GENETICS
Molecular Identification of Ceratitis capitata (Diptera: Tephritidae) Using DNA Sequences of the COI Barcode Region N. B. BARR,1,2 M. S. ISLAM,3 M. DE MEYER,4
AND
B. A. MCPHERON3
Ann. Entomol. Soc. Am. 105(2): 339Ð350 (2012); DOI: http://dx.doi.org/10.1603/AN11100
ABSTRACT The utility of the cytochrome oxidase I gene barcode region for diagnosis of the Mediterranean fruit ßy, Ceratitis capitata (Weidemann), is evaluated using African fruit ßy collections. The method fails to discern C. capitata from its close relative Ceratitis caetrata Munro, based on genetic distances, parsimony networks, or nucleotide diagnostic characters observed in the DNA barcode sequences. When treated as a single taxon, it is possible to discern the C. capitata ⫹ C. caetrata lineage from other Ceratitis species. Levels of intraspeciÞc diversity vary within the genus Ceratitis and multiple copies of the mitochondrial gene are reported for Ceratitis cosyra (Walker). The DNA barcoding method based on genetic distance is compared with a molecular identiÞcation method using restriction fragment length polymorphism. The DNA barcode and restriction fragment-length polymorphism methods provide similar identiÞcation results, but the DNA sequence information is more suitable for quantitative analysis of the information. KEY WORDS DNA barcode, restriction fragment-length polymorphism, diagnostic, Tephritidae
The fruit ßy genus Ceratitis MacLeay (Diptera: Tephritidae) comprises 89 described species that are native to Africa (De Meyer 2000a, De Meyer and Copeland 2005, Barr and McPheron 2006). The genus includes several highly polyphagous species that are of economic importance (White and Elson-Harris 1992, Yuval and Hendrichs 2000, Copeland et al. 2006) and recognized as invasive or potentially invasive (De Meyer et al. 2008). Of these species, Ceratitis capitata (Wiedemann), commonly called the Mediterranean fruit ßy, is regarded as the most serious international pest because of its broad range of hosts, nearly worldwide distribution, and impact on trade (White and Elson-Harris 1992, Copeland et al. 2002, Vera et al. 2002, De Meyer et al. 2008, Barr 2009). Accurate identiÞcation of C. capitata is important for implementing proper quarantine and pest management practices. Consequently, morphological tools have been developed to diagnose C. capitata and other Ceratitis species using adult characters (De Meyer 1996, 1998, 2000b; De Meyer and Freidberg 2005a). Unfortunately, at ports of entry, intercepted larvae cannot be reliably identiÞed to the species-level by using morphology. White and Elson-Harris (1992) developed keys for third-instar larvae, but these keys include only 11 species and are based on small sample sizes for most of the species. 1 Center for Plant Health Science and Technology, Mission Laboratory, USDA-APHIS, Moore Air Base, Edinburg, TX 78541. 2 Corresponding author, e-mail:
[email protected]. 3 Department of Entomology, Pennsylvania State University, University Park, PA 16802. 4 Royal Museum for Central Africa, Leuvensesteenweg 13, 3080 Tervuren, Belgium.
Although numerous studies using molecular methods to identify C. capitata have been published, most include insufÞcient sampling of species to demonstrate taxonomic speciÞcity in the techniques. For example, Douglas and Haymer (2001) and KakouliDuarte et al. (2001) each reported a technique to discern C. capitata from C. rosa Karsch, and Armstrong and Ball (2005) a technique to discern C. capitata from C. rosa Karsch and C. cosyra (Walker), but did not test additional Ceratitis species. Some studies only compare C. capitata to fruit ßies from other genera (Armstrong et al. 1997, Huang et al. 2009). Despite the limited taxon sampling, these tools can be of value in fruit ßy identiÞcation when nonmolecular information can be integrated in the diagnosis process. To date, the most taxonomically comprehensive molecular tool published for Ceratitis species identiÞcation is a Polymerase Chain ReactionÐRestriction Fragment Length Polymorphism (polymerase chain reaction [PCR]-restriction fragment-length polymorphism [RFLP]) method by Barr et al. (2006). This tool is based on relatively good, although not complete, sampling of the genus and includes collections from its ancestral home range in Africa (De Meyer et al. 2004). These African collections provide a better estimate of genetic variation within C. capitata than do collections derived from introduced populations (Barr 2009). In addition, the data set includes samples of Ceratitis caetrata Munro and Ceratitis pinax Munro, two close relatives of C. capitata that are useful for evaluating the speciÞcity of the diagnostic tool (De Meyer 2005, Barr and McPheron 2006, Barr and Wiegmann 2009). The PCR-RFLP method, developed using the 12S rRNA, 16S rRNA, and NADH-dehydrogenase subunit
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
340
ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
6 mitochondrial DNA loci, was not able to distinguish C. capitata from C. caetrata but it could distinguish C. pinax. However, differences in the internal transcribed spacer region 1 (ITS-1) of C. capitata and C. caetrata were able to discriminate the two species based on DNA sequences and PCR amplicon length (Barr et al. 2006). Many of the DNA samples used in the PCR-RFLP study also were used to estimate the phylogeny of the genus (Barr and McPheron 2006, Barr and Wiegmann 2009). Although DNA sequences were generated to estimate phylogenies and locate restriction enzyme recognition sites in the restriction fragment-length polymorphism diagnostic tool, the ability to use the DNA sequences as diagnostic information has yet to be evaluated. Recent studies have demonstrated that a DNA sequencing approach to identiÞcation of C. capitata geographic populations was superior to a PCR-RFLP approach when using the same mitochondrial gene region (Lanzavecchia et al. 2008, Barr 2009). There is growing interest in the use of DNA barcodes to diagnose pest arthropod species (Armstrong 2010, Floyd et al. 2010). The technology is being investigated as a tool for identiÞcation of species that have economic, ecological, and health impacts (Ball and Armstrong 2006, Scheffer et al. 2006, Yancy et al. 2008, Lowenstein et al. 2009, deWaard et al. 2010, Naro-Maciel et al. 2010). The Þrst published application of DNA barcoding for tephritid fruit ßy identiÞcation focused on the genus Bactrocera Macquart (Armstrong and Ball 2005). Currently, DNA barcoding of tephritid fruit ßies is an objective of international barcode campaigns such as the Quarantine Barcoding of Life (QBOL, www.qbol.org/UK/) and the Tephritid Barcode Initiative (TBI, http://www. barcoding.si.edu/major_projects.html). Although