explaining the sawtooth: latitudinal periodicity in a

Explaining the sawtooth: latitudinal periodicity in a circadiangene correlates with shifts in generation number

R. C. LEVY*1 , G. M. KOZAK*1 , C. B. WADSWORTH*, B. S. COATES† & E. B. DOPMAN**Department of Biology, Tufts University, Medford, MA, USA

†USDA-ARS, Corn Insects and Crop Genetics Research Unit, Genetics Laboratory, Iowa State University, Ames, IA, USA

Keywords:

circadian rhythms;

developmental timing;

diapause;

latitudinal cline;

phenology;

seasonality;

voltinism.

Abstract

Many temperate insects take advantage of longer growing seasons at lowerlatitudes by increasing their generation number or voltinism. In someinsects, development time abruptly decreases when additional generationsare fit into the season. Consequently, latitudinal ‘sawtooth’ clines associatedwith shifts in voltinism are seen for phenotypes correlated with develop-ment time, like body size. However, latitudinal variation in voltinism hasnot been linked to genetic variation at specific loci. Here, we show a patternin allele frequency among voltinism ecotypes of the European corn borermoth (Ostrinia nubilalis) that is reminiscent of a sawtooth cline. We charac-terized 145 autosomal and sex-linked SNPs and found that period, a circa-dian gene that is genetically linked to a major QTL determining variation inpost-diapause development time, shows cyclical variation between voltinismecotypes. Allele frequencies at an unlinked circadian clock gene crypto-chrome1 were correlated with period. These results suggest that selection ondevelopment time to ‘fit’ complete life cycles into a latitudinally varyinggrowing season produces oscillations in alleles associated with voltinism, pri-marily through changes at loci underlying the duration of transitionsbetween diapause and other life history phases. Correlations among clockloci suggest possible coupling between the circadian clock and the circannu-al rhythms for synchronizing seasonal life history. We anticipate that latitu-dinal oscillations in allele frequency will represent signatures of adaptationto seasonal environments in other insects and may be critical to understand-ing the ecological and evolutionary consequences of variable environments,including response to global climate change.

Introduction

Due to seasonal fluctuations in temperate climates,organisms face continuous challenges in synchronizingtheir life cycles to resource availability (Tauber & Tau-ber, 1981; Forrest & Miller-Rushing, 2010). Fitness andsurvival will be maximized when growth and reproduc-tion coincide with favourable climatic conditions andpeaks in resources such as food or available mates. Sim-ilarly, survival can be improved by synchronizing dor-mancy with periods of diminished resources (Tauber

et al., 1986). In this way, environmental fluctuationsover time result in annual rhythms in the major lifehistory phases of growth, reproduction and dormancy.Overlaid on seasonal variation in climate is geographic

variation in the length of time in which growth andreproduction can take place (season length). Seasonlength has additional consequences for life cycle syn-chronization. At high latitudes and altitudes, long win-ters result in extended periods of dormancy compared tobrief periods of growth and reproduction (Montesinos-Navarro et al., 2011). As latitude decreases, season lengthincreases and the critical environmental conditions thatinduce dormancy occur later in the year, whereas theconditions that signal dormancy termination occur ear-lier (Forrest & Miller-Rushing, 2010; Hut et al., 2013).These changes to dormancy can subsequently shift the

Correspondence: Erik B. Dopman, Department of Biology, Tufts

University, 200 Boston Ave., Ste. 4700, Medford MA 02155, USA.

Tel.: 617 627 4890; fax: 617 627 3805; e-mail: [email protected] authors

40© 20 1 4 E u r op e a n So c i e t y F o r E v o l u t i o n a r y B i o l o g y . J . E v o l . B i o l . 28 ( 2 0 1 5 ) 4 0 – 5 3

JOURNAL OF EVOLUT IONARY B IO LOGY © 2014 EUROPEAN SOC I E TY FOR EVOLUT IONARY B IO LOGY

doi: 10.1111/jeb.12562

timing of growth and reproduction and possibly also pro-duce variation in the number of generations producedper season (Tauber & Tauber, 1981; Tauber et al., 1986).Establishing the connection between climate, generationnumber and the evolution of life history phases is criticalto understanding adaptation to current climates and pre-dicting responses to future global climate change.In insects, research on the number of generations per

season (voltinism) has described how voltinism varieswith latitude and the possible connection to latitudinalvariation in dormancy. Insect dormancy (diapause) ischaracterized by low metabolic activity and develop-mental arrest, with the timing of diapause determiningthe subsequent timing of metamorphosis and adultemergence (Tauber & Tauber, 1981; Tauber et al.,1986). Changes in diapause timing have often beenlinked to biotic factors such as host plant phenology(Feder et al., 1993; Dambroski & Feder, 2007). How-ever, changes in diapause timing are also often corre-lated with latitudinal variation in abiotic factors such asphotoperiod and temperature (reviewed in Hut et al.,2013). For example, latitudinal variation exists withrespect to the critical photoperiod required to inducediapause among a wide range of insects, includingwasps (Nasonia: Paolucci et al., 2013), moths (Ostrinia:Showers, 1981; Huang et al., 2013; Chiasmia: V€alim€akiet al., 2013), butterflies (Papilio: Valella & Scriber, 2003;Sericinus: Wang et al., 2012), crickets (Pteronemobius:Masaki, 1978a) and mosquitos (Wyeomyia: Bradshawet al., 2003). The parallel clines in voltinism and dia-pause timing support diapause as a key life historyphase in life cycle synchronization.Latitudinal clines in dormancy-related traits are often

continuous. However, in many insects, they showstepped ‘sawtooth’ clines that reflect how voltinism andthe duration of dormant life stages change with increas-ing season length (Roff, 1980; Tauber & Tauber, 1981).Generally, the length of time to complete a given lifestage may gradually increase along with season lengthuntil a critical point is reached at which the seasonlength supports an additional generation. At this tippingpoint, an additional generation will be selectivelyfavoured to maximize the intrinsic rate of naturalincrease (r) of the population (Fisher, 1958), but to ‘fit’all generations into the season, selection will alsofavour decreases in the duration of one or more lifestage(s) (Roff, 1980). For example, successful comple-tion of an additional generation may require reaching adormant life stage before the onset of winter or syn-chronizing growth with host plant phenology. Underthis seasonal fitting model, abrupt shifts from long toshort developmental periods are expected to repeat lati-tudinally as subsequent critical points for additionalgenerations are encountered (Masaki, 1978b; Roff,1980). Evidence supports the existence of sawtoothclines in developmental duration and the length of thediapausing life stage (Blanckenhorn & Fairbairn, 1995),

sawtooth patterns in correlated traits such as body size(reviewed in Kivel€a et al., 2011; Shelomi, 2012) andthe contribution of temperature and host plant avail-ability to these patterns (Scriber & Lederhouse, 1992;Ayres & Scriber, 1994; Scriber, 2002). When sawtoothclines are present, we expect genes associated with lifestage duration to show cyclical changes in allele fre-quency corresponding to transitions from long to shortduration, with environmental variation producing grad-ual increases in the trait and contributing to theobserved ‘sawtooth’ phenotypic pattern. However, thisidea that periodicity in allele frequency could be relatedto shifts in voltinism has not been explicitly tested.In contrast to the sawtooth clines work, a substantial

body of research has quantified allele frequency varia-tion related to continuous changes in dormancy-relatedtraits. Most of this research has been on model systemssuch as Arabidopsis (Stinchcombe et al., 2004; Grilloet al., 2013) and Drosophila (Muona & Lumme, 1981;Schmidt & Paaby, 2008; Tyukmaeva et al., 2011). Forexample, adult Drosophila enter reproductive diapauseduring the winter and resume oogenesis in the spring.The onset of diapause varies continuously with latitude(Muona & Lumme, 1981; Schmidt & Paaby, 2008;Tyukmaeva et al., 2011) and has been correlated withlinear allele frequency changes in insulin receptors(Paaby et al., 2010) and circadian genes (includingperiod: Costa et al., 1992; timeless: Sandrelli et al., 2007;Tauber et al., 2007; and couch potato: Cogni et al., 2014).Applying a similar approach to species that vary latitu-dinally in voltinism could relate specific genetic changesto synchronization with seasonal environments and theevolution of voltinism.An ideal system for associating clinal genetic variation

directly with changes in voltinism is the European cornborer moth (Ostrinia nubilalis). The European corn borer(ECB) was introduced to North America in the early20th century with imported crops from Europe (Caffrey& Worthley, 1927). Since its introduction to the AtlanticCoast, ECB has spread west to the Rocky Mountains andsouth to the Gulf Coast and has remained a persistentpest of cultivated corn crops (Beck & Apple, 1961; Show-ers, 1981). Two different pheromone strains were intro-duced to North America. E-strain females produce andmales respond to a pheromone comprised mainly of theE-isomer of 11-tetradecenyl acetate, whereas Z-strainmoths use a blend that is primarily the Z-isomer(Kochansky et al., 1975). Differences in pheromoneblend contribute to reproductive isolation betweenstrains (Kluns, 1975; Dopman et al., 2010), as does vol-tinism, which varies within and between strains fromone (univoltine) to three or more generations (multivol-tine) (Showers, 1981; Glover et al., 1992; Dopman et al.,2010). E and Z strains coexist along the Atlantic Coast.All Midwestern populations are Z-strain, and popula-tions are univoltine in the north and increase in voltin-ism further south (Fig. 1a; Showers, 1981).

