Comparison of Landmark- and Outline-Based Geometric...

Research ArticleComparison of Landmark- and Outline-BasedGeometric Morphometrics for Discriminating MosquitoVectors in Ratchaburi Province Thailand

Tanawat Chaiphongpachara

College of Allied Health Science Suan Sunandha Rajabhat University Samut Songkhram 75000ailand

Correspondence should be addressed to Tanawat Chaiphongpachara tanawatchssruacth

Received 3 May 2018 Accepted 25 October 2018 Published 6 November 2018

Academic Editor Gabriele Gentile

Copyright copy 2018 Tanawat ChaiphongpacharaThis is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

It is often challenging to identify mosquito vectors in the field based on morphological features due to their similar morphologiesand difficulties in obtaining undamaged samples but is required for their successful control Geometric morphometrics (GM)overcomes this issue by analyzing a suite of traits simultaneously and has the added advantages of being easy to use low cost andquickTherefore this research compared the efficiency and precision of landmark- and outline-basedGM techniques for separatingspecies of mosquitoes in Huay Nam Nak village Ratchaburi Province Thailand This research collected 273 individuals belongingto seven species Anopheles barbirostris An subpictus Culex quinquefasciatus Cx vishnui Cx whitmorei Aedes aegypti and Aealbopictus Both landmark-based and outline-based GM techniques could identify malaria vectors in this area to the genus levelsuccessfully and were also very effective for identifying the malaria vectorsAnopheles spp and the dengue vectorsAedes spp to thespecies level However they were less effective for distinguishing between species of Culex Therefore GM represents a valuabletool for the identification of mosquito vectors in the field which will facilitate their successful control

1 Introduction

Mosquitoes (Diptera Culicidae) are medically importantsmall insects that transmit many diseases to humans particu-larly in tropical and subtropical regions [1] Globally there are3400 species of mosquitoes in 42 genera many of which areimportant vectors of pathogens [2]Mosquito-borne diseaseswhich include dengue fever (DF) chikungunya malariaJapanese encephalitis and filariasis [3] account for 17 of allinfectious diseases worldwide and are responsible for approx-imately 14 million deaths per year [4] making them a seriouspublic health issue

Mosquito-borne diseases are one of the most importanthealth concerns in Thailand According to the Bureau ofEpidemiology ofThailand DF had the highest incidence ratein 2014 at 3567 cases per 100000 people followed by malaria(1748) chikungunya (029) filariasis (003) and Japaneseencephalitis (002) [5] Ratchaburi Province on the western

border of Thailand has one of the highest incidences of mos-quito-borne diseases particularly DF and malaria with mor-bidity rates of 3068 and 895 cases per 100000 populationrespectively Therefore it shows that Ratchaburi Provincehas an epidemic of mosquito-borne diseases still makingit important that control methods are found to reduce thenumber of patients in this area

Mosquito control is one important strategy for control-lingmosquito-borne disease epidemics [6ndash8] However sincedifferent species of mosquitoes have different characteristicssuch as behaviors breeding sites andepidemiologies a know-ledge of which vector species are present in the area [9 10]and the epidemiological patterns of disease transmission isrequired to ensure the planned control method is appropriate[11 12] The It is can be challenging to identify mosquitovectors