � 0.0646), with reductions incaptures of 73 � 6% for CA mills and of 47 � 11% forTX/LA mills. Fumigations resulted in a 52 � 6% re-duction in proportion of traps capturing at least onebeetle during the monitoring period. Before fumiga-tion, the proportion of traps with captures was 0.54 �0.04,whichwas reduced to0.28�0.04after fumigation(Table 2). Reduction in proportion of traps with cap-tures did not signiÞcantly vary among mills (F� 0.31;df � 5, 19; P � 0.9024) or between regions (t � 0.69,df � 23, P� 0.4970). Reduction in beetle capture andproportion of traps with captures were not signiÞ-cantly correlated with mean or minimum temperatureor maximum or mean wind speed during fumigation.Reduction in proportion of traps with captures was notcorrelated with maximum temperature; however, re-duction in beetle capture was signiÞcantly, positively

correlated with maximum temperature during fumi-gation (� � 0.477, n � 25, P � 0.0160). FumigationefÞcacy, reduction in trap capture and reduction inproportion of traps with captures, were not signiÞ-cantly correlated with prefumigation trap captures.

Rebound in trap captures and in proportion of trapswith captures after fumigation in CA mills seemed tobe inßuenced by seasonal temperature ßuctuations.Regardless of the date of fumigation, there is a con-sistent reduction in both beetles per trap and propor-tionof trapswithcapturesat theCAmills fromJanuaryto March, with a corresponding increase from May toSeptember (Fig. 3a and c). This same pattern wasobserved in TX/LA mills, but the trend was less clearlyevident (Fig. 3b and d).

After SF fumigation, the threshold of 2.5 beetles perinside trap, the median beetles per trap observed inßour mills, was reached after 270 � 31 d. However, ofthe 25 fumigations, only 11 reached this thresholdbefore the next fumigation or the end of the moni-toring period (Fig. 4a). Time to reach 1.6 beetles perinside trap, the median beetles per trap immediately

Fig. 1. Mean (� SEM) T. castaneum per inside trap (bar graphs) with outside (gray line) and inside (black line)temperatures (�C) for the seven mills. Inside temperatures were only available for part of the monitoring period for millsCA2, CA3, LA1, and TX3. Arrows and change in color of bars indicate a fumigation. Mill TX3 was not fumigated while beingmonitored.


before fumigation in the mills was 189 � 28 d, with 18of the 25 fumigation events reaching this thresholdbefore the next fumigation or termination of moni-toring (Fig. 4b). Time to reach the threshold of 50% oftraps capturing at least one beetle, median beforefumigation in ßour mills, was shorter, only 183 � 30 d,and was reached after 18 of the 25 fumigations (Fig.4c). It took 202 � 31 d to reach the threshold of 53%of traps capturing at least one beetle, the prefumiga-tion median in the rice mills and 16 of the 25 fumi-gations reached this threshold before the next fu-migation or the end of the monitoring period (Fig.4d). The time to reach these thresholds did notsigniÞcantly vary between mill regions (P � 0.05).Rate of rebound to the prefumigation median of 53%of traps capturing at least one beetle was signiÞcantly,positively correlated with minimum temperatureduring fumigation (� � 0.713, n � 16, P � 0.0019).Rebound rate was not signiÞcantly correlated withprefumigation trap captures or with maximum tem-perature, mean temperature, or wind speed duringfumigation.

One fumigation at mill LA1 was with MB rather thanSF (Table 2). EfÞcacy of the two fumigants could notbe compared directly due to the lack of replication.However, the MB fumigation resulted in reductions intrap capture and proportion of traps with captures thatwere similar to the mean for the SF fumigations and tothe earlier SF fumigation at mill LA1 (Table 2). Thepattern of rebound after the MB fumigation was alsosimilar to that of the SF fumigations at the TX/LA mills(Fig. 3).