© 2014 EUROPEAN SOC I E TY FOR EVOLUT IONARY B IO LOGY . J . E VOL . B I OL . 2 8 ( 2 0 15 ) 4 0 – 53JOURNAL OF EVOLUT IONARY B IO LOGY © 2 0 14 EUROPEAN SOC I E TY FOR EVOLUT IONARY B IOLOGY

Periodicity in a circadian gene 41

Voltinism in ECB is determined in part by diapauseinduction and termination. The onset of diapause istriggered in larvae by a combination of photoperiodicand temperature cues (Beck & Apple, 1961; McLeod &Beck, 1963). Diapause termination is also photoperiodand temperature dependent and shows variation inECB (Fig. S1). A 30-day shift in post-diapause develop-mental (PDD) timing exists between bivoltine (shortPDD) and univoltine (long PDD) populations in NewYork. Variation in PDD has been mapped to a QTLcalled Pdd on the Z chromosome (Glover et al., 1992;Dopman et al., 2004, 2005; Wadsworth et al., 2013,2015). Given this known variation in the duration ofdiapause termination, we predicted that this trait mayshow a sawtooth-like pattern across latitude associatedwith shifts in voltinism (Fig. 1b).By studying genetic variation along a latitudinal

cline, where populations are known to differ in voltin-ism, we can begin to understand the genetic basis forthe evolution of seasonal timing. In this study, wesampled variation in Z chromosome and autosomalgenes in Midwestern populations of ECB along a clinein voltinism ecotypes (Fig. 1a). We predicted that lociinvolved in seasonal adaptation could show severalpatterns. First, response to continuous environmentalgradients of temperature or photoperiod could producea linear relationship between allele frequency and lati-tude. Second, given the changes in voltinism observedin North American populations of ECB, we might see

abrupt shifts in frequencies corresponding to changesin voltinism ecotype (Fig. 1b,c). Third, due to the pos-sibility that the duration of post-diapause developmentcould vary along with season length, we predicted thatany sawtooth patterns in developmental durationmight produce periodic oscillations in allele frequen-cies, such that selection for short and long develop-mental durations would produce a cyclical patternacross the latitudinal cline (Fig. 1b,d). In addition, weused pedigree mapping families to determine whetherassociations among markers of interest were the resultof physical linkage to each other or with the Pddlocus.

Materials and methods

Sample collection

Adult male moths were collected in 2005 using phero-mone traps as described in Kim et al. (2009) at ten loca-tions spaced at 80.5 km (50 mile) intervals across a724.2 km (450 mile) north–south transect centred atAmes, Iowa, USA, and extending south into Missouri(38.8°N) and north into Minnesota (45.3°N) (Fig. 1a;Table S1). A total of 24 individuals from each location(n = 240) were subsampled from those described inKim et al. (2009), corresponding to a brief time windowof 21 days during the presumed second-generationmating flight of 2005 (19 July through 9 August).

(a) (b) (c)

(d)

Fig. 1 Voltinism in European corn borer. (a) Sampling locations along a north–south transect in Minnesota, Iowa and Missouri overlaid on

a map of voltinism patterns (number of generations per year indicated in white; adapted from Showers, 1981). (b) Schematic of how

season length increases with latitude, causing increases in voltinism (indicated by dashed lines). As latitude increases, changes in dormancy

timing occur due to shifts in the critical conditions for entering (induction) and exiting (termination) dormancy. Variation in the duration

of post-diapause development (PDD) may also occur within a voltinism type with short PDD (blue) in the north and long PDD further

south (red). a (orange vertical line) indicates populations sampled in areas of univoltine–bivoltine overlap, b (yellow vertical line) indicates

populations sampled at bivoltine only sites, and c (green vertical line) indicates populations sampled in areas of bivoltine–multivoltine

overlap. (c) Alleles associated with generation number are predicted to show a disjunct pattern corresponding to voltinism types (dashed

lines). (d) Alleles associated with PDD are predicted to show a cyclical pattern corresponding to short and long PDD depicted in (b).

© 2 0 14 EUROPEAN SOC I E TY FOR EVOLUT IONARY B IOLOGY . J . E VOL . B IO L . 28 ( 2 0 1 5 ) 4 0 – 5 3JOURNAL OF EVOLUT IONARY B IO LOGY © 2014 EUROPEAN SOC I E TY FOR EVOLUT IONARY B IO LOGY

42 R. C. LEVY ET AL.

Genotyping assays

Adult moths were stored at !20 °C, and individualDNA extractions were conducted using Bio-Rad’s AquaPure isolation kit (Bio-Rad, Hercules, CA, USA) (seeKim et al., 2009). Individuals were genotyped using 194single nucleotide polymorphism (SNP) markers in sevenPCR multiplex reactions on a Sequenom MassARRAY!

(Sequenom, San Diego, CA, USA). Five of these assayswere previously designed from polymorphic sites inexpressed sequence tags (ESTs) as described in Coateset al. (2011) and included seven Z chromosome and187 autosomal markers (labelled by contig number).Individuals were also genotyped for 42 SNP loci onnewly designed Sequenom assays. A total of 42 SNPloci from 32 PCR products were incorporated into twomultiplex assays. Polymorphic SNPs were identifiedfrom aligned Sanger sequencing reads of PCR-amplifiedproducts from 46 unique primer pairs. Specifically,genomic DNA from eight individuals (two individualsfrom each of four locations 45.3°, 43.1°, 40.9°, 38.8°)was used as PCR template, and the resulting productswere sequenced at the University of Chicago Compre-hensive Cancer Center DNA Sequencing Facility.Sequence data were aligned using CLC Main Work-bench 6 (CLC Bio, Aarhus, Denmark) to identify puta-tive polymorphic sites. Sequenom assays weredeveloped for polymorphic SNPs (Sequenom AssayDesign Suite 1.0, Sequenom, San Diego, CA, USA). Ofthese markers, 29 were putatively located on the Zchromosome, 12 were in putatively diapause-associatedat autosomal genes, and one was in mitochondrial DNA(mtDNA) (Table S2; loci labelled by primer name). AllSNP assays were run at the Iowa State University Cen-ter for Plant Genomics (ISU-CPG). One additional auto-somal locus, cryptochrome1, was genotyped in allindividuals for an insertion–deletion polymorphism byagarose gel electrophoresis of differentially sized PCRproducts. Two individuals homozygous for each of thethree allele types were Sanger-sequenced to verify thatthese products were alleles of cryptochrome1.

Genetic linkage of markers

Physical linkage of markers was determined by geno-typing four ECB Z-strain pedigree families. UnivoltineZ-strain moths were derived from a stock line at theGeneva Agricultural Station and maintained at TuftsUniversity, and bivoltine Z-strain moths were collectedfrom a field site in East Aurora, NY, in 2011. Univoltinemales were crossed to bivoltine females, and F1 hybridmale progeny were then backcrossed to bivoltinefemales. DNA was isolated using DNeasy kits (Qiagen,Venlo, The Netherlands). For each family, the initialparents, backcross parents and all female offspring weregenotyped on the seven Sequenom assays at the ISU-CPG. Only female offspring were genotyped due to

expected hemizygosity at all Z-linked SNP markers,which allowed differentiation from heterozygous auto-somal markers. We genotyped four families: Family 1(67 female offspring), Family 2 (78 offspring), Family 3(66 offspring) and Family 4 (75 offspring). Data fromfamilies 1–3 were used to construct autosomal linkagegroups; data from families 1–4 were used to determineZ-linkage (Family 4 was genotyped only for putativelyZ-linked loci). Cryptochrome1 was genotyped via Sangersequencing in ten individuals per family for families 2and 4. Genetic linkage maps were constructed sepa-rately in each family using R (R Core Team, 2014) andthe R/QTL package (Broman & Sen, 2009; Arends et al.,2010). Linkage groups were formed using the estimatemap function, a minimum LOD of 3 and a maximumrecombination frequency of 0.35. Linkage groups withmore than two markers were exported for each family.Linkage groups were merged between families based onshared markers, and additional markers were added tothe consensus linkage map from the SNP-based geneticmap previously generated for O. nubilalis (Table S3 inCoates et al., 2013).Female backcross offspring were also phenotyped for

post-diapause development (PDD) time. Diapause wasinduced by rearing larvae from eggs under short-dayconditions (12:12 light:dark, 23 °C). After 35 days, di-apausing 5th instar larvae were transferred to long-dayconditions (16:8, 26 °C). Larvae were checked everyother day, and PDD was measured as the number ofdays from being placed in long-day conditions untilpupation.