the number of public DNA barcode records is increasing for Ceratitis species (e.g., 317 specimens with barcode records representing 45 taxa were reported on the Barcode of Life Database, BOLD, on 3 May 2011), no publications have evaluated the utility of these sequences. We sequence the proposed barcode region (Hebert et al. 2003) of the cytochrome oxidase subunit 1 gene (COI) using the Ceratitis specimens included in the Barr et al. (2006) restriction fragment-length polymorphism study. These collections represent expertly identiÞed material collected in Africa. We use these COI data to 1) test whether the COI barcode region can reliably identify C. capitata, and 2) compare the diagnostic utility of the DNA sequence method to the published PCR-RFLP methodology for diagnoses within the genus Ceratitis. Materials and Methods Samples. All specimens were from collections made on the African mainland or Re´ union. Collection information per species is provided in Table 1. Collection information per specimen is provided in Supp. Table S1. The majority of samples were generated as
Vol. 105, no. 2
part of a study of Kenyan fauna (De Meyer et al. 2002, Copeland et al. 2006). With the exception of four C. capitata specimens (codes 620 Ð 623, Kenyan samples 1382 by R. Copeland), DNA isolates were reported previously in the Barr et al. (2006) and Barr and McPheron (2006) studies. In addition to Ceratitis, species representing the genera Trirhithrum Bezzi, Capparimyia Bezzi, Carpophthoromyia Austen, and Notomma Bezzi are included for comparison to the restriction fragment-length polymorphism study. De Meyer and Freidberg (2005b) have shown that the Capparimyia melanspis Bezzi collection included in Barr et al. (2006) study includes two species: C. melanaspis and Capparimyia aenigma De Meyer and Freidberg. These samples are treated as a mixture of species and reported as Capparimyia sp. in the current study. Over 600 samples were available for analysis, but only a subset of 249 samples was selected for DNA barcoding to minimize research costs. For each species included in the Barr et al. (2006) study, at least Þve samples were selected for barcode analysis (Table 1). Specimens were selected based on collection location and restriction fragment-length polymorphism genotypes to maximize geographic and genetic variation within each species. In addition, several species represented by single specimens in the Barr and McPheron (2006) phylogenetic study also were included to increase taxonomic coverage in the data base. Every specimen is numerically coded based on position in the database (Supp. Table S1). All material was identiÞed to species based on adult morphology (Supp. Table S1). Pinned vouchers (of entire adult body) representative of each Kenyan collection series are maintained at the Royal Museum for Central Africa (KMMA, Tervuren, Belgium), the Frost Entomological Museum at the Pennsylvania State University (PSUC, University Park, PA), or both. For most DNA barcoded samples (Table 1), a tissue voucher is maintained in ethanol at the APHISCPHST lab in Edinburg, TX. The tissue vouchers include wings and abdomens from each specimen (the head, thoraces, and legs were used in the DNA isolation process). A description of DNA isolation procedures is reported in Barr et al. (2006). PCR and DNA Sequencing. The COI gene was ampliÞed from all DNA isolates using the LCO-1490 (5⬘GGTCAACAAATCATAAAGATATTGG-3⬘) and HCO2198 (5⬘-TAAACTTCAGGGTGACCAAAAAATCA-3⬘) primers reported by Folmer et al. (1994). Samples that failed to generate usable PCR product or DNA sequences were re-analyzed using the LCO and HCO primers. A portion of the DNA isolates that failed to generate good results using the LCO and HCO primer pair after two trials were then retested using the primer pair TY-J-1460 (5⬘-TACAATTTATCGCCTAAACTTCAGCC-3⬘) and C1-N-2191 (5⬘-CCCGGTAAAATTAAAATATAAACTTC-3⬘) as reported by Simon et al. (1994). PCRs were performed using the Applied Biosystems Gene Amp PCR System 9700 (Applied Biosystems, Foster City, CA).
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
March 2012 Table 1.
BARR ET AL.: DNA BARCODING THE MEDITERRANEAN FRUIT FLY
341
General Barcode Information for Taxa included in study
Taxon
NPOP
NIND
NSEQ
NHAP
NVOU
Country (NIND)
GenBank Accessions
C. anonae C. argenteobrunnea C. bremii C. caetrata C. capitata
3 1 1 2 23
6 5 1 27 40
6 5 1 25 34
3 4 1 16 27
6 5 1 25 22
JN705083ÐJN705088 JN705177ÐJN705181 JN705240 JN705043ÐJN70568 JN705009ÐJN705042
C. colae C. contramedia C. copelandi C. cornuta C. cosyra C. cristata C. curvata C. ditissima
4 1 2 1 4 1 1 4
8 5 5 5 11 10 5 5
8 5 4 5 10 10 5 5
1 1 3 1 7 7 4 3
8 5 4 5 10 10 5 5
C. divaricata C. fasciventris C. flexuosa C. gravinotata C. marriotti C. millicentae C. oraria C. penicillata C. pennitibialis C. perseus C. pinax C. podocarpi C. querita C. rosa
1 4 2 1 4 1 1 1 1 4 1 2 1 7
6 5 7 1 7 5 6 1 1 5 5 5 5 7
6 5 7 1 7 5 6 1 1 5 5 5 5 7
3 4 5 1 2 4 6 1 1 3 3 2 4 4
6 5 7 1 7 5 6 1 1 5 5 5 5 7
C. rubivora C. simi C. stictica C. stipula C. venusta Ceratitis nsp. 1195 Carpophthoromyia dimidiata Capparimyia sp. Notomma sp. (1000) T. coffeae T. culcasiae T. demeyeri T. meladiscum T. nigerrimum T. occipitale T. senex T. teres Total
3 1 1 1 1 1 1
5 1 5 1 6 5 5
5 1 5 1 6 5 5
4 1 3 1 1 1 5
5 1 5 1 6 5 5
Kenya (6) Kenya (5) Kenya (1) Kenya Kenya (25), Malawi (2), Reunion (2), Ghana (5) Ghana (8) Kenya (5) Kenya (5) Kenya (5) Kenya (6), Mali (4) Kenya (10) Kenya (5) Kenya (3), Ghana (1), Nigeria (1) Kenya (6) Kenya (5) Kenya (7) Kenya (1) Kenya (7) Kenya (5) Kenya (6) Nigeria (1) Kenya (1) Kenya (5) Kenya (5) Kenya (5) Kenya (5) Kenya(5), Malawi (1), Reunion (1) Kenya (5) Kenya (1) Kenya (5) Kenya (1) Kenya (6) Kenya (5) Kenya (5)
1 1 1 1 1 1 2 1 1 1 98
5 5 1 1 1 1 5 1 1 1 249
5 5 1 1 1 1 5 1 1 1 239
2 2 1 1 1 1 3 1 1 1 151
5 5 1 1 1 1 5 1 1 1 227
Kenya (5) Kenya (5) Kenya (1) Kenya (1) Kenya (1) Kenya (1) Kenya (5) Kenya (1) Kenya (1) Kenya (1)
JN705137ÐJN705144 JN705167ÐJN705171 JN705111ÐJN705114 JN705235ÐJN705239 JN705157ÐJN705166 JN705195ÐJN705204 JN705127ÐJN705131 JN705074ÐJN705077 JN705151ÐJN705156 JN705089ÐJN705093 JN705115ÐJN705121 JN705231 JN705182ÐJN705188 JN705078ÐJN705082 JN705145ÐJN705150 JN705233 JN705232 JN705101ÐJN705105 JN705069ÐJN705073 JN705122ÐJN705126 JN705132ÐJN705136 JN705094ÐJN705100 JN705106ÐJN705110 JN705234 JN705172ÐJN705176 JN705230 JN705189ÐJN705194 JN705225ÐJN705229 JN705215ÐJN705219 JN705210ÐJN705214 JN705220ÐJN705224 JN705242 JN705241 JN705247 JN705243 JN705205ÐJN705209 JN705246 JN705244 JN705245
Sample sizes given for number of populations sampled (NPOP), individuals sampled (NIND), sequences generated (NSEQ), haplotypes identiÞed (NHAP), and tissue vouchers saved (NVOU).