There was a signiÞcant, positive relationship be-tween mean captures per trap inside mills and meancaptures per trap outside mills (all mills combined:r2 � 0.439, df � 394; MSE � 1.86; P � 0.0001; Fig. 5).Thus, increased captures in traps located outside themill was associated with increased captures inside themill. Both the intercept and slope of the regression linewere signiÞcantly different from zero (intercept: F�9.53, df � 7, P � 0.0001; slope: F � 37.19, df � 7, P �0.0001). Contrasts comparing the slope of the regres-sion lines among mills indicated that mill TX2 had asigniÞcantly greater slope, 1.822 � 0.413, than mills

Fig. 2. Proportion of inside traps capturing at least oneT. castaneumduring a monitoring period for the seven mills. Changein color of bars indicates a fumigation. Mill TX3 was not fumigated while being monitored.


CA1 (slope � 0.147 � 0.017,P� 0.0001), CA2 (slope �0.074 � 0.012,P� 0.0001), CA3 (slope � 0.153 � 0.014,P� 0.0001), TX1 (slope � �0.056 � 0.268,P� 0.0002),and LA1 (slope � 0.123 � 0.068, P� 0.0001) (Fig. 3).Mills CA1 (P � 0.0005) and CA3 (P � 0.0001) hadsigniÞcantly greater slopes than mill CA2 (Fig. 5). Theslope of the regression line for mill TX3 was not sig-niÞcantly different from that of any other mill, andmills TX1 and TX3 were the only mills with negativeslopes; but the negative slopes were not signiÞcantlydifferent from zero (P � 0.05) (Fig. 5) indicating norelationship between inside and outside captures.

Outside temperature and mean beetle capture perinside trap had a signiÞcant, positive relationship(r2 � 0.199, df � 395; MSE � 2.64; P� 0.0001; Fig. 6),indicating that as outdoor temperatures increased,trap captures inside the mill also increased. The in-tercept was not signiÞcantly different from zero (F�1.59, df � 7, P � 0.1362); however, the slope of theregression line was signiÞcantly different from zero(F� 9.72, df � 7,P� 0.0001). Contrasts comparing the

slopes of the regression lines did not indicate signiÞ-cant differences in estimated slope among the mills.Outside temperature and proportion of inside trapswith captures also had a signiÞcant, positive relation-ship (r2 � 0.426, df � 395; MSE � 0.03;P� 0.0001) andcontrasts indicated estimated slopes of regression linessigniÞcantly varied among mills. Mill CA2 had a sig-niÞcantly greater slope, 0.0234 � 0.0029, than mill LA1,0.0067 � 0.0040 (P � 0.0007). All other mills hadsimilar slope estimates.

In those mills where inside temperature was mon-itored, inside and outside temperatures were similar,regardless of time of year (Fig. 1). Linear regressionindicated that the relationship was signiÞcant and pos-itive (r2 � 0.934, df � 2896; MSE � 2.67; P � 0.0001;Fig. 7). Contrasts indicated the estimated slopes ofregression lines varied among mills, with mills LA1 andTX3 having signiÞcantly larger slopes 0.8830 � 0.008and 0.8847 � 0.008, respectively, than mills CA2 andCA3, 0.7368 � 0.009 and 0.6434 � 0.009, respectively(P� 0.0001 for all comparisons). In addition, slope for

Fig. 3. Rebound by month of the year of T. castaneum captured per trap (A and B) and proportion of traps with captures(C and D) after fumigation at CA mills (A and C) and TX/LA mills (B and D).


mill CA2 was signiÞcantly greater than that of mill CA3(P � 0.0001). The large r2 values and slopes close toone indicate that temperatures inside the mill followthose outside the mill closely, but that CA2 and CA3tended to stay warmer at cooler outside temperaturesthan LA1 and TX3.

Discussion

Seasonality andEfficacy ofFumigation.Trends inT.castaneum captures inside the seven rice mills fol-lowed seasonal patterns in which captures and pro-portion of traps capturing at least one beetle increasedthroughout spring, peaked in summer months, andthen dropped off during the colder winter months.This is a different seasonal pattern than observedin trap captures in midwestern ßour mills; beetle cap-tures in ßour mills tended to increase, albeit at differ-ent rates, throughout the year (Campbell et al. 2010a).In the ßour mills, fumigations appeared to have astronger impact than seasonality on rebound, whereasin rice mills season seems to have a stronger inßuenceon beetle captures than time after fumigation. Sea-sonality may have a weaker inßuence on beetle cap-tures in ßour mills because ßour mill interiors tendedto be buffered from extreme outside conditions thatresulted in a nonlinear relationship between insideand outside temperature (Campbell et al. 2010a). Ricemills did not exhibit similar buffering of indoor tem-peratures (Fig. 7), the possible causes of which arediscussed below; thus, beetle populations would havebeen exposed to climatic extremes.