Marker annotation

Marker annotations were obtained using the whole gen-ome assembly of the related silkworm, Bombyx mori(Integrated sequences, September 2008 version, accessedat KAIKObase: http://sgp.dna.affrc.go.jp/pubdata/ge-nomicsequences.html). The BLASTN algorithm was usedto determine synteny between ECB linkage groups andthe B. mori chromosomes. In addition, marker sequenceswere queried against the NCBI nonredundant proteindatabase using the BLASTX algorithm. Sequences formarkers of interest were aligned using Clustal Omega(Sievers et al., 2011).

Data analysis

Allele frequency variation with latitudeThe relationship between allele frequency and latitudewas tested for each marker using linear regression in R.To graphically compare the magnitude of the relation-ship between latitude and allele frequency acrossmarkers, the estimated slope from the linear regres-sion for each marker was used to calculate a ‘standard-ized’ allele frequency at three points (!1, 0, 1). Allslopes were plotted so they were increasing. Pearson



product-moment correlation tests were used to comparepatterns of variation between markers with significantrelationships with latitude. Plots were made using the R

packages GGPLOT2 (Wickham, 2009) and CORRPLOT (Wei,2012). False discovery rate correction for multipletesting (q-value) was calculated in the FDRTOOL package(Strimmer, 2008).

Allele frequency variation with voltinismAt least three distinct voltinism ecotypes have beenidentified in Midwestern ECB that produce one, two orthree generations per season (Showers et al., 1975).Populations from the one to two generation region maycontain a mixture of the one and two generation eco-types or may be a distinct ecotype (producing a secondgeneration in some years but not others: Showers,1981), but for simplicity, we refer to this as the 1–2generation ecotype. Therefore, grouping populations byvoltinism resulted in three ecotypes: (i) one to two gen-erations (the two northernmost sites: 44.6° and 45.3°N),(ii) two generations only (the central six sites: 40.2–43.9°) and (iii) two to three generations (the twosouthernmost sites: 38.8° and 39.5°). Genetic differenti-ation between ecotypes was measured by calculatingFST’ among voltinism groups using the Weir and Cock-erham correction (Weir & Cockerham, 1984) in the R

package HIERFSTAT (Goudet, 2005). Loci with greaterthan expected differentiation were detected in twoways. First, the significance of differentiation betweenvoltinism ecotypes was assessed for each locus by Fish-er’s allelic goodness-of-fit test in HierFstat (Goudetet al., 1996; Goudet, 2005). P-values associated withthese G statistics were determined using 200 permuta-tions of random assignment of the data into alternategroups. Second, an explicit FST outlier test was per-formed using Fdist as implemented in the JAVA programLOSITAN (Beaumont & Nichols, 1996; Antao et al.,2008). Data were converted from Fstat format using theprogram PGDSPIDER (Lischer & Excoffier, 2012). In Losit-an, neutral FST was estimated from the data set in aninitial run, and then 50 000 simulations were per-formed to identify outliers. Outliers for directionalselection were loci in which the locus-specific FST wasgreater than the population-level FST in 97.5% or moreof the simulations.

Allele frequency variation with populationTotal and pairwise FST’ were estimated for all ten popu-lations for all markers in HierFstat. Isolation-by-distance(IBD) models assume that genetic differentiation ispositively related to the geographic distance betweenpopulations (Wright, 1943). Evidence for a linear rela-tionship between geographic distance and pairwise pop-ulation FST’ estimates was tested using a Mantel test inthe program IBDWS version 3.23 (Jensen et al., 2005).Mantel tests were calculated using (i) pairwise FST’ forall markers and (ii) pairwise FST’ calculated using only

the 131 markers without a significant linear relation-ship between latitude and allele frequency. In addition,outlier analyses were performed among all 10 popula-tions in HierFstat and Fdist (similar to those performedamong voltinism ecotypes).

Cyclical patterns in allele frequencyIt was hypothesized that some loci might show cyclicalor periodic patterns associated with differences in devel-opmental duration that would not be easily detected ina linear regression of allele frequencies on latitude(Fig. 1d). Periodicity in allele frequency was testedusing the Jonckheere-Terpstra-Kendall algorithm, anonparametric test of periodicity as implemented in theR package JTK CYCLE (Hughes et al., 2010; Deckard et al.,2013). Our data set contained 10 data points (popula-tions) and JTK cycle tested for cyclic patterns that com-pleted one full cycle between 5 and 9 data points,estimating the cycle and amplitude and then calculatinga Bonferroni-corrected P-value for periodicity.

Results

Single nucleotide polymorphism (SNP) genotyping

Samples were genotyped for one mtDNA, 38 Z-linkedand 155 autosomal SNPs using Sequenom assays. All194 markers produced base calls for at least one sample.The mean number of individuals successfully genotypedwas 210.8 " 42.9 per SNP, with a median of 225 indi-viduals. After quality filtering of SNPs with low poly-morphism (see Table S3), 145 SNP markers (Z-linked:35 markers from 21 loci, autosomal: 109 markers from101 loci, mtDNA: 1 locus) and one autosomal polymor-phic indel locus were retained for statistical analyses.

Genetic linkage of molecular markers

We generated 30 linkage groups with more than threemarkers using the combined data from our pedigreefamilies (mean markers per linkage group = 9.6,median = 5, range = 3–35). Correspondence of ourgroups to B. mori chromosomes is listed in Table S4. Allputative Z-linked markers, with one exception (537),showed evidence for sex linkage in our pedigrees basedon lack of heterozygous genotypes for hemizygousfemales. Sanger sequence data of cryptochrome1 in fami-lies 2 and 4 suggested that the marker was autosomaldue to observed female heterozygotes, but this markercould not be placed in a linkage group.

Allele frequency variation with latitude

Eleven of the 110 polymorphic autosomal markers(10.0%) and four of the 35 Z-linked markers (11.4%)showed significant correlations between allele fre-quency and latitude (Table S5; P < 0.05; FDR corrected



0.133 ≤ q ≤ 0.497). The magnitudes of the standardizedslopes between allele frequency and latitude are com-pared in Fig. S2a; marker-specific slopes are shown inFig. S3. Cryptochrome1 (cry1) showed three indel alleles,and the major allele showed a significant correlationbetween allele frequency and latitude (r = !0.75,P = 0.012, q = 0.249). Other annotated markers withsignificant correlations with latitude included four Z-linked markers: Lactate-dehydrogenase (Ldh), Shaker (Shk)and Glutamine synthetase (680 and contig00968).These 15 latitude-associated markers were responsi-

ble for the overall genomewide pattern of isolation bydistance among our populations (Fig. S2b); when theywere removed from the data set, no significant rela-tionship was detected between pairwise FST’ and geo-graphic distance (Fig. S2c). Allele frequency wassignificantly correlated among some markers (Fig. S4).The four Z-linked markers that showed an associationwith latitude (Ldh, Shk, 680, contig00968) were corre-lated with one another, with contig07183 and withcontig01517. Cry1 major allele frequency was correlatedwith period (per), contig01517 and contig00845. Thefirst minor allele of cry1 was also correlated with per.Many of these correlations also had an FDR correctedq < 0.05 (Fig. S4).

Allele frequency variation with voltinism

HierFstat analysis of genetic differentiation between thethree voltinism ecotypes estimated a relatively low glo-bal FST’ = 0.0015 (95% CI: !0.001 to 0.0033). Eightmarkers, including cry1, contig05909 (CG32521) andper, had significant genetic differentiation based onFisher’s G statistic (Fig. 2; Table 1; P < 0.05, FDR cor-rected 0.238 ≤ q ≤ 0.677). Using Fdist, cry1(FST = 0.0547) and contig05909 (FST = 0.116) wereidentified as FST outliers that may be influenced bydirectional selection.