PCR at the APHIS lab was performed in 50-l reaction volumes by using 1 l of DNA template or water (as a negative control). Each reaction had a Þnal concentration of 1X buffer, 2.5 mM MgCl2, 0.2 mM of each dNTP, 0.2 M of each primer, and 0.625U of Ex-Taq polymerase (Takara). Cycling conditions were 3 min at 94⬚C followed by 39 cycles of 94⬚C (20 s)/50⬚C (20 s)/72⬚C (30 s) and an extension of 5 min at 72⬚C. PCR product was visualized on 1% agarose gels and samples were puriÞed using QIAquick puriÞcation columns (Qiagen, Valencia, CA). DNA sequences were generated for each primer (i.e., sense and anti-sense strands) using an ABI 3730XL DNA analyzer at Davis Sequencing (Davis, CA), a Beckman-Coulter CEQ8000 at the APHIS Mission Lab (Edinburg, TX), or both.
At PSU, PCR was performed in 25-l reaction volumes using 1 l of DNA template or water as a negative control. Each reaction had a Þnal concentration of 1X buffer, 1.5 mM MgCl2, 0.2 mM of each dNTP, 0.3 M of each primer, and 0.625 U of Qiagen Taq polymerase. Cycling conditions were 3 min at 94⬚C followed by 39 cycles of 94⬚C (1 min)/50⬚C (1 min)/72⬚C (1 min) and an extension of 10 min at 72⬚C. PCR product was visualized on 1.5% agarose gels and samples were puriÞed using ExoSAP-IT (USB Corp.). DNA sequences were generated for each primer (i.e., sense and anti-sense strands) using an ABI 3730XL DNA analyzer at the Huck InstituteÕs Nucleic Acid Facility at Penn State University (University Park, PA).
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
342
ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
The TA cloning kit (Invitrogen, Carlsbad, CA) was used to clone amplicons of the COI barcode region. This cloning was done for samples that failed to generate reliable sequences using direct sequencing. Transformed colonies were grown and selected on LB-agar plates with kanamycin and ampicillin. Plasmid DNA was puriÞed from transformed colonies using a miniprep kit (Qiagen) and sequenced using universal M13 primers at the Penn State Genomics Core Facility (University Park, PA). DNA Editing and Analysis. All DNA sequence trace Þles were edited using Sequencher v.4.10 (Gene Codes, Ann Arbor, MI). For each specimen, bidirectional sequences were assembled into a contig and used to corroborate base calls and identify conßicting signal. Sequences with evidence of polymorphism (dual peaks at a site in the trace Þles) in the mitochondrial COI marker were included in the analysis if the frequency of polymorphism was low (⬍1%). Specimens that failed to generate usable sequences in both directions were removed from the analysis. Consensus sequences from Sequencher were imported into MEGA4 (Kumar et al. 2008) and aligned by hand. Sequences that generated evidence of pseudogene copies, based on the presence of reading frame mutations that resulted in premature stop codons, were removed from analysis. The edited consensus Þles were submitted to GenBank with associated trace Þles. Unique haplotypes in the data set were identiÞed by Þrst generating a haplotype data Þle by using DnaSP v5.10 (Librado and Rozas 2009). These haplotypes then were conÞrmed or further collapsed by comparing pair-wise p-distances in MEGA. The program TCS 1.21 (Clement et al. 2000) was used to generate statistical parsimony networks of the haplotypes. The default connection limit of 95% was applied to all analyses. Intra- and inter-speciÞc variation values were estimated for each species by using MEGA. Intra-speciÞc variation was estimated as a range using minimum and maximum genetic distances separating any two barcodes within a species and as the mean genetic distance among barcodes of conspeciÞc specimens. To obtain inter-speciÞc estimates, all conspeciÞc pairwise comparisons were excluded from estimates of mean, minimum, and maximum p-distance values. For example, the p-distances between a C. pinax barcode sequence and each of the 234 non-C.pinax barcodes in the alignment were calculated. This was repeated for the other four C. pinax barcodes in the database. The 1,170 p-distance values were then used to identify the mean, minimum, and maximum inter-speciÞc variation for C. pinax. An uncorrected p-distance estimate based on the observed proportion of differences between sequences was selected for estimating variation in the COI data set instead of the more commonly used Kimura 2-Parameter (K2P) model. Based on previous research the p-distance estimate is an appropriate measure for analysis of closely related species (Nei and Kumar 2000, Srivathsan and Meier 2011). The minimum inter-speciÞc distance value was used to identify the “nearest neighboring species” in the data
Vol. 105, no. 2
set and the “nearest neighboring specimens” (i.e., most similar sequences) between the two neighboring species. The average number of nucleotide substitutions per site between populations (Dxy) was calculated between each pair of nearest neighboring species using DnaSP. The minimum and maximum intra-speciÞc and inter-speciÞc values were graphed for each taxon. The separation between the maximum intraspeciÞc and minimum inter-speciÞc values is the barcode gap (Meyer and Paulay 2005). The diagnostic utility of the COI barcode was assessed for C. capitata by evaluating the barcode gap, inter-speciÞc connections in the haplotype network, and possible character-based diagnostic sites using the nucleotide diagnostic (ND) process as described by Wong et al. (2009). The genetic variation across the COI gene region for the entire data set was estimated in a sliding window of 100 bases with one base steps using DnaSP. The effect of sequence length on the utility of a distancebased identiÞcation was tested for C. capitata by performing sliding window analyses. To measure the effect of window size on the barcode gap between C. capitata and its nearest neighboring species, Dxy values were calculated between populations using window sizes of 10, 50, 100, 200, 300, 400, and 500 bases and a constant step size of one base. These minimum and maximum values for intra-speciÞc and inter-speciÞc variation were plotted against window size. The Dxy values estimate the mean variation between populations but not between two specimens. As a result, the sliding window using Dxy values could fail to detect loss of a barcode gap between two speciÞc sequences. Therefore, the sliding window analysis experiment was repeated using p-distance () estimates between the nearest neighboring specimens from different species (i.e., minimum inter-speciÞc values). For comparison, intra-speciÞc variation for each species was estimated using the most dissimilar specimens within a species (i.e., maximum intra-speciÞc variation). The resulting barcode database was tested against 27 unique COI sequences reported previously by Barr (2009) for C. capitata. These datasets were compared using minimum genetic distances, character differences between haplotypes, and connectivity of haplotypes in a statistical parsimony network. The Barr (2009) sequences were shorter (467 bp) than the sequences generated for this study (Table 1). Because of this difference in length and the fact that the 2009 records were generated from specimens lacking any series or tissue vouchers, these sequences were not combined in the general analysis of diagnostic utility. Results General PCR and Sequencing Success. All 249 samples selected for barcode analysis generated visible PCR product of the expected size (c. 650bp). The majority of these PCR products generated usable DNA sequences. Six of the C. capitata samples (011, 022, 023, 024, 025, and 027), however, generated se-
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
March 2012
BARR ET AL.: DNA BARCODING THE MEDITERRANEAN FRUIT FLY