Fumigations in the ßour mills were more effectiveat reducing trap captures with a mean reduction of85 � 5% (Campbell et al. 2010a) at mills in the mid-western United States and 95 � 1% at mills in theUnited Kingdom (Small 2007), both higher than themean reduction observed in rice mills of 66 � 6%(Table 2). Differences in efÞcacy of fumigations mighthave been caused by differences in fumigant used; ricemills were fumigated with SF, whereas ßour mills were

Fig. 4. Time-to-event curves for rebound to threshold of2.5 T. castaneum per trap (A), 1.6 T. castaneum per trap (B),50% of traps capturing at least one T. castaneum (C), and 53%of traps capturing at least one T. castaneum (D) over astandardized 14-d sampling interval. Data from the six millsundergoing fumigations during the monitoring period werecombined. Closed circles indicate censored events (the nextfumigation occurred before reaching the threshold or mon-itoring was terminated before reaching the threshold).

Fig. 5. Relationship between mean T. castaneum capture in traps located inside the mill and outside the mill. Estimatedslope of regression lines included below panel label. Different letters after the regression line indicate the difference in theestimate of the slope of the regression line is signiÞcant.


fumigated with MB (Campbell et al. 2010a; Table 2).However, when SF and MB fumigations were com-pared directly in wheat (Triticum aestivum L.) ßourmills, efÞcacy of SF and MB fumigations measured asreduction in captures immediately after treatment

have been comparable (Small 2007, Tsai et al. 2011).Multiple studies have reported that higher doses of SFare required to control the Tribolium spp. egg stagecompared with other beetle developmental stages(Bell et al. 1999, Hartzer et al. 2010). However, trapcaptures in the sampling period immediately afterfumigation will initially reßect mortality of adults andpupae, as larvae and eggs would not have becomeadults by this time and therefore would not have beencaptured in traps. Thus, if the dosage of SF was nothigh enough to cause egg mortality, T. castaneum eggsurvival would not become apparent for weeks ormonths after fumigation, depending on temperatureand food quality.

There are two likely reasons why observed fumi-gation efÞcacy was lower in the rice mills compared

Fig. 6. Relationship between mean daily outside tem-perature (�C) and mean inside T. castaneum capture (leftcolumn of panels) and proportion of inside traps capturing atleast one T. castaneum (right column of panels). Estimatedslopes of regression lines included below panel label. Slopesdid not signiÞcantly vary among mills for mean T. castaneumcapture. For proportion of traps with captures, regressionlines followed by different letters indicate a signiÞcant dif-ference between estimated slopes of the regression lines.

Fig. 7. Relationship between inside and outside temper-atures (�C) in those mills where inside temperatures weremonitored. Regression lines followed by different lettersindicate signiÞcant differences between estimated slopes ofregression lines.


with similar treatments in ßour mills. First, lower bee-tle densities were observed at the rice mills beforefumigation, 2.9 � 0.7 beetles per trap (Table 2), com-pared with prefumigation captures at the ßour mills of17.4 � 6.2 and 5.8 � 2.9 (Campbell et al. 2010a).Because beetle captures at the rice mills were fre-quently low(i.e., less than twobeetlesper trap)beforefumigation, reducing captures even further would notnecessarily be expected (Table 2). At such low ratesof capture, capture rates were likely difÞcult to quan-tify accurately and stochastic events may have had alarger inßuence on captures, both of which wouldobscure detection of a reduction in beetle capture.Furthermore, at Midwestern ßour mills, when cap-tures before fumigation were less than two per trap,captures were reduced by 76 � 11%; however, whenprefumigation captures were more than two per trap,captures were reduced by 91 � 2% (Campbell et al.2010a).