Allele frequency variation with population

The genetic differentiation among all 10 populationswas also low with an estimated global FST’ = 0.0022(95% CI: !0.0002 to 0.0046). HierFstat results identi-fied eleven markers that showed significant differentia-tion among populations (Table 1), of which only two(cry1 and contig06459) were also among the HierFstatvoltinism outliers. Five markers were identified as FSToutliers among populations in Fdist: 2076, contig05909,contig06459, contig00382 and alpha-amylase (amy).Only contig05909 was also present among the voltinismoutliers.

Cyclical patterns in allele frequency

In addition to showing genetic differentiation amongvoltinism types, per also showed a significant cyclical pat-

tern (Fig. 2c; Table 1; Bonferroni-corrected P = 0.033,cycle = 5, amplitude = 0.12). Per showed an oscillatingpattern with peaks in allele frequency at 39° and 42°

latitude and troughs at 41° and 45°. The pattern of varia-tion in per was correlated with cry1, and the first minorallele of cry1 also showed a cyclical pattern (Table 1;Fig. 2a; Bonferroni-corrected P = 0.023, cycle = 6,amplitude = 0.05). Three other markers also had signifi-cant cyclical patterns: contig05725 (Eukaryotic translationinitiation factor 3 subunit H), contig00280 (Heat shockprotein hsp21.4) and contig05979 (probable cysteinedesulfurase).

Period and post-diapause development time

Our primers amplified at 519-bp portion of period con-taining both coding and noncoding sequence. A poly-morphic G/T SNP at position 80 in the sequence wasgenotyped in all individuals via our Sequenom assay.This SNP corresponds to an amino acid change fromthe hydrophobic alanine to the negatively chargedaspartic acid (Fig. 3a). In our latitudinal cline, the Tallele was at high frequency in the far south(38.8–39.5°, Fig. 2c) and at low frequency in the north(44.6–45.3°) where the G allele was at higher fre-quency. Pedigree families 1, 2 and 4 were polymorphicfor this SNP. In these three families, females whoinherited the G allele from the univoltine grandparenthad long post-diapause development time, whereasfemales who inherited the T allele from the bivoltinegrandparent had short development time (Fig. 3b; Fam-ily 1: G allele PDD time = 40.10 " 1.85 days, T allelePDD time = 27 " 1.01, two-tailed unequal variancet44 = 6.23, P < 0.0001; Family 2: G PDDtime = 45.87 " 1.55, T PDD time = 35.47 " 1.19,t70 = 5.29, P < 0.0001; Family 4: G PDDtime = 38.42 " 1.45, T PDD time = 29.95 " 1.15,t56 = 4.59, P < 0.0001).

Discussion

Our results suggest that photoperiod is an importantabiotic factor shaping latitudinal variation in voltinismin the European corn borer moth. Many of the outlierloci between voltinism ecotypes were related to thecircadian clock pathway (HierFstat: 4 of 8 outliers; Fdistanalyses: 2 of 2 outliers). The circadian pathway senseschanges from short- to long-day photoperiods throughlight-dependent patterns of gene expression (Stoleruet al., 2007; Meuti & Denlinger, 2013). Allelic variationin the circadian genes cryptochrome1 and period wereassociated with clinal differences in voltinism. Crypto-chrome1 allele frequency showed a significant correla-tion with latitude, was an FST outlier locus between ourthree voltinism ecotypes and showed evidence of acyclical pattern. Period showed variation betweenthe voltinism ecotypes as well as variation within the



bivoltine ecotype, producing a significant cyclical pat-tern (Fig. 2c). Contig05909 was another FST outlierlocus and was predicted to be homologous to Drosophilaprotein CG32521, a downstream target of the retinaldevelopment transcription factor eyeless (Ostrin et al.,2006). Contig01517 showed a strong association withlatitude and cry1 and encodes a gamma subunit ofguanine nucleotide-binding protein (G-protein). Aswith the beta subunit (RACK1), the gamma subunitmay function in photoreception and the regulation ofcircadian genes like cry1 (Hamasaka et al., 2005; Wang& Montell, 2007; Robles et al., 2010). Therefore, thecircadian pathway appears to be important for evolu-tionary changes in seasonal timing.Cryptochrome is a blue light photoreceptor that con-

tributes to the regulation of circadian rhythms in plants

and animals (Cashmore et al., 1999; Stoleru et al.,2007). In Drosophila, light stimulates CRYPTOCHROMEto degrade the protein TIMELESS (TIM), preventingdimers from forming between TIM and PERIOD (PER),and thus activating the expression of other circadiangenes (Stanewsky et al., 1998; Cashmore et al., 1999).Drosophila have a single cryptochrome, but Lepidopterahave two gene copies (cry1 and cry2) through geneduplication (Zhu et al., 2005). The ECB cryptochromegene sequence in this study showed greater homologyto the Lepidopteran CRY1 protein sequence (BLASTXfor CRY1: score = 81, e-value < 1 9 10!15; CRY2:score = 51.52, e-value < 3 9 10!10). Cry1 is ortholo-gous to Drosophila cryptochrome. In contrast, cry2 is simi-lar to the mammalian protein (cry-m), which bindsdirectly to PER to function as a transcriptional repressor

(a)

(b)

(c) Fig. 2 Outlier loci among voltinism

ecotypes. Allele frequency, latitude and

voltinism ecotype (indicated by vertical

dashed lines) plotted for (a)

cryptochrome1 voltinism outlier

(HierFstat and Fdist) and cyclical outlier

which had three alleles: major allele

(solid line; significant linear pattern),

minor allele 1 (dashed line; significant

cyclical pattern) and minor allele 2

(dotted line); (b) contig5909

(c05909.680/CG32521) voltinism

outlier (HierFstat and Fdist); (c) period

voltinism outlier (HierFstat) and cyclical

outlier.



of the circadian clock (Zhu et al., 2008; Meuti & Den-linger, 2013). In D. melanogaster, variation in crypto-chrome has no known association with latitude; ratherpolymorphism appears to be maintained by balancingselection in Europe (Pegoraro et al., 2014). Cry1 hasbeen linked to eclosion time in D. melanogaster (Pegor-aro et al., 2014) and reproductive diapause in D. triau-raria (Yamada & Yamamoto, 2011). In bean bugs(Riptortus pedestris), cry2 is related to pupal diapauseincidence (Ikeno et al., 2011). Our results are the firstto specifically link polymorphism in cryptochrome1 toinsect voltinism.We found oscillations in period allele frequency across

latitude consistent with spatially varying selection,which is notably different from the linear allele fre-quency clines in period that have been described in Dro-sophila (Sawyer et al., 2006; Kyriacou et al., 2008). Inother insects, phenotypic sawtooth clines have beenextensively documented (reviewed in Kivel€a et al.,2011; Shelomi, 2012), but our current study is the firstto find a cyclical pattern in allele frequency correspond-ing to changes in voltinism. Variation in period has pre-viously been associated with temperature sensitivity ofreproductive diapause in Drosophila (Huang et al., 1995)and pupal diapause incidence in flesh flies (Han & Den-linger, 2009) and linden bugs (Dole"zel et al., 2007).

This oscillating pattern in period is consistent with ourexpectation for an allele underlying a sawtooth pheno-typic cline. Although further sampling is needed to ver-ify that the oscillating pattern in genotype correspondsdirectly to a sawtooth cline in diapause phenotype, ourresults are concordant with earlier work that found aperiodic pattern in diapause induction across thesesame latitudes (Beck & Apple, 1961). Under controlledlaboratory conditions (photoperiod = 14.5 hours light,temperature = 26 °C), ECB from Wisconsin populations(latitude = 43°) had higher diapause incidence(DI = 90%) than populations from Iowa (lati-tude = 42°, DI = 80%). Diapause incidence increasedfurther south in Kansas populations (latitude = 39°,DI = 96%) before decreasing again in Missouri (lati-tude = 37°, DI = 45%; Fig S5). Based on this previouswork, we propose that the mechanism generating thiscyclical pattern may be an association between periodand some genetic factor associated with diapause and/or the duration of the dormant life stage.In ECB, period, the QTL controlling variation in dia-

pause termination time and the QTL for diapauseinduction all map to the Z chromosome (Glover et al.,1992; Dopman et al., 2004, 2005; Ikten et al., 2011;Wadsworth et al., 2015). Therefore, period may be phys-ically linked to one or more of these QTL. The QTL for

Table 1 Summary of loci that show significant differentiation between voltinism ecotypes or between populations.