quences with evidence of intra-individual polymorphism at many (ⱖ5%) sites. In addition, poor quality DNA sequences were generated from two C. caetrata samples (047 and 056), one Ceratitis copelandi De Meyer and Freidberg sample, and one Ceratitis cosyra sample (417). Eight Ceratitis cristata Bezzi samples generated poor sequences (507, 508, 511, 514, 516, 517, 518, and 519) using the Folmer primers. One Capparimyia sp. sample (551) generated ⬎1% polymorphic sites (8/603). The C. capitata, C. caetrata, and C. copelandi samples with evidence of high polymorphism levels and poor quality sequence were removed from the analysis. The Capparimyia sp. sample was included in the analysis but will not meet the GenBank criteria as an ofÞcial barcode record according to NCBI standards. The C. cristata and C. cosyra samples were re-ampliÞed and sequenced using the Simon et al. (1994) COI primer pair (see Materials and Methods). This primer substitution generated good quality barcode sequences for the C. cristata samples but not for the C. cosyra sample. The C. cosyra PCR product that failed to generate good sequence data was cloned using the TA-cloning kit (Invitrogen). From that cloning event, four transformed colonies were selected for DNA sequencing. Three unique genotypes were observed in the four clones (JN715791Ð 4). One genotype (JN715793) had two reading frame mutations. The other two genotypes (JN715791 and JN715792) had amino acid sequences similar to other C. cosyra sequences (2Ð3 amino acid substitutions), but p-distances distinct from the other con-speciÞc barcodes (⬎11%). These sequences were most similar to Ceratitis fruit ßy DNA using NCBI BLAST searchers. Based on these results the specimen was not included in further barcode analyses of the species. The Þnal data set included 239 barcode sequences representing 44 species and resulted in a 603-bp alignment containing no gaps. There were a few examples of single amino acid substitutions within species (i.e., C. capitata: I/T; C. perseus De Meyer and Copeland: A/T; C. oraria De Meyer and Copeland: T/A; C. stictica Bezzi: M/V), but no evidence of reading frame mutations. C. capitata Identification. The alignment included 34 C. capitata barcode sequences of which 27 represented unique haplotypes. Based on the statistical parsimony network, the C. capitata haplotypes are connected in a network with 16 C. caetrata haplotypes (Fig. 1). No haplotypes were shared between the two species. However, the positioning of haplotypes in the “C. capitata⫹C. caetrata” (CAP-CAET) network does not suggest a clear separation of gene pools for the two species. There were no diagnostic character states in the alignment that could distinguish C. capitata from C. caetrata. Wong et al. (2009) deÞne these characters as nucleotide diagnostics (ND). The two species lacked both simple ND (sND) and compound ND (cND) characters. The cND search identiÞed eleven variable sites (60, 99, 259, 342, 462, 486, 508, 537, 543, 579, and
343
Fig. 1. Network of CAP-CAET. The white circles represent C. capitata haplotypes and gray circles represent C. caetrata haplotypes.
582; Supp. Table S2) that were insufÞcient to diagnose the species because of shared characters between species. Consistent with the character-based diagnostics, the C. capitata and C. caetrata sequences lacked a “barcode gap” using genetic distances (Fig. 2). The minimum genetic separation between the two species was 0.17%, and the maximum variation within either species was ⱖ1.5%. In comparison, a barcode gap was observed for the majority of tested Ceratitis species (Fig. 2). For example, a strong barcode gap was observed for Ceratitis pinax, a close relative of C. capitata and C. caetrata. Fig. 3 displays barcode gap information after redeÞning taxonomic units in the data set for some of the species. Recalculation of genetic distances when using a taxon, called “CAP-CAET,” that collapses the species C. capitata and C. caetrata resulted in a barcode gap (Fig. 3). The maximum genetic variation in the CAPCAET taxon was 2% and the minimum separation between the taxon and its nearest neighbor, C. pinax, was 6.7% (Supp. Fig. S3). Separation of the taxa CAPCAET and C. pinax was also supported by 32 Þxed differences and two unconnected networks. A sliding window analysis of genetic distance (Dxy) between the two taxa, CAP-CAET and C. pinax, demonstrates that divergence values can vary across the COI gene barcode region (Fig. 4). The trend for increasing variation toward the 3⬘ end is consistent with genetic diversity estimates () for the entire COI data set (Fig. 4). Additional sliding window analyses show that a window size of 200 bases should generate a barcode gap between the CAP-CAET and C. pinax taxa (Fig. 5). Although a barcode gap was observable using a 100 base window and Dxy estimates, a comparison of p-distances of the two most similar interspeciÞc sequences (C. capitata 015 and C. pinax 060) lacked the gap at this window size (Fig. 6). All 27 of the C. capitata COI haplotypes documented by Barr (2009) were genetically similar to barcodes in the data base. Seven of the 2009 haplo-
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
344
ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
Vol. 105, no. 2
Fig. 2. Variation estimates and barcode gap for species. (Online Þgure in color.)
types were identical to a C. capitata barcode in the new data set. Although no COI sequences were shared between C. capitata and C. caetrata for the 50 sequences generated in the current study, one of the Barr (2009) C. capitata sequences was identical to a C. caetrata sequence. The other Barr (2009) sequences were ⬎99% similar to the CAP-CAET barcodes. Barcoding of Other Ceratitis Species. Barcode gaps were observed for 25 of the other species in the study. The barcode gap value, calculated between nearest neighboring taxa, was between 3 and 4% for ten species and over 4% for 14 species. The smallest barcode gap (0.95%) was for Ceratitis cosyra and it resulted from a high level of intra-speciÞc variation (9%). This large variation is the result of divergence between a single haplotype, observed in two individuals (403, 404), and the other six haplotypes for the species. As expected, partitioning the haplotypes of the species into genetic clusters (i.e., C. cosyra A and B) helped to generate a barcode gap (Fig. 3). This is consistent with the unconnected networks and large number of mutations separating haplotypes of the species (Fig. 7). The two divergent C. cosyra genetic types were generated from expertly identiÞed material. Although it is possible that the name C. cosyra includes cryptic spe-
Fig. 3. Variation and barcode gaps for taxa collapsed or split. (Online Þgure in color.)
cies, morphological evidence has not been documented to support revision of the taxon. The taxa Capparimyia sp. and Carpophthoromyia dimidiata Bezzi generated barcode gaps over 4%, but also exhibited relatively high levels of intra-speciÞc variation (6 Ð7%). No congeners are included in the study for these two taxa, so it is uncertain if the observed barcode gaps are biologically meaningful. Although De Meyer (2006) reports that Carpophthoromyia vittata (F.) has a parapatric distribution in Kenya with C. dimidiata and the two species share a host range in Kenya, there is no evidence that this second species is present in our collection. The possibility that this nominal species comprises multiple cryptic species, however, cannot be excluded without further study. Ceratitis argenteobrunnea Munro also exhibited a high level of intra-speciÞc variation (6.6%) but the value overlapped with the minimum inter-speciÞc divergence (5.14%). Similar to the variation results for C. cosyra, partitioning the haplotypes into genetic clusters (i.e., C. argenteobrunnea A and B) helped to generate a barcode gap (Fig. 3) and the haplotypes failed to generate a single network (Fig. 7). The other three species that failed to generate a barcode gap were C. fasciventris (Bezzi), C. anonae Graham, and C. rosa. Although the maximum intraspeciÞc variation was not exceptionally high (0.7Ð 2.2%) for these species, the minimum inter-speciÞc variation was low (0 Ð 0.5%). These three species are members of the FAR species complex (Barr and McPheron 2006, Virgilio et al. 2008) and expected to exhibit low levels of genetic separation. Haplotypes from FAR complex species connect into a single network (Fig. 8) and one haplotype is shared between C. fasciventris and C. rosa. Virgilio et al. (2008) report additional information on shared mitochondrial haplotypes and genetic similarity based on more extensive sampling of the three FAR species. Thirteen of the species in the study were represented by a single specimen and could not be tested
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
March 2012
BARR ET AL.: DNA BARCODING THE MEDITERRANEAN FRUIT FLY
345
Fig. 4. Sliding window of total data set. 100 base window (one step) for all sequences (n ⫽ 239) in study and a comparison between C. capitata/C. caetrata samples versus C. pinax samples.