Second, efÞcacy may have been lower for these ricemills because evidence suggests that beetle popula-tions were not restricted to the mill, in contrast topopulations in ßour mills. A signiÞcant positive rela-tionship between inside and outside trap captures inÞve of the seven mills and higher beetle capturesoutside the mill compared with inside the mill in fourof the mills supports the conclusion that T. castaneumpopulations may move freely between indoor andoutdoor habitats (Fig. 5; Campbell and Arbogast 2004;Campbell et al. 2010a). Thus, these nonfumigated out-door areas could serve as a source for colonization ofthe mills after fumigation, lowering the apparent ef-Þcacy. These two factors alone and in combinationmight explain the lower immediate reduction in cap-ture after fumigation.

Although nonsigniÞcant, the difference in efÞcacyof fumigation between the two rice growing regionsshowed a trend of higher efÞcacy in the CA mills, 73 �6% reduction in trap capture, compared with theTX/LA mills, 47 � 11% reduction in trap capture(Table 2). Many factors could have contributed to thisdisparity, including differences in building size orstructure, SF dosage, weather conditions, differencesin pest management strategies, or a combination. Fora fumigation to be effective, the fumigant must bemaintained at the appropriate concentration for therequired length of time to achieve the prescribedaccumulated dosage within the structure to kill thetarget pest. Differences in treatment methodology,mill structure, or availability of refugia may have re-duced the accumulated dosage to which beetles wereexposed, possibly contributing to differences in efÞ-cacy observed between the two regions. Although acomparison of achieved accumulated dosages andHLTs between regions, among mills, and among fu-migation events might help to explain the observeddifferences in efÞcacy, we do not have enough speciÞctreatment data to make such comparisons. However,we do know that all SF fumigations were made ac-cording to the label (i.e., the Fumiguide calculation);thus, our conclusions in regard to the efÞcacy of theSF fumigations are based on the label rate and not a

speciÞc accumulated dosage. Number of SF fumiga-tions performed might also have been a factor explain-ing differences between regions, although this is dif-Þcult to quantify. Our data set indicated a largernumber of consecutive treatments with SF in the CAmills than at the TX/LA mills. The higher level ofexperience using this fumigant might have led to moreeffective SF fumigations in the CA facilities, especiallyconsidering that at the TX/LA mills some treatmentsrepresent Þrst experience with this fumigant. Thus,there are many factors related to treatment with thepotential to affect fumigation efÞcacy which were notmeasured during fumigations included in our analyses.It is possible that a combination of these factors con-tributed to the relatively large, though nonsigniÞcant,differences in fumigation efÞcacy between regions.

Weather conditions during fumigation, such as tem-perature and wind speed, also have the potential toaffect efÞcacy of fumigation. Increasing the temper-ature from 25 to 30�C reduced the required accumu-lated dosage of SF to kill T. castaneum eggs from 1,700g-h/m3 to 1,150 g-h/m3 (Bell et al. 1999). Thus, atlower temperatures, it may be difÞcult to reach therequired accumulated dosage to kill all the life stagesof T. castaneum. In the rice mills studied, maximumdaily temperature exhibited a signiÞcant, positive cor-relation with reduction in beetle capture after fumi-gation. This supports the positive relationship be-tween efÞcacy of fumigation and temperature. Highwind speeds may also increase the rate of SF losswithin fumigated structures which may lead to re-duced accumulated dosages and reductions in efÞ-cacy. However, we did not detect any signiÞcant re-lationships between maximum or mean wind speedsand fumigation efÞcacy or rate of rebound in thisstudy.Population Rebound After Fumigation. The rate at

which beetle populations rebound to prefumigationlevels is an additional measure of fumigation efÞcacy.Rebound rate depends on beetle survival after fumi-gation, beetle immigration and emigration duringand/or after fumigation, mill temperature, and addi-tional management tactics used in the mill. Increasedbeetlemortality, preventionofbeetle immigrationandemigration, cooler temperatures, and use of additionalmanagement such as sanitation can all slow the re-bound rate. We used thresholds of 1.6 and 2.5 beetlescaptured per trap, thresholds that refer to medianbeetles captured per trap during the prefumigationsampling periods for rice and ßour mills (Campbell etal. 2010b), respectively, and 53% and 50% of traps withcaptures, values referring to median proportion oftraps with captures during the prefumigation samplingperiods for rice and ßour mills (Campbell et al. 2010b),respectively, to analyze rebound rates. Rebound ratesat the rice mills were highly variable, ranging from 19 dto 369 d, depending on the threshold applied (Fig. 4).In addition, in multiple instances beetle capture didnot reach the threshold before the next fumigation(Fig. 4). Although these thresholds are somewhat ar-bitrary in that they are not based on balancing eco-nomic loss due to product contamination with loss due


to the cost of treatment, they are useful in assessingrebound rate and in comparing rebound rates be-tween mill types.