Marker

Linkage

group ECB

Location

B. mori

Voltinism groups (3) Populations (10)Cyclical

HierFstatFdist

HierFstatFdist JTK cycle

FST’

G stat

P-value Percentile FST’

G stat

P-value Percentile

Bonferroni

P-value

cry1 NA Chr17 0.0507 0.005 0.989377 0.0166 0.025 0.95635 0.02282

contig01517.490 15 Chr23 0.0353 0.005 0.940448

shp_A 5 NA 0.0401 0.005 0.96851

contig05909.680 18 Chr17 0.1055 0.01 0.999471 0.0431 0.105 0.999916

contig06459.185 22 Chr19 0.0357 0.02 0.958401 0.0277 0.045 0.988713

contig06616.192 5 Chr5 0.0317 0.03 0.941477

per_A Z Z chr 0.0277 0.035 0.8695 0.03312

contig05822.227 2 NA 0.0092 0.045 0.743966 0.0242 0.04 0.939272

2076_C Z Z chr 0.0523 0.005 0.999722

contig00382.237 14 NA 0.0348 0.015 0.994137

contig01874.600 21 Chr3 0.0171 0.03 0.818727

HR3_B NA Chr2 0.019 0.03 0.852556

contig06879.246 5 Chr5 0.0292 0.035 0.970767

622_C Z Z chr 0.013 0.04 0.506357

contig03796.413 NA Chr3 0.0191 0.045 0.898642

cp146_B NA Chr22 0.0237 0.045 0.925301

amy_B NA Chr4 0.0224 0.065 0.985336

contig05725.351 2 Chr4 0.00403

contig05979.160 16 Chr28 0.02282

contig00280.181 9 Chr13 0.03312

Results from HierFstat (FST’ and P-value of G statistic) and Fdist (percentile of simulations in which the simulated FST < locus-specific FST)

shown. Significant outliers highlighted in bold (P < 0.05 for HierFstat, percentile > 0.975 for Fdist). The last column shows the Bonferroni

-corrected P-value for loci with significant cyclic patterns (determined by JTK cycle).



PDD differences between bivoltine E-strain and univol-tine Z-strain ECB (Pdd) is linked to the Tpi locus, withone allele corresponding to a short pupation time of15.3 days and the other allele leading to a long pupa-tion time of 43.7 days (Glover et al., 1992; Dopmanet al., 2005). Recent work found that period co-segre-gates with Pdd and Tpi in these populations (Wads-worth et al., 2015). In the univoltine and bivoltine Z-strain pedigree families in our study, there was also astrong association between period and PDD time. OneSNP allele was associated with long PDD time and theother with short PDD (Fig. 3b). Interestingly, the shortPDD allele we identified is predicted to cause an aminoacid change (from alanine to the negatively chargedaspartic acid) two residues away from the estimatedstart of the Per-Arnt-Sim-B (PAS-B) domain and down-stream of the first TIMELESS interacting site (TIS-1) inthe PERIOD sequence of the silkworm, B. mori (Iwaiet al., 2006). Ostrinia furnacalis, the Asian corn borer,also has alanine in this position of the protein (Regieret al., 2012). The PAS-B domain contributes to the sta-bility of PER-TIM heterodimers in Drosophila (Henniget al., 2009) and PER-PER homodimers in mammals(Kucera et al., 2012), suggesting that a change in thisregion may affect these protein–protein interactions.If period is associated with Pdd, the oscillating pattern

we observed could be explained by spatially varyingselection on the timing of diapause emergence. In thenorthernmost sites (44.6° and 45.3°), we found the long

Pdd allele at higher frequency. This may because atthese latitudes, the long allele is prevalent in univoltinepopulations, whereas the short Pdd allele could be pre-valent in bivoltine populations to accommodate thedevelopment of two complete generations in a singleshort growing season (indicated by a in Fig. 1b). If oursample contained a majority of univoltine individuals,it would explain the high frequency of the long alleleat these northern sites. Consistent with this idea, sam-ples from northerly bivoltine populations between 42and 44° latitude also had the short Pdd allele at highfrequency. At southerly bivoltine locations in Iowa(40–42°), we again found the long Pdd allele at higherfrequency. This could be due to the fact that growingseasons are longer but insufficient to accommodate athird generation, which may lead to increases in thefrequency of the long Pdd allele (Fig. 1b-b). At latitudesfurther south, a third generation can be produced andshort Pdd alleles might allow for fitting of all three gen-erations into a season (Fig. 1b-c). These data, togetherwith the oscillating pattern Beck & Apple (1961) foundin diapause incidence (Fig. S5), suggest that seasonalfitting produces oscillations in alleles associated withdiapause in ECB, primarily through changes in theduration of transitions between diapause and other lifehistory phases. Work in tiger swallowtail butterflies(Papilio canadensis and P. glaucus) has also found differ-ences in voltinism which are associated with a Z-linkedfactor and that period shows genetic differentiation

(a)

(b)

Fig. 3 Period alleles and PDD in ECB.

(a) Alignment of amino acid sequence

of PERIOD using silkworm (Bombyx

mori), European corn borer

transcriptome (ECB Transcript:

Wadsworth & Dopman in preparation),

Asian corn borer (ACB, Ostrinia

furnacalis; Regier et al., 2012), European

corn borer G allele coding sequence and

ECB T allele. The amino acid change

between ECB alleles caused by the

nucleotide change from G to T is noted

by an arrow. TIMELESS interacting site

1 (TIS-1) and protein interacting Per-

Arnt-Sim-B (PAS-B) domain in the

B. mori sequence denoted by rectangles

(Iwai et al., 2006). (b) Distribution of

post-diapause development (PDD) time

(in days) in female backcross offspring

in mapping Family 1 based on their

genotype at the period SNP; open bars

represent the G allele, and filled bars,

the T allele.



between these Papilio species (Putnam et al., 2007; Ku-nte et al., 2011), suggesting period could be associatedwith voltinism in distantly related Lepidoptera.Colonization of North America by ECB likely

included bouts of rapid adaptation to seasonal environ-ments. Moths from this study were collected ~ 60 yearsafter their introduction to the Midwest in the 1940s(Beck & Apple, 1961; Palmer et al., 1985). It is possiblethat certain neutral processes could cause clinal pat-terns at some loci by chance. One such process mightbe rapid population expansion, which can cause differ-entiation via gene surfing along the expansion front(Excoffier & Ray, 2008). Another possibility is thatadmixture of populations from different locations inEurope during colonization of North America led tononrandom assortment of alleles (Barton & Hewitt,1985). Different allelic variants in North America likelyarose through admixture between colonizing popula-tions from Italy and Hungary (Caffrey & Worthley,1927), but admixture is unlikely to provide an explana-tion for the specific clinal patterns we observed. Underneutral processes, we would not expect loci that showclinal and oscillating patterns to be involved in any spe-cific pathways. Loci associated with circadian rhythmsand diapause loci represented 7.5% of our molecularmarkers, with the other 92.5% being anonymousZ-linked or autosomal loci. However, circadian and dia-pause-associated markers represented 14% of markersshowing linear associations with latitude, 37.5–50% ofFST outliers between voltinism groups and 40% of thecyclic outliers. More importantly, period allele frequencyshowed a strong oscillating pattern with latitude, andwe found that this locus co-segregated with post-dia-pause termination time phenotype in crosses betweenunivoltine and bivoltine ECB populations. Such anassociation between the genotype of a cyclic allele anda diapause-associated phenotype is unlikely to be dueto chance. The magnitude of allele frequency variationin cryptochrome1 (0.29) and period (0.25) over our 6.5o

latitudinal cline is comparable to the variation observedin period (Thr-Gly)20 alleles in Australian Drosophila(allele frequency variation of 0.4 over an 25o cline) thathas evolved via natural selection in the past 200 years(Sawyer et al., 2006; Kyriacou et al., 2008). Thus, itseems likely that latitudinal patterns in circadian geneallele frequency in ECB are driven by adaptation to sea-sonal environments. This association could be con-firmed by sampling replicate latitudinal transects acrossNorth America and Europe in future studies.Our results suggest that differences in voltinism in

ECB are associated with changes in the circadian clockpathway. Allele frequency changes for cryptochrome1and period were strongly correlated with each other,more strongly than period allele frequencies were corre-lated with other markers known to be physically linkedto period on the Z chromosome (such as Ldh; Fig. S4).Rather than being physically linked in ECB, cry1 and

period could show correlations as a consequence of epi-static selection, possibly because of their interactionswith timeless, a gene showing clinal variation in Dro-sophila (Tauber et al., 2007). It is unclear why multiplecircadian genes show correlated changes associated withvoltinism in ECB and which gene(s) in the pathwaymight be the primary target(s) of natural selection.Thus, the precise contribution of cry1, period, timeless andother circadian genes to circannual rhythms, patterns ofvoltinism and seasonal timing warrant further study. Wemust also confirm tentative associations between periodand diapause emergence timing and/or induction pheno-types. Changes in phenology in many plants and animalsinvolve changes in both dormancy induction and termi-nation (Forrest & Miller-Rushing, 2010). Therefore,understanding how these two seasonal timing pheno-types evolve in concert to enable adaptation to variableand rapidly changing environments represents a majorquestion in biological research.