for barcode gaps. The inter-speciÞc values are reported for these species in Supp. Fig. S4. Results from parsimony network analyses support the barcode gap data. For example, the haplotypes of species with low intra-speciÞc variation and barcode gaps ⱖ3% generated species-speciÞc haplotype networks (Supp. Fig. S5), the haplotypes of species that lacked adequate inter-speciÞc variation generated multi-species networks (Figs. 1 and 8), and the haplotypes of species that exhibited high levels of intraspeciÞc variation generated multiple (unconnected) networks (Fig. 7). When the three closely related FAR complex species were grouped and treated as a single taxon in the barcode study, a small barcode gap was measurable (Fig. 3). Unlike the CAP-CAET taxon, the barcode gap was small (⬍1%) relative to the intra-taxonomic diversity (2.5%) of FAR. As expected, partitioning the haplotypes of species with high intra-speciÞc genetic diversity into genetic
clusters (i.e., C. cosyra A and B, and C. argenteobrunnea A and B) helped to generate barcode gaps (Fig. 3). This is consistent with the unconnected networks and large number of mutations separating haplotypes of a species (Fig. 7). Based on the data set, these intraspeciÞc separations were similar to observed genetic distances that separate Ceratitis species. Discussion Identification of C. capitata Using a DNA Barcode. It is not possible to diagnose a C. capitata sample to the level of species by using only a COI DNA barcode sequence. Consistent with previous genetic evidence (Barr and McPheron 2006), DNA barcodes were similar for specimens of C. capitata and its sister taxon, C. caetrata. The two species lacked a “barcode gap” (Meyer and Paulay 2005) for COI, precluding a phenetic-based identiÞcation by using genetic distances. Comparison of the 27 C. capitata and 16 C. caetrata
Fig. 5. Barcode gap between CAP-CAET and C. pinax populations using different fragment sizes. Comparison of sliding window results for taxa as the window size increases. The mean genetic distance is reported for variation within C. pinax (pop-P, n ⫽ 5) and C. capitata/C. caetrata (pop-C, n ⫽ 59) samples and between the two taxonomic groups (pop-CvP, n ⫽ 64). (Online Þgure in color.)
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
346
ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
Fig. 6. Barcode gap between CAP-CAET and C. pinax individuals using different fragment sizes. Comparison of sliding window results for pairwise comparisons between specimens as the window size increases. The range of distances is reported for comparisons between two C. pinax (PvP) samples, one C. capitata and one C. caetrata sample (CvC), and one C. capitata sample and one C. pinax sample (CvP). The PvP (071Ð072) and CvC (033Ð044) samples were selected to maximize genetic distance and the CvP (015Ð069) samples were selected to minimize the distance between taxa. (Online Þgure in color.)
haplotypes in our data set against 848,360 species barcode records in the Barcode of Life Database, BOLD (Ratnasingham and Hebert 2007; www.barcodinglife. org/), resulted in no species level matches because of the high sequence similarity between these two species [search performed 20 January 2011]. Hart and Sunday (2007) proposed that statistical parsimony networks could provide an alternative approach to distinguish species based on DNA barcode data. Using a 95% connection limit as deÞned by Templeton et al. (1992), it is possible to statistically sort haplotypes of different species into unconnected networks (Chen et al. 2010). This approach, however,
Vol. 105, no. 2
Fig. 8. Network of FAR complex species.
connected all C. capitata and C. caetrata haplotypes in a single network. Sarkar et al. (2002, 2008) and Wong et al. (2009) described procedures to identify diagnostic character states within a barcode data set. Using Wong et al.Õs process, our data failed to identify a Þxed diagnostic difference between the two species using individual sites (i.e., simple nucleotide diagnostic, sND) or a combination of states at multiple sites (compound nucleotide diagnostic, cND). Wong et al. (2009) proposed that in the absence of a Þxed cND, it is still possible to identify private haplotypes called “conditional” NDs. Using our barcode data set, all haplotypes for these species are private (not shared between species) and conditional cNDs are present. However, based on the haplotype network and genetic diversity estimates, we suspect that the private haplotypes are an artifact of sampling and not useful as diagnostic
Fig. 7. Haplotypes that did not connect into a single network according to taxonomy.
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
March 2012
BARR ET AL.: DNA BARCODING THE MEDITERRANEAN FRUIT FLY
information. A comparison of our data with C. capitata COI sequences published previously by Barr (2009) identiÞed a C. capitata sequence that is identical to a C. caetrata barcode. In contrast, the phenetic-based and character-based barcoding methods work well to identify a taxon comprising C. capitata and C. caetrata. By merging these species into one taxon, called CAP-CAET, the next most similar species is C. pinax. These lineages can be distinguished by a barcode gap, unconnected networks, and Þxed character differences. The barcodes reported in our study include ⬇600 bases, but it is possible for data generated during a query of the database to be shorter. For example, poor PCR or sequencing reactions (caused by poor quality sample or introduced error) could result in the exclusion of bases during the editing process or require the use of alternate primers to generate the data. To improve PCR or sequencing success rates with degraded DNA of historical samples, Van Houdt et al. (2009) proposed the use of several shorter (minibarcode) regions of the COI gene to generate fruit ßy barcodes. Consequently, the minimum fragment size required to provide a correct identiÞcation of Ceratitis capitata was calculated by testing sliding windows for evidence of a barcode gap. Based on our data, any COI barcode fragment ⱖ300 bases should be adequate for providing a reliable identiÞcation of the CAP-CAET taxon. A comparison of correct identiÞcations across insect orders by using mini-barcodes also noted a reduction in performance when fragment size decreased (Virgilio et al. 2010). Although it is not possible to provide an absolute identiÞcation of C. capitata samples by using the COI barcode, the ability to limit the pool of possible Cetratitis species in a query to just two species (C. capitata or C. caetrata) is useful. If these are the only two possible species, then the ITS-1 locus can be used to distinguish the species (Barr et al. 2006). Based on differences in the biogeography and ecology of C. capitata and C. caetrata, it is also possible to make a species identiÞcation that is conditional on nonmolecular information. For instance, unlike the invasive Mediterranean fruit ßy, C. caetrata has a more restricted host range that includes only indigenous wild fruits (but not commercially grown fruits) and has not been detected outside of Kenya (De Meyer 2001, De Meyer et al. 2002). The likelihood of intercepting C. capitata at U.S. ports of entry should be higher than that for C. caetrata. Conditional identiÞcations, however, are context dependent (i.e., the diagnostic process for one port or country may not be the same for another port or country) and should be justiÞable with scientiÞc information. Comparison of the COI Barcode and Original Restriction Fragment-length Polymorphism Method for Identification of Ceratitis species. From an operational perspective, the proposed “universal” barcode primers of Folmer et al. (1994) did not work consistently for the genus Ceratitis because some species such as C. cristata required alternate primers to generate usable barcodes. Similarly, the restriction fragment-length