Rebound in ßour mills occurred more rapidly thanin rice mills, which may be due to higher prefumiga-tion beetle densities in ßour mills, identiÞed as a factorrelated to rebound rate (Fig. 4; Campbell et al. 2010b).Warmer temperatures observed in ßour mills, partic-ularly over the winter, also may have contributed tothe more rapid rebound. Furthermore, rebound inßour mills was more consistent in that T. castaneumcaptures were sharply reduced immediately after fu-migation and then steadily increased until the nextfumigation. These trends were particularly evidentwhen season of fumigation was taken into account(Campbell and Arbogast 2004; Campbell et al. 2010b).The same trends were not observed in rice mills asfumigations did not always lead to sharp declines inbeetle captures nor was rebound consistent over time(Figs. 1 and 2). Small (2007) monitored T. confusumpopulations for up to 84 d postfumigation, and for bothSF and MB fumigations, beetle populations did notrebound to prefumigation densities within this timeframe.

Relationships between temperatures outside andinside of mills varied between rice mills in California,Texas, and Louisiana and Midwest ßour mills (Fig. 7;Campbell et al. 2010a). Inside temperatures in ricemills were more similar to outdoor temperatures andthis disparity was most pronounced during the winter,whereas ßour mills tended to follow outside temper-atures only during the warm season. Temperatures inrice mills are not as buffered from outside conditions(Fig. 7), and rice mills do not seem to be heated duringthe cooler months. In addition, the machinery used tomill rice may produce less heat than in ßour millsbecause of a reliance on belts and gravity rather thanthe pneumatics and motorized equipment used inßour mills and the reduced amount of processing ofthe grain that occurs. In addition, many rice mills aretypically under negative building air pressure. Allthree of these differences could contribute to thedifferences in temperature ßuctuations between riceand wheat ßour mills. Flour mill interior temperaturesduring the cool season were generally stable due toheating of the mills and maintained temperatures thatsupport beetle development year-round (Campbell etal. 2010a). In warm season, ßour mill temperaturestended to be warmer than outside, whereas in ricemills inside and outside temperatures were more sim-ilar. Cooler temperatures inside rice mills would havedirectly affected population growth inside mills andalso indirectly affected indoor populations via a cor-responding reduction in outside activity and move-ment of individuals into mills; both may have contrib-uted to the increased time to rebound observed.

Food quantity and quality also may have causedbeetle captures in ßour mills to rebound more quicklythan in rice mills. The amount of Þne material pro-duced when grinding grain into ßour would be ex-pected to be larger, allowing for larger accumulationsof food material in ßour mills compared with rice mills.

However, this is a factor that still needs to be directlymeasured. T. castaneum development is also delayedon rice products when compared with wheat millingby-products (Imura 1991). Via (1991) found that twostrains ofT. castaneum tended to develop more quicklyand reach larger population sizes on wheat ßour com-pared with rice ßour, although the improvement wasnonsigniÞcant for either strain. When reared on rice,ßour beetle fecundity was reduced as the extent ofmilling was increased (McGaughey 1974). Rice millsproducing primarily polished white rice will have lessof the preferred brown rice readily available as a foodsource. Over several generations such small increasesin development time or reductions in fecundity couldcontribute to the lower beetle densities and slowerrebound rates observed in the rice mills (Throne1989).