Conclusions

The extremely rapid evolution of allelic variationobserved in North American ECB strongly suggests thatnatural selection is driving changes between and withinvoltinism ecotypes in order to fit additional generationsinto a season. We propose that selective pressures onseasonal fitting generate sawtooth phenotypic clines inthe duration of life stages and associated oscillations inallele frequency in developmental timing genes. Saw-tooth clines have been the subject of over 30 years ofresearch (Masaki, 1978a,b; Roff, 1980), and we antici-pate that latitudinal patterns of cyclic genetic variationmay be key signatures of seasonal timing traits and rep-resent sawtooth clines at the genetic level. Contempo-rary increases in temperature are driving local increasesin season length and generation number in Lepidoptera(Breed et al., 2013; Scriber, 2014; Scriber et al., 2014),suggesting that these same genes may show cyclicalpatterns across decades. Thus, scouting for such pat-terns in other taxa will allow us to identify the genesassociated with seasonal environments, understandhow populations adapt to climatic variation throughouttheir geographic ranges and predict how populationswill respond to changes in climate.

Acknowledgments

We thank Mitzi Wilkening and the other staff at theIowa State University Genomic Technologies Facility forperforming the Sequenom Assay. Roxana Khozein andXinnan Li assisted with primer design and Sequenomassay marker development; Charles Linn providedinsect stocks; Thomas Sappington and Kyung Seok Kimprovided field collected samples; Shoshanna Kahneassisted with figures and pedigree family genotyping;and Gabriel Golczer provided help with R scripts. Tho-



mas Flatt, Mark Scriber and two anonymous reviewersprovided helpful feedback on the manuscript. Thisstudy was funded by National Science Foundation(EBD: DEB-1257251) and United States Department ofAgriculture (EBD: 2010-65106-20610). A portion ofthis research was also supported by the United StatesDepartment of Agriculture, Agricultural Research Ser-vice (BSC: CRIS Project 3625-22000-017-00) and theIowa Agriculture and Home Economics ExperimentStation, Ames, IA (BSC: Project 3543).

Data accessibility

Allele frequency data: Dryad doi:10.5061/dryad.cg2gg.http://datadryad.org/review?wfID=32515&token=55a79920-202a-4654-b96d-75cf2971e971

References

Antao, T., Lopes, A., Lopes, R.J., Beja-Pereira, A. & Luikart, G.

2008. Lositan: a workbench to detect molecular adaptation

based on a Fst-outlier method. BMC Bioinformatics 9: 323.Arends, D., Prins, P., Jansen, R.C. & Broman, K.W. 2010. R/QTL: high-throughput multiple QTL mapping. Bioinformatics

26: 2990–2992.Ayres, M.P. & Scriber, J.M. 1994. Local adaptations to regionalclimates in Papilio canadensis (Lepidoptera: Papilionidae). Ecol.

Monogr. 64: 465–482.Barton, N.H. & Hewitt, G.M. 1985. Analysis of hybrid zones.

Annu. Rev. Ecol. System. 16: 113–148.Beaumont, M.A. & Nichols, R.A. 1996. Evaluating loci for use

in the genetic analysis of population structure. Proc. R. Soc. B

263: 1619–1626.Beck, S.D. & Apple, J.W. 1961. Effects of temperature andphotoperiod on voltinism of geographical populations of the

European corn borer, Pyrausta nubilalis. J. Econ. Entomol. 54:550–558.

Blanckenhorn, W.U. & Fairbairn, D.J. 1995. Life history adap-

tation along a latitudinal cline in the water strider Aquarius

remigis (Heteroptera: Gerridae). J. Evol. Biol. 8: 21–41.Bradshaw, W.E., Quebodeaux, M.C. & Holzapfel, C.M. 2003.Circadian rhythmicity and photoperiodism in the pitcher-

plant mosquito: adaptive response to the photic environ-

ment or correlated response to the seasonal environment?

Am. Nat. 161: 735–748.Breed, G.A., Stichter, S. & Crone, E.E. 2013. Climate-driven

changes in northeastern US butterfly communities. Nat. Clim.

Chang. 3: 142–145.Broman, K.W. & Sen, S. 2009. A Guide to QTL Mapping withR/QTL. Springer, New York.

Caffrey, D.J. & Worthley, L.H. 1927. A progress report on the

investigations of the European corn borer. U.S. Department ofAgriculture Department Bulletin 1476.

Cashmore, A.R., Jarillo, J.A., Wu, Y.-J. & Liu, D. 1999. Crypto-

chromes: blue light receptors for plants and animals. Science

284: 760–765.Coates, B.S., Bayles, D.O., Wanner, K.W., Robertson, H.M.,

Hellmich, R.L. & Sappington, T.W. 2011. The application

and performance of single nucleotide polymorphism markers

for population genetic analyses of Lepidoptera. Front. Genet.2: 38.

Coates, B.S., Sumerford, D.V., Siegfried, B.D., Hellmich, R.L. &

Abel, C.A. 2013. Unlinked genetic loci control the reduced

transcription of aminopeptidase-n-1 and 3 in the Europeancorn borer and determine tolerance to Bacillus thuringiensis

Cry1Ab toxin. Insect Biochem. Mol. Biol. 43: 1152–1160.Cogni, R., Kuczynski, C., Koury, S., Lavington, E., Behrman,E.L., O’Brien, K.R., et al. 2014. The intensity of selection act-

ing on the couch potato gene: spatial-temporal variation in a

diapause cline. Evolution 68: 538–548.Costa, R., Peixoto, A.A., Barbujani, G. & Kyriacou, C.P. 1992.A latitudinal cline in a Drosophila clock gene. Proc. R. Soc. B

250: 43–49.Dambroski, H.R. & Feder, J.L. 2007. Host plant and lati-

tude-related diapause variation in Rhagoletis pomonella: a testfor multifaceted life history adaptation on different stages of

diapause development. J. Evol. Biol. 20: 2101–2112.Deckard, A., Anafi, R.C., Hogenesch, J.B., Haase, S.B. & Harer,J. 2013. Design and analysis of large-scale biological rhythm

studies: a comparison of algorithms for detecting periodic

signals in biological data. Bioinformatics 29: 3174–3180.Dole"zel, D., "Sauman, I., Ko"st’#al, V. & Hodkova, M. 2007.Photoperiodic and food signals control expression pattern of

the clock gene, period, in the linden bug, Pyrrhocoris apterus.

J. Biol. Rhythms 22: 335–342.Dopman, E.B., Bogdanowicz, S.M. & Harrison, R.G. 2004.Genetic mapping of sexual isolation between E and Z phero-

mone strains of the European corn borer (Ostrinia nubilalis).

Genetics 167: 301–309.Dopman, E.B., Perez, L., Bogdanowicz, S.M. & Harrison, R.G.

2005. Consequences of reproductive barriers for genealogical

discordance in the European corn borer. Proc. Natl. Acad. Sci.

USA 102: 14706–14711.Dopman, E.B., Robbins, P.S. & Seaman, A. 2010. Components

of reproductive isolation between North American phero-

mone strains of the European corn borer. Evolution 64: 881–902.

Excoffier, L. & Ray, N. 2008. Surfing during population expan-

sions promotes genetic revolutions and structuration. Trends

Ecol. Evol. 23: 347–351.Feder, J.L., Hunt, T.A. & Bush, L. 1993. The effects of climate,host plant phenology and host fidelity on the genetics of

apple and hawthorn infesting races of Rhagoletis pomonella.

Entomol. Exp. Appl. 69: 117–135.Fisher, R.A. 1958. The Genetical Theory of Natural Selection.

Dover Publications, New York.