347
polymorphism method of Barr et al. (2006) also requires analysis with multiple primer sets for most species in the tool. The rate of PCR-RFLP artifacts or failures using the restriction fragment-length polymorphism method was around 2%, whereas 6% of the samples included in the COI barcode study failed to generate usable barcodes on the Þrst attempt using the Folmer et al. primers. There is evidence of within-genome polymorphism in the COI gene for individuals of some species. In C. cosyra, there was evidence of at least one pseudogene copy of the COI barcode. The presence of polymorphisms and pseudogenes of mitochondrial DNA has been documented previously for various loci in the genus for C. rosa (using the ND6 gene), C. marriotti Munro (using the 3⬘ half of COI gene), and C. venusta (Munro) (using the 3⬘ half of the COI gene) (Barr and McPheron 2006). Magnacca and Brown (2010) discuss the problems associated with DNA barcoding polymorphic and heteroplasmic markers. Although there were no conÞrmed instances of pseudogene copies in the Barr et al. (2006) study, PCR-RFLP artifacts suggested polymorphism in the 12S rDNA and NADHdehydrogenase 6 markers. The COI barcode and restriction fragment-length polymorphism diagnostic tools performed equally well to diagnose most species. Neither method could distinguish between C. capitata and C. caetrata or among the three species of the FAR complex (C. fasciventris, C. anonae, and C. rosa). As expected, these unresolved taxonomic clusters represent two lineages of closely related species (Barr and McPheron 2006, Virgilio et al. 2008). As described previously, it is possible to use either molecular method to diagnose the FAR taxon or the CAP-CAET taxon. Although it is possible to identify C. argenteobrunnea and C. cosyra by using the restriction fragmentlength polymorphism method, these species lack a barcode gap suitable for a distance-based identiÞcation. This difference in identiÞcation capabilities is in part because of the use of different gene regions and how we deÞne the diagnostic characters. The restriction fragment-length polymorphism method is based on the concept of private alleles or “conditional” haplotypes (Wong et al. 2009), and the character differences that separate species are not Þxed within each species. When DNA barcodes are analyzed as private or conditional haplotypes, it is also possible to identify these two species. Using restriction fragment-length polymorphism, the Þve C. argenteobrunnea samples included in our study generate three distinct mitochondrial haplotypes. Using the COI barcode, the Þve samples generate four distinct DNA sequence haplotypes. Although it is possible to estimate which restriction fragment-length polymorphism haplotypes are most similar, it is difÞcult to understand the evolutionary signiÞcance of these similarities because of the limited number of characters in the data. Based on the COI barcode haplotypes, it is clear that the most distant haplotype is separated by at least 37 mutational steps (⬎6% divergence). This is unexpectedly high in com-
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
348
ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
parison to COI barcode variation measured in the other Ceratitis species. In the restriction fragmentlength polymorphism method, this divergent haplotype is just another character, but in the barcode data set this haplotype is ßagged as a problem because of the quantitative distance and haplotype network estimations. The ability to detect variation within species is an advantage of the sequencing approach. DNA barcodes provide a greater level of quantitative information than the restriction fragment-length polymorphism method and an opportunity to estimate uncertainty in the reference database. The restriction fragment-length polymorphism method documented two diagnostic forms for C. cosyra, but barcodes were generated only from specimens with the dominant form. This is because the sample with the uncommon restriction fragmentlength polymorphism form generated multiple copies of the COI gene and was excluded from further analysis. Although the other 10 C. cosyra samples shared an identical restriction fragment-length polymorphism haplotype, two of these ßies had barcode sequences at least 52 mutational steps away from the other eight ßies. Although these data may be useful for identifying cryptic species and guiding systematic revisions within the genus, it is important that the name C. cosyra stay associated with the barcode records until a formal taxonomic change is made. DNA barcodes, like other molecular diagnostic methods, are designed to serve as surrogate characters for recognized species (Carew et al. 2005). Hebert et al. (2003, 2004) proposed using a standardized threshold value as a way to identify genetic distances inconsistent with intra-speciÞc variation. Although a threshold of ⬇3% had been estimated by Hebert et al. for several animal lineages, this value is supposed to be estimated empirically for different taxa. Despite problems with using a standard cut-off threshold value to identify a large group of species (DeSalle et al. 2005, Cognato 2006, Meier et al. 2006, Wiemers and Fiedler 2007, Little and Stevenson 2007), the utility of threshold values for diagnosing pests likely will be context dependent (Armstrong 2010). Based on maximum intra-speciÞc variation, our data lack a standard “cut-off” value for the genus. Using ten times the mean intra-speciÞc value for the entire data set, as proposed by Hebert et al. (2004), the threshold would be near 8%. This is too high for most of our sampled species. Wong et al. (2009) explained that the 95% statistical parsimony connection limit that deÞnes network connectivity also functions as a threshold. Therefore, when species with high intra-speciÞc variation (i.e., C. cosyra and C. argenteobrunnea) were divided into multiple operational taxonomic units (e.g., C. cosyra-A versus C. cosyra-B) based on networks, the maximum intra-speciÞc values for taxa were consistent with the proposed 3% threshold for animal species. Additional information is required to conÞrm that COI diversity within C. cosyra segregates into two diagnosable networks (i.e., two genetic groups for the same species) or if genotypes linking these two networks have yet to be sampled (i.e., there
Vol. 105, no. 2
is one large network for the species). Proper sampling will be important for deciding how to use barcodes for identiÞcation of Ceratitis species and other economic pests. When evaluated using character and distance information, the DNA barcode method performs as well as the restriction fragment-length polymorphism method to identify Ceratitis species. Although the restriction fragment-length polymorphism method is still a more economical approach to molecular identiÞcation based on reagents and instrumentation, other factors can affect the operational costs for high throughput analysis of samples. From a diagnostics standpoint, the DNA barcode method enables better quantitative analysis, provides more information for detecting and assessing false positives and false negatives, and uses a data set that can be easily shared and accessed by the greater research community (Barr 2009). Acknowledgments We thank Lisa Ledezma and Amanda Cook for technical assistance in the experiments, and Raul Ruiz and two anonymous reviewers for suggestions that have improved the manuscript.