Interestingly, at the ßour mills, the rate of beetlerebound was signiÞcantly affected by the season dur-ing which the fumigation was performed, with re-bound occurring most rapidly after summer fumiga-tions and least rapidly after fall fumigations (Campbellet al. 2010a). In rice mills, however, the season duringwhich fumigation was performed did not seem toinßuence the rate of rebound. Most of the rice millfumigations tookplaceduring the summer(62%),withfewer in the spring (23%) and fall (15%) (Table 2).With so few fumigations represented for the springand fall, there was probably not enough replication tomake an accurate comparison in rebound rate amongthe seasons at the rice mills. With additional rebounddata from spring and fall fumigations in rice mills, aseasonal trend may emerge, which would be expectedconsidering the strong impact of seasonality on trapcaptures (Fig. 1).Inside Versus Outside T. castaneumCapture.A sig-

niÞcant, positive relationship between mean insidetrap captures and mean outside trap captures wasobserved, with higher mean captures in traps locatedoutside of four of the seven rice mills (Fig. 5). It isdifÞcult toassesswhetherbeetles capturedoutside themill were moving into or out of the mill. Higher beetlecapture outside of the mill would suggest movementinto the mill at CA1, CA2, CA3, and LA1. Two of theTexas mills, TX1 and TX3, had a negative relationshipbetween inside and outside mean trap capture andTX2 had higher trap captures inside the mill thanoutside the mill, suggesting that at these mills, emi-gration from the mill may have been occurring (Fig.5). However, for all fumigated mills, outside beetlecapture was consistently reduced in the sampling pe-riod immediately after fumigation implying that fumi-gated areas served as a population source for outdoorpopulations and fumigation eliminated or reduced thissource. Thus, although we cannot conclusively deter-mine whether beetles were moving primarily into orout of the mill, we can conclude that beetle popula-tions inside and outside the mill were interconnected.Similar monitoring at ßour mills showed lower beetlecaptures outside the mill than inside the mill, suggest-ing that beetle populations were likely established inand mostly conÞned to the mill (Campbell and Arbo-


gast 2004). The presence of adult beetles shortly afterfumigation at the ßour mills was likely not due tobeetle movement into mills from outside, but ratherdue to survival of fumigation or movement from non-fumigated areas within the mill (Campbell et al.2010a).Role of Temperature. Temperature is an important

consideration for pest management in rice mills be-cause beetles require ambient temperatures abovecertain minima to develop and to disperse. Coolertemperatures slow beetle population expansion (Erd-man 1964); thereby slowing the rate of populationrebound after a fumigation. Inside trap captures andproportion of traps with captures tended to increaseas outside temperature increased in all seven mills(Fig. 6). This trend would be expected considering thestrong impact of seasonality on beetle captures. Gen-erally, the relationship between proportion of trapswith captures and outside temperature was strongerthan between mean inside trap captures and outsidetemperature. This is not surprising, because as outsidetemperatures increased, so did inside temperatures(Fig. 7), that likely led to an increase in beetle activity.More rapid beetle development and population ex-pansion, coupled with increased movement at highertemperatures would be expected to cause increasedbeetle contact with traps. As beetles migrated to newhabitat patches more often, a larger proportion oftraps were likely to catch at least one beetle. Thepositive relationship between temperature and meanbeetles per inside trap was weaker probably due towide variations in trap captures among trap locations,possibly masking more signiÞcant trends (Fig. 6).

The relationship between inside and outside tem-peratures at rice mills was very close. The regressionof inside temperatures on outside temperaturesyielded high r2 values and slopes close to 1, indicatingthat a 1�C change in outside temperature led to anearly identical change in temperature inside thesemills (Fig. 7). Mills LA1 and TX3 had signiÞcantlylarger estimated slopes of the regression line wheninside mean temperature was regressed on outsidetemperature (Fig. 7). Thus, an increase in outsidetemperature at these mills resulted in a larger increasein inside temperature than an identical change atthe CA2 or CA3 mills. This difference could have beendue to differences in the structure of the buildings.The TX3 mill is a metal and wooden structure makingit difÞcult to seal for fumigation, which is why it wasnot fumigated during the monitoring period.T. castaneum cease embryonic development at tem-

peratures below 17.5�C (Howe 1956); thus, T. casta-neum population growth at temperatures below thisthreshold would not occur. At the midwestern ßourmills, inside temperatures rarely fell below this min-imum over the 7 yr of the study (Campbell et al.2010a). However, at the four rice mills where indoortemperatures were monitored, mean daily tempera-tures frequently fell below 17�C and, during the win-ter, remained below this threshold for extended pe-riods (Fig. 1).