Forrest, J. & Miller-Rushing, A.J. 2010. Toward a synthetic

understanding of the role of phenology in ecology and evo-lution. Phil. Trans. R. Soc. B 365: 3101–3112.

Glover, T.J., Robbins, P.S., Eckenrode, C.J. & Roelofs, W.L.

1992. Genetic control of voltinism characteristics in Euro-

pean corn borer races assessed with a marker gene. Arch.Insect Biochem. Physiol. 20: 107–117.

Goudet, J. 2005. HIERFSTAT, a package for R to compute and

test hierarchical F-statistics. Mol. Ecol. Notes 5: 184–186.Goudet, J., Raymond, M., de Mee€us, T. & Rousset, F. 1996.

Testing differentiation in diploid populations. Genetics 144:1933–1940.

Grillo, M.A., Li, C., Hammond, M., Wang, L. & Schemske,D.W. 2013. Genetic architecture of flowering time differenti-



ation between locally adapted populations of Arabidopsis tha-liana. New Phytol. 197: 1321–1331.

Hamasaka, Y., Wegener, C. & N€assel, D.R. 2005. GABA modu-

lates Drosophila circadian clock neurons via GABAb receptors

and decreases in calcium. J. Neurobiol. 65: 225–240.Han, B. & Denlinger, D.L. 2009. Length variation in a specific

region of the period gene correlates with differences in pupal

diapause incidence in the flesh fly, Sarcophaga bullata.J. Insect Physiol. 55: 415–418.

Hennig, S., Strauss, H.M., Vanselow, K., Yildiz, €O., Schulze, S.,

Arens, J., et al. 2009. Structural and functional analyses of

PAS domain interactions of the clock proteins DrosophilaPERIOD and mouse PERIOD2. PLoS Biol. 7: e1000094.

Huang, Z.J., Curtin, K.D. & Rosbash, M. 1995. PER protein

interactions and temperature compensation of a circadian

clock in Drosophila. Science 267: 1169–1172.Huang, L.-L., Chen, C., Xiao, L., Xia, Q.-W., Hu, L.-T. & Xue,

F. 2013. Geographical variation and inheritance of the

photoperiodic response controlling larval diapause in twodistinct voltine ecotypes of the Asian corn borer Ostrinia fur-

nacalis. Physiol. Entomol. 38: 126–132.Hughes, M.E., Hogenesch, J.B. & Kornacker, K. 2010.

JTK_CYCLE: an efficient nonparametric algorithm for detect-ing rhythmic components in genome-scale data sets. J. Biol.

Rhythms 25: 372–380.Hut, R.A., Paolucci, S., Dor, R., Kyriacou, C.P. & Daan, S.

2013. Latitudinal clines: an evolutionary view on biologicalrhythms. Proc. R. Soc. B 280: 20130433

Ikeno, T., Numata, H. & Goto, S.G. 2011. Photoperiodic

response requires mammalian-type cryptochrome in the beanbug Riptortus pedestris. Biochem. Biophys. Res. Commun. 410:394–397.

Ikten, C., Skoda, S.R., Hunt, T.E., Molina-Ochoa, J. & Foster, J.E.

2011. Genetic variation and inheritance of diapause inductionin two distinct voltine ecotypes of Ostrinia nubilalis (Lepidop-

tera: Crambidae). Ann. Entomol. Soc. Am. 104: 567–575.Iwai, S., Fukui, Y., Fujiwara, Y. & Takeda, M. 2006. Structure

and expressions of two circadian clock genes, period and time-less in the commercial silkmoth, Bombyx mori. J. Insect Physiol.

52: 625–637.Jensen, J.L., Bohonak, A.J. & Kelley, S.T. 2005. Isolation by

distance, web service. BMC Genet. 6: 13.Kim, K.S., Bagley, M.J., Coates, B.S., Hellmich, R.L. & Sapp-

ington, T.W. 2009. Spatial and temporal genetic analyses

show high gene flow among European corn borer (Lepidop-tera: Crambidae) populations across the central US corn belt.

Environ. Entomol. 38: 1312–1323.Kivel€a, S.M., V€alim€aki, P., Carrasco, D., M€aenp€a€a, M.I. & Ok-

sanen, J. 2011. Latitudinal insect body size clines revisited: acritical evaluation of the saw-tooth model. J. Anim. Ecol. 80:1184–1195.

Kluns, J. 1975. Insect sex pheromones: intraspecific phero-

monal variability of Ostrinia nubilalis in North America andEurope. Environ. Entomol. 4: 891–894.

Kochansky, J., Card#e, R., Liebherr, J. & Roelofs, W. 1975. Sex

pheromone of the European corn borer, Ostrinia nubilalis(Lepidoptera: Pyralidae), in New York. J. Chem. Ecol. 1: 225–231.

Kucera, N., Schmalen, I., Hennig, S., €Ollinger, R., Strauss,

H.M., Grudziecki, A., et al. 2012. Unwinding the differencesof the mammalian PERIOD clock proteins from crystal struc-

ture to cellular function. Proc. Natl. Acad. Sci. 109: 3311–3316.

Kunte, K., Shea, C., Aardema, M.L., Scriber, J.M., Juenger,

T.E., Gilbert, L.E., et al. 2011. Sex chromosome mosaicism

and hybrid speciation among tiger swallowtail butterflies.PLoS Genet. 7: e1002274.

Kyriacou, C.P., Peixoto, A.A., Sandrelli, F., Costa, R. & Tauber,

E. 2008. Clines in clock genes: fine-tuning circadian rhythmsto the environment. Trends Genet. 24: 124–132.

Lischer, H. & Excoffier, L. 2012. PGDSpider: an automated data

conversion tool for connecting population genetics and

genomics programs. Bioinformatics 28: 298–299.Masaki, S. 1978a. Climatic adaptation and species status in the

lawn ground cricket. Oecologia 35: 343–356.Masaki, S. 1978b. Seasonal and latitudinal adaptations in the

life cycles of crickets. In: Evolution of Insect Migration andDiapause (H. Dingle, ed.), pp. 72–100, Proceedings in life

sciences. Springer, US.

McLeod, D. & Beck, S.D. 1963. Photoperiodic termination ofdiapause in an insect. Biol. Bull. 124: 84–96.

Meuti, M.E. & Denlinger, D.L. 2013. Evolutionary links

between circadian clocks and photoperiodic diapause in

insects. Integr. Comp. Biol. 53: 131–143.Montesinos-Navarro, A., Wig, J., Xavier Pico, F. & Tonsor, S.J.

2011. Arabidopsis thaliana populations show clinal variation

in a climatic gradient associated with altitude. New Phytol.

189: 282–294.Muona, O. & Lumme, J. 1981. Geographical variation in the

reproductive cycle and photoperiodic diapause of Drosophila

phalerata and D. transversa (Drosophilidae: Diptera). Evolution35: 158–167.

Ostrin, E.J., Li, Y., Hoffman, K., Liu, J., Wang, K., Zhang, L.,

et al. 2006. Genome-wide identification of direct targets of

the Drosophila retinal determination protein eyeless. GenomeRes. 16: 466–476.

Paaby, A.B., Blacket, M.J., Hoffmann, A.A. & Schmidt, P.S.

2010. Identification of a candidate adaptive polymorphism

for Drosophila life history by parallel independent clines ontwo continents. Mol. Ecol. 19: 760–774.

Palmer, D.F., Schenk, T. & Chiang, H.-C. 1985. Dispersal and

voltinism adaptation of the European corn borer in North

America, 1917–1977.Paolucci, S., van de Zande, L. & Beukeboom, L.W. 2013. Adap-

tive latitudinal cline of photoperiodic diapause induction in

the parasitoid Nasonia vitripennis in Europe. J. Evol. Biol. 26:705–718.

Pegoraro, M., Noreen, S., Bhutani, S., Tsolou, A., Schmid, R.,

Kyriacou, C.P., et al. 2014. Molecular evolution of a perva-

sive natural amino-acid substitution in Drosophila crypto-chrome. PLoS ONE 9: e86483.

Putnam, A.S., Scriber, J.M. & Andolfatto, P. 2007. Discordant

divergence times among Z-chromosome regions between

two ecologically distinct swallowtail butterfly species. Evolu-tion 61: 912–927.

R Core Team 2014. R: A language and environment for statis-

tical computing. R Foundation for Statistical Computing,Vienna, Austria. URL http://www.R-project.org/

Regier, J.C., Mitter, C., Solis, M., Hayden, J.E., Landry, B.,

Nuss, M., et al. 2012. A molecular phylogeny for the pyra-

loid moths (Lepidoptera: Pyraloidea) and its implications forhigher-level classification. Syst. Entomol. 37: 635–656.