References Cited Armstrong, K. 2010. DNA barcoding: a new module in New ZealandÕs plant biosecurity diagnostic toolbox. OEPP/ EPPO Bull. 40: 91Ð100. Armstrong, K. F., and S. L. Ball. 2005. DNA barcodes for biosecurity: invasive species identiÞcation. Phil. Trans. R. Soc. B 360: 1813Ð1823. Armstrong, K. F., C. M. Cameron, and E. R. Frampton. 1997. Fruit ßy (Diptera: Tephritidae) species identiÞcation: a rapid molecular diagnostic technique for quarantine application. Bull. Entomol. Res. 87: 111Ð118. Ball, S. L., and K. F. Armstrong. 2006. DNA barcodes for insect pest identiÞcation: a test case with tussock moths (Lepidoptera: Lymantriidae). Can. J. For. Res. 36: 337Ð 350. Barr, N. B. 2009. Pathway analysis of Ceratitis capitata (Diptera: Tephritidae) using mitochondrial DNA. J. Econ. Entomol. 102: 401Ð 411. Barr, N. B., and B. A. McPheron. 2006. Molecular phylogenetics of the genus Ceratitis (Diptera: Tephritidae). Mol. Phylogenet. Evol. 38: 216 Ð230. Barr, N. B., and B. M. Wiegmann. 2009. Phylogenetic relationships of Ceratitis fruit ßies inferred from nuclear CAD and tango/ARNT gene fragments: testing monophyly of the subgenera Ceratitis (Ceratitis) and C. (Pterandrus). Mol. Phylogenet. Evol. 53: 412Ð 424. Barr, N. B., R. S. Copeland, M. De Meyer, D. Masiga, H. G. Kibogo, M. K. Billah, E. Osir, R. A. Wharton, and B. A. McPheron. 2006. Molecular diagnostics of economically important Ceratitis fruit ßy species (Diptera: Tephritidae) in Africa using PCR and RFLP analyses. Bull. Entomol. Res. 96: 505Ð521. Carew, M. E., V. Pettigrove, and A. A. Hoffman. 2005. The utility of DNA markers in classical taxonomy using cytochrome oxidase I markers to differentiate Australian Cladopelma (Diptera: Chironomidae) midges. Ann. Entomol. Soc. Am. 98: 587Ð594.
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
March 2012
BARR ET AL.: DNA BARCODING THE MEDITERRANEAN FRUIT FLY
Chen, H., M. Strand, J. L. Norenburg, S. Sun, H. Kajihara, A. V. Chernyshev, S. A. Maslakova, and P. Sundberg. 2010. Statistical parsimony networks and species assemblages in Cephalotrichid Nemerteans (Nemertea). PLoS ONE 5: e12885. Cognato, A. I. 2006. Standard percent DNA sequence difference for insects does not predict species boundaries. J. Econ. Entomol. 99: 1037Ð1045. Copeland, R. S., R. A. Wharton, Q. Luke, and M. De Meyer. 2002. Indigenous hosts of Ceratitis capitata (Diptera: Tephritidae) in Kenya. Ann. Entomol. Soc. Am. 95: 672Ð 694. Copeland, R. S., R. A. Wharton, Q. Luke, M. De Meyer, S. Lux, N. Zenz, P. Machera, and M. Okumu. 2006. Geographic distribution, host fruit, and parasitoids of African fruit ßy pests Ceratitis anonae, Ceratitis cosyra, Ceratitis fasciventris, and Ceratitis rosa (Diptera: Tephritidae) in Kenya. Ann. Entomol. Soc. Am. 99: 261Ð278. Clement, M., D. Posada, and K. A. Crandall. 2000. TCS: a computer program to estimate gene genealogies. Mol. Ecol. 9: 1657Ð1660. De Meyer, M. 1996. Revision of the subgenus Ceratitis (Pardalaspis) Bezzi, 1918 (Diptera: Tephritidae: Ceratitini). Syst. Entomol. 21: 15Ð26. De Meyer, M. 1998. Revision of the subgenus Ceratitis (Ceratalaspis) Hancock (Diptera: Tephritidae). Bull. Entomol. Res. 88: 257Ð290. De Meyer, M. 2000a. Phylogeny of the genus Ceratitis (Dacinae: Ceratitidini), pp. 409 Ð 428. In M. Aluja and A. L. Norrbom (eds.), Fruit ßies (Tephritidae): phylogeny and evolution of behavior. CRC, Boca Raton, FL. De Meyer, M. 2000b. Systematic revision of the subgenus Ceratitis MacLeay s. s. (Diptera: Tephritidae). Zool. J. Linn. Soc. 128: 439 Ð 467. De Meyer, M. 2001. Distribution patterns and host-plant relationships within the genus Ceratitis MacLeay (Diptera: Tephritidae) in Africa. Cimbebasia 17: 219 Ð228. De Meyer, M. 2005. Phylogenetic relationships within the fruit ßy genus Ceratitis MacLeay (Diptera: Tephritidae), derived from morphological and plant evidence. Insect Syst. Evol. 36: 459 Ð 480. De Meyer, M. 2006. Systematic revision of the fruit ßy genus Carpophthoromyia Austen (Diptera, Tephritidae). Zootaxa 1235: 1Ð 48. De Meyer, M., and R. S. Copeland. 2005. Description of new Ceratitis MacLeay (Diptera: Tephritidae) species from Africa. J. Nat. Hist. 39: 1283Ð1297. De Meyer, M., and A. Freidberg. 2005a. Revision of the subgenus Ceratitis (Pterandrus) Bezzi (Diptera: Tephritidae). Isr. J. Entomol. 35-. 36: 197Ð315. De Meyer, M., and A. Freidberg. 2005b. Revision of the fruit ßy genus Capparimyia (Diptera, Tephritidae). Zoologica Scr. 34: 279 Ð303. De Meyer, M., R. S. Copeland, S. A. Lux, M. Mansell, S. Quilici, R. Wharton, I. M. White, and N. J. Zenz. 2002. Annotated check list of host plants for afrotropical fruit ßies (Diptera: Tephritidae) of the genus Ceratitis. Documentations Zoologiques, Musee Royal de lÕAfrique Centrale 27: 1Ð91. De Meyer, M., R. S. Copeland, R. A. Wharton, and B. A. McPheron. 2004. On the geographical origin of the medßy, Ceratitis capitata (Wiedemann), pp. 45Ð53. In B. Barnes (ed.), Proceedings, the 6th International Symposium on Fruit Flies of Economic Importance, 6 Ð10 May 2002, Stellenbosch, South Africa. Isteg ScientiÞc Publications, Irene, South Africa. De Meyer, M., M. P. Robertson, A. T. Peterson, and M. W. Mansell. 2008. Ecological niches and potential distribu-
349