FutureDirections.We have used trap captures andproportion of traps capturing at least one beetle as ameasure of T. castaneum population density; however,it is unknown how pheromone trap captures are re-lated to actual insect densities, further complicatingthe assessment of fumigation efÞcacy. Adult beetlecaptures in traps were used in this study to compareinfestations among different facilities, inside and out-side a facility, and within a speciÞc facility over time.However, because this is an indirect sampling methodcapturing dispersing individuals, the data are poten-tially prone to inßuence by differences in landscapeand abiotic conditions among locations. Variation inprobability of capturing insects may result in capturesnot being related to population trends and may affecttreatment efÞcacy evaluations at the detriment tomanaging pest infestations. Research exploring howbeetle captures vary among locations within and be-tween food processing facilities and the relationshipsbetween captures in traps and other metrics of pestabundance would improve the utility of trapping dataand help validate the trends reported here.

Although the ability to relate trap capture to actualpest densities would allow the development of eco-nomic thresholds, this is unlikely to be achieved con-sidering the difÞculty in quantifying deÞnite densitieswithin structures. Controlled experimentation ex-panding the understanding of factors that inßuenceinsect trap capture would help mill personnel deter-mine whether changes in capture indicate changes inactual density or are caused by other confoundingfactors. Such an increased understanding of pest dy-namics has the potential to prevent unnecessary fu-migations. With the phasing out of MB, any reductionin fumigation frequency will ease the transition toalternative pest management tactics.

Acknowledgments

We thank the management at the rice mills for allowingaccess to facilities and for cooperation in trap shipment. Weare grateful to R. Hammel for technical support with theassistance of J. Ediger, Z. Goldstein, M. Hartzer, C. Persson,D. Robbins, J. Staats, and M. Van Tuyl. We thank threeanonymous reviewers for helpful comments on an earlierdraft. This study was supported in part by USDA RAMP(agreement 2007-51101-18407) and USDA Methyl BromideAlternative (agreement 2011-51102-31125) grants. This arti-cle is contribution 13-013-J from the Kansas AgriculturalExperiment Station.

References Cited

Bell, C.H.,N. Savvidou, andT. J.Wontner Smith. 1999. Thetoxicity of sulfuryl ßuoride (Vikane) to eggs of insectpests of ßour mills, pp. 345Ð350. In Z. Jin, Q. Liang, Y.Liang, X. Tan, and L. Guan (eds.), Proceedings of the 7thInternational Working Conference on Stored-ProductProtection, 14Ð19 October 1998, Beijing, China. SichuanPublishing House of Science and Technology, Chengdu,China.

Campbell, J. F. 2008. Evaluating sources of stored-productinsect infestation, pp. 137Ð157. In R. Mancini, M. O. Car-valho, B. Timlick, and C. Adler (eds.), Contribution for


Integrated Management of Stored Rice Pests. IICT-Insti-tuto de Investigacao CientõÞca Tropical, Lisboa, Portugal.

Campbell, J. F., and R. T. Arbogast. 2004. Stored-productinsects in a ßour mill: population dynamics and responseto fumigation treatments. Entomol. Exp. Appl. 112: 217Ð225.

Campbell, J. F., M. D. Toews, F. H. Arthur, and R. T. Arbo-gast. 2010a. Long-term monitoring of Tribolium casta-neum in two ßour mills: seasonal patterns and impact offumigation. J. Econ. Entomol. 103: 991Ð1001.

Campbell, J. F., M. D. Toews, F. H. Arthur, and R. T. Arbo-gast. 2010b. Long-term monitoring of Tribolium casta-neum populations in two ßour mills: rebound after fumi-gation. J. Econ. Entomol. 103: 1002Ð1011.

Childs, N., and A. Burdett. 2000. The U.S. rice export mar-ket, pp. 48Ð54. In Rice Situation and Outlook. RCS-2000.U.S. Dep. Agric. Economic Research Service.

Erdman, H. E. 1964. Sexual precocity of male ßour beetleTribolium castaneum (Herbst) and inßuence of temper-atures on reproduction during early adult life. Can. En-tomol. 96: 656.

Fields, P. G., and N.D.G. White. 2002. Alternatives tomethyl bromide treatments for stored-product and quar-antine insects. Annu. Rev. Entomol. 47: 331Ð359.

Hartzer,M.,B. Subramanyam,W.Chayaprasert,D.E.Maier,S. Savodelli, J. F. Campbell, and P. W. Flinn. 2010.Methyl bromide and sulfuryl ßuoride effectivenessagainst red ßour beetle life stages, pp. 365Ð370. InM. O.Carvalho, P. G. Fields, C. S. Adler, F. H. Arthur, C. G.Athanassiou, J. F. Campbell, F. Fleurat-Lessard, P. W.Flinn, R. J. Hodges, A. A. Isikber, et al. (eds.), Proceedingsof the 10th International Working Conference on StoredProduct Protection, 27 JuneÐ2 July 2010, Estoril, Portugal.Julius Kuhn-Institut, Berlin, Germany.

Hodges, R. J., R. Robinson, and D. R. Hall. 1996. Quinonecontamination of dehusked rice by Tribolium castaneum(Herbst) (Coleoptera: Tenebrionidae). J. Stored Prod.Res. 32: 31Ð37.

Howe, R. W. 1956. The effect of temperature and humidityon the rate of development and mortality of Triboliumcastaneum (Herbst) (Coleoptera, Tenebrionidae). Ann.Appl. Biol. 44: 356Ð368.

Imura, O. 1991. A comparative study of the feeding habitsof Tribolium freemani Hinton and Tribolium castaneum

(Herbst) (Coleoptera: Tenebrionidae). Appl. Entomol.Zool. 26: 173Ð182.

McGaughey,W.H. 1970. Effect of degree of milling and ricevariety on insect development in milled rice. J. Econ.Entomol. 63: 1375Ð1376.

McGaughey,W.H. 1974. Insect development in milled rice:effectsofvariety,degreeofmilling,parboilingandbrokenkernels. J. Stored Prod. Res. 10: 81Ð86.

Mutters, R.G., and J. F. Thompson. 2009. Rice quality hand-book. Publication 3514, University of California Agricul-ture and Natural Resources, Oakland, CA.

Richardson, J. W., and J. L. Outlaw. 2010. Economic con-tributions of the US rice industry to the US economy.AFPC Research Report 10-3. Agricultural & Food PolicyCenter, Texas A&M University, College Station, TX.

Small, G. J. 2007. A comparison between the impact of sul-furyl ßuoride and methyl bromide fumigations on stored-product insect populations in UK ßour mills. J. StoredProd. Res. 43: 410Ð416.

Snyder, C. S., and N. A. Slaton. 2001. Rice production in theUnited StatesÑan overview. Better Crops Plant Food. 85:3Ð7.

Throne, J. E. 1989. Effects of noncatastrophic control tech-nologies that alter life history parameters on insect pop-ulation growth: a simulation study. Environ. Entomol. 18:1050Ð1055.

Toews, M. D., F. H. Arthur, and J. F. Campbell. 2005. Roleof food and structural complexity on capture ofTriboliumcastaneum (Herbst) (Coleoptera: Tenebrionidae) in sim-ulated warehouses. Environ. Entomol. 34: 164Ð169.

Toews, M. D., F. H. Arthur, and J. F. Campbell. 2009. Mon-itoring Tribolium castaneum (Herbst) in pilot-scale ware-houses treated with �-cyßuthrin: are residual insecticidesand trapping compatible? Bull. Entomol. Res. 99: 121Ð129.

Tsai, W.-T., L. J. Mason, W. Chayaprasert, D. E. Maier, andK. E. Ileleji. 2011. Investigation of fumigant efÞcacy inßour mills under real-world fumigation conditions. J.Stored Prod. Res. 47: 179Ð184.

Via, S. 1991. Variation between strains of the ßour beetleTribolium castaneum in relative performance on Þveßours. Entomol. Exp. Appl. 60: 173Ð182.

Received 16 July 2012; accepted 23 October 2012.


Documents

S Tribolium castaneum (Coleoptera: Tenebrionidae