Robles, M.S., Boyault, C., Knutti, D., Padmanabhan, K. &Weitz, C.J. 2010. Identification of RACK1 and protein kinase

ca as integral components of the mammalian circadian clock.

Science 327: 463–466.Roff, D. 1980. Optimizing development time in a seasonalenvironment: the ‘ups and downs’ of clinal variation. Oecolo-

gia 45: 202–208.Sandrelli, F., Tauber, E., Pegoraro, M., Mazzotta, G., Cisotto,P., Landskron, J., et al. 2007. A molecular basis for natural

selection at the timeless locus in Drosophila melanogaster. Sci-

ence 316: 1898–1900.Sawyer, L.A., Sandrelli, F., Pasetto, C., Peixoto, A.A., Rosato,E., Costa, R., et al. 2006. The period gene Thr-Gly polymor-

phism in Australian and African Drosophila melanogaster pop-

ulations: implications for selection. Genetics 174: 465–480.Schmidt, P.S. & Paaby, A.B. 2008. Reproductive diapause andlife-history clines in North American populations of Drosoph-

ila melanogaster. Evolution 62: 1204–1215.Scriber, J.M. 2002. Latitudinal and local geographic mosaics inhost plant preferences as shaped by thermal units and vol-

tinism in Papilio spp. (Lepidoptera). Eur. J. Entomol. 99: 225–239.

Scriber, J.M. 2014. Review: climate-driven reshuffling of spe-cies and genes; potential conservation roles for species trans-

locations and recombinant hybrid genotypes. Open Access

Insects 5: 1–61.Scriber, J.M. & Lederhouse, R.C. 1992. The thermal environ-ment as a resource dictating geographic patterns of feeding

specialization of insect herbivores. In: Effects of Resource Distri-

bution on Animal-Plant Interactions (M.R. Hunter, T. Ohgushi& P.W. Price, eds), pp. 429–466. Academic Press, San Diego,

California.

Scriber, J.M., Eliot, B., Maher, E., McGuire, M. & Niblack, M.

2014. Adaptations to ‘thermal time’ constraints in Papilio:Latitudinal and local size clines differ in response to regional

climate change. Open Access Insects 5: 199–226. doi:10.3390/insects5010199.

Shelomi, M. 2012. Where are we now? Bergmann’s rule sensulato in insects. Am. Nat. 180: 511–519.

Showers, W. 1981. Geographic variation of the diapause

response in the European corn borer. In: Insect Life History

Patterns (R. Denno & H. Dingle, eds), pp. 97–111, Proceed-ings in life sciences. Springer, New York.

Showers, W., Chiang, H., Keaster, A., Hill, R., Reed, G.,

Sparks, A., et al. 1975. Ecotypes of the European corn borerin North America. Environ. Entomol. 4: 753–760.

Sievers, F., Wilm, A., Dineen, D., Gibson, T.J., Karplus, K., Li,

W., et al. 2011. Fast, scalable generation of high-quality pro-

tein multiple sequence alignments using Clustal omega. Mol.Syst. Biol. 7: 539.

Stanewsky, R., Kaneko, M., Emery, P., Beretta, B.,

Wager-Smith, K., Kay, S.A., et al. 1998. The cryb mutation

identifies cryptochrome as a circadian photoreceptor inDrosophila. Cell 95: 681–692.

Stinchcombe, J.R., Weinig, C., Ungerer, M., Olsen, K.M.,

Mays, C., Halldorsdottir, S.S., et al. 2004. A latitudinal clinein flowering time in Arabidopsis thaliana modulated by the

flowering time gene FRIGIDA. Proc. Natl. Acad. Sci. USA 101:4712–4717.

Stoleru, D., Nawathean, P., Fern#andez, M.D.L.P., Menet, J.S.,Ceriani, M.F. & Rosbash, M. 2007. The Drosophila circadian

network is a seasonal timer. Cell 129: 207–219.

Strimmer, K. 2008. Fdrtool: a versatile R package for estimat-ing local and tail area-based false discovery rates. Bioinfor-

matics 24: 1461–1462.Tauber, C.A. & Tauber, M.J. 1981. Insect seasonal cycles:

genetics and evolution. Annu. Rev. Ecol. Syst. 12: 281–308.Tauber, M.J., Tauber, C.A. & Masaki, S. 1986. Seasonal Adapta-

tions of Insects. Oxford University Press, New York.

Tauber, E., Zordan, M., Sandrelli, F., Pegoraro, M., Osterwal-der, N., Breda, C., et al. 2007. Natural selection favors a

newly derived timeless allele in Drosophila melanogaster. Science

316: 1895–1898.Tyukmaeva, V.I., Salminen, T.S., Kankare, M., Knott, K.E. &Hoikkala, A. 2011. Adaptation to a seasonally varying envi-

ronment: a strong latitudinal cline in reproductive diapause

combined with high gene flow in Drosophila montana. Ecol.

Evol. 1: 160–168.Valella, P. & Scriber, M.J. 2003. Latitudinal variation in photo-

periodic induction of pupal diapause in the Spicebush Swal-

lowtail Butterfly, Papilio troilus (Lepidoptera: Papilionidae).Holarctic Lepidoptera 10: 37–41.

V€alim€aki, P., Kivel€a, S. & M€aenp€a€a, M. 2013. Temperature-

and density-dependence of diapause induction and its life

history correlates in the geometrid moth Chiasmia clathrata(Lepidoptera: Geometridae). Evol. Ecol. 27: 1217–1233.

Wadsworth, C.B., Woods, W.A., Hahn, D.A. & Dopman, E.B.

2013. One phase of the dormancy developmental pathway is

critical for the evolution of insect seasonality. J. Evol. Biol.26: 2359–2368.

Wadsworth, C.B., Li, X. & Dopman, E.B. 2015. A recombina-

tion suppressor contributes to ecological speciation in Ostri-nia moths. Heredity, Accepted.

Wang, T. & Montell, C. 2007. Phototransduction and retinal

degeneration in Drosophila. Pfl€ug. Arch. Eur. J. Physiol. 454:821–847.

Wang, X.-P., Yang, Q.-S., Dalin, P., Zhou, X.-M., Luo, Z.-W. &

Lei, C.-L. 2012. Geographic variation in photoperiodic dia-

pause induction and diapause intensity in Sericinus montelus

(Lepidoptera: Papilionidae). Insect Sci. 19: 295–302.Wei, T. 2012. Package ‘corrplot’ – Visualization of a correlation

matrix.

Weir, B.S. & Cockerham, C.C. 1984. Estimating F-statistics for

the analysis of population structure. Evolution 38: 1358–1370.

Wickham, H. 2009. ggplot2: Elegant Graphics for Data Analysis.

Springer, New York.Wright, S. 1943. Isolation by distance. Genetics 28: 114.Yamada, H. & Yamamoto, M.-T. 2011. Association between

circadian clock genes and diapause incidence in Drosophila

triauraria. PLoS ONE 6: e27493.Zhu, H., Yuan, Q., Froy, O., Casselman, A. & Reppert, S.M.

2005. The two CRYs of the butterfly. Curr. Biol. 15: R953–R954.

Zhu, H.S., Sauman, I., Yuan, Q., Casselman, A., Emery-Le, M.,Emery, P., et al. 2008. Cryptochromes define a novel circadian

clock mechanism in monarch butterflies that may underlie

sun compass navigation. PLoS Biol. 6: 138–155.

Supporting information

Additional Supporting Information may be found in theonline version of this article:



Figure S1 Life cycle and timing of diapause of bivoltineand univoltine European corn borers.Figure S2 Linear associations with latitude.Figure S3 Locus-specific associations with latitude.Figure S4 Correlation matrix among markers showinga relationship with latitude.Figure S5 Diapause incidence in ECB.Table S1 Sampling locations, listed from north to south,with corresponding latitude (as noted in Figure 1), num-ber of European corn borer generations per season, andmean degree days for the 2005 and 2006 seasons.

Table S2 Sequenom assay markers.Table S3 Summary of Sequenom assay results.Table S4 ECB linkage groups from pedigree families.Table S5 Summary of correlation and linear model statis-tics for all markers with statistically significant correla-tions between allele frequency and latitude.

Data deposited at Dryad: doi:10.5061/dryad.cg2gg

Received 29 August 2014; revised 21 November 2014; accepted 24

November 2014



explaining the sawtooth: latitudinal periodicity in a

Documents