tions of Mediterranean fruit ßy (Ceratitis capitata) and Natal fruit ßy (Ceratitis rosa). J. Biogeogr. 35: 270 Ð281. DeSalle, R., M. G. Egan, and M. Siddall. 2005. The unholy trinity: taxonomy, species delimitation and DNA barcoding. Phil. Trans. Soc. B 360: 1905Ð1916. deWaard, J. R., A. Mitchell, M. A. Keena, D. Gopurenko, L. M. Boykin, K. F. Armstrong, M. G. Pogue, J. Lima, R. Floyd, R. H. Hanner, et al. 2010. Towards a global barcode library for Lymantria (Lepidoptera: Lymantriinae) tussock moths of biosecurity concern. PLoS ONE 5: e14280. Douglas, L. J., and D. S. Haymer. 2001. Ribosomal ITS1 polymorphism in Ceratitis capitata and Ceratitis rosa (Diptera: Tephritidae). Ann. Entomol. Soc. Am. 94: 726 Ð 731. Floyd, R., J. Lima, J. deWaard, L. Humble, and R. Hanner. 2010. Common goals: policy implications of DNA barcoding as a protocol for identiÞcation of arthropod pests. Biol. Invasions 12: 2947Ð2954. Folmer, O., M. Black, W. Hoeh, R. Lutz, and R. Vrijenhoek. 1994. DNA primers for ampliÞcation of mitochondrial cytochrome c oxidase I from diverse metazoan invertebrates. Mol. Marine Biol. Biotechnol. 3: 294 Ð299. Hart, M. W., and J. Sunday. 2007. Things fall apart: biological species form unconnected parsimony networks. Biol. Lett. 3: 509 Ð512. Hebert, P.D.N., A. Cywinska, S. L. Ball, and J. R. deWaard. 2003. Biological identiÞcations through DNA barcodes. Proc. R. Soc. Lond. B 270: 313Ð321. Hebert, P.D.N., A. Cywinska, S. L. Ball, and J. R. deWaard. 2004. Biological identiÞcations through DNA barcodes. Proc. R. Soc. Lond. B 270: 313Ð321. Huang, C.-G., J.-C. Hsu, D. S. Haymer, G.-C. Lin, and W.-J. Wu. 2009. Rapid identiÞcation of the Mediterranean fruit ßy (Diptera: Tephritidae) by loop-mediated isothermal ampliÞcation. J. Econ. Entomol. 102: 1239 Ð1246. Kakouli-Duarte, T., D. G. Casey, and A. M. Burnell. 2001. Development of a diagnostic DNA probe for the fruit ßies Ceratitis capitata and Ceratitis rosa (Diptera: Tephritidae) using ampliÞed fragment-length polymorphism. J. Econ. Entomol. 94: 989 Ð997. Kumar, S., M. Nei, J. Dudley, and K. Tamura. 2008. MEGA: a biologist-centric software for evolutionary analysis of DNA and protein sequences. Brief. Bioinform. 9: 299 Ð306. Lanzavecchia, S. B., J. L. Cladera, P. Faccio, N. Petit Marty, J. C. Vilardi, and R. O. Zandomeni. 2008. Origin and distribution of Ceratitis capitata mitochondrial DNA haplotypes in Argentina. Ann. Entomol. Soc. Am. 101: 627Ð 638. Librado, P., and J. Rozas. 2009. DnaSP version 5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25: 1451Ð1452. Little, D. P., and D. W. Stevenson. 2007. A comparison of algorithms for the identiÞcation of specimens using DNA barcodes: examples from gymnosperms. Cladistics 23: 1Ð21. Lowenstein, J. H., G. Amato, and S.-O. Kolokotronis. 2009. The real maccoyii: identifying tuna sushi with DNA barcodes Ð contrasting characteristic attributes and genetic distances. PLoS ONE 4: e7866. Magnacca, K. N., and M.J.F. Brown. 2010. Tissue segregation of mitochondrial haplotypes in heteroplasmic Hawaiian bees: implications for DNA barcoding. Mol. Ecol. Res. 10: 60 Ð 68. Meier, R., K. Shiyang, G. Vaidya, and P.K.L. Ng. 2006. DNA barcoding and taxonomy in Diptera: a tale of high intraspeciÞc variability and low identiÞcation success. Syst. Biol. 55: 715Ð728.
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018
350
ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
Meyer, C. P., and G. Paulay. 2005. DNA barcoding: error rates based on comprehensive sampling. PLoS Biol. 3: e422. Nei, M., and S. Kumar. 2000. Molecular evolution and phylogenetics. Oxford University Press, New York. Naro-Maciel, E., B. Reid, N. N. Fitzsimmons, M. Le, R. DeSalle, and G. Amato. 2010. DNA barcodes for globally threatened marine turtles: a registry approach to documenting biodiversity. Mol. Ecol. Res. 10: 252Ð263. Ratnasingham, S., and P.D.N. Hebert. 2007. BOLD: the Barcode of Life Data System (www.barcodinglife.org). Mol. Ecol. Notes 7: 355Ð364. Sarkar, I. N., P. J. Planet, T. E. Bael, S. E. Stanley, M. Siddall, R. DeSalle, and D. H. Figurski. 2002. Characteristic attributes in cancer microarrays. J. Biomed. Inform. 35: 111Ð122. Sarkar, I. N., P. J. Planet, and R. DeSalle. 2008. CAOS software for use in character-based DNA barcoding. Mol. Ecol. Res. 8: 1256 Ð1259. Scheffer, S. J., M. L. Lewis, and R. C. Josh. 2006. DNA barcoding applied to invasive leafminers (Diptera: Agromyzidae) in the Philippines. Ann. Entomol. Soc. Am. 99: 204 Ð210. Simon, C., F. Frati, A. Beckenbach, B. Crespi, H. Liu, and P. Flook. 1994. Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved polymerase chain reaction primers. Ann. Entomol. Soc. Am. 87: 651Ð701. Srivathsan, A., and R. Meier. 2011. On the inappropriate use of Kimura 2-parameter (K2P) divergences in the DNAbarcoding literature. Cladistics 27: 1Ð5. Templeton, A. R., K. A. Crandall, and C. F. Sing. 1992. A cladistics analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. Genetics 132: 619 Ð 633. Van Houdt, J.K.J., F. C. Breman, M. Virgilio, and M. De Meyer. 2009. Recovering full DNA barcodes from nat-
Vol. 105, no. 2
ural history collections of Tephritid fruitßies (Tephritidae, Diptera) using mini barcodes. Mol. Ecol. Res. 10: 459 Ð 465. Vera, M. T., R. Rodriguez, D. F. Segura, J. L. Cladera, and R. W. Sutherst. 2002. Potential geographical distribution of the Mediterranean fruit ßy, Ceratitis capitata (Diptera: Tephritidae), with emphasis on Argentina and Australia. Environ. Entomol. 31: 1009 Ð1022. Virgilio, M., T. Backeljau, N. Barr, and M. De Meyer. 2008. Molecular evaluation of nominal species in Ceratitis fasciventris, C. anonae, C. rosa species complex (Diptera: Tephritidae). Mol. Phylogenet. Evol. 48: 270 Ð280. Virgilio, M., T. Backeljau, B. Nevado, and M. De Meyer. 2010. Comparative performances of DNA barcoding across insect orders. BMC Bioinform. 11: 206. White, I. M., and M. M. Elson-Harris. 1992. Fruit ßies of economic signiÞcance: their identiÞcation and bionomics. CAB, Wallingford, United Kingdom. Wiemers, M., and K. Fiedler. 2007. Does the DNA barcoding gap exist? Ð a case study in blue butterßies (Lepidoptera: Lycaenidae). Front. Zool. 4: 8. Wong, E.H.-K., M. S. Shivji, and R. H. Hanner. 2009. Identifying sharks with DNA barcodes: assessing the utility of a nucleotide diagnostic approach. Mol. Ecol. Res. 9: 243Ð 256. Yancy, H. F., T. S. Zemlak, J. A. Mason, J. D. Washington, B. J. Tenge, N.-L.T. Nguyen, J. D. Barnett, W. E. Savary, W. E. Hill, M. M. Moore, et al. 2008. Potential use of DNA barcodes in regulatory science: applications of the Regulatory Fish Encyclopedia. J. Food Prot. 71: 210 Ð217. Yuval, B., and J. Hendrichs. 2000. Behavior of ßies in the genus Ceratitis (Dacinae: Ceratitidini), pp. 429 Ð 457. In M. Aluja and A. L. Norrbom (eds.), Fruit ßies (Tephritidae): Phylogeny and evolution of behavior. CRC, Boca Raton, FL. Received 7 June 2011; accepted 4 October 2011.
Downloaded from https://academic.oup.com/aesa/article-abstract/105/2/339/120778 by guest on 05 February 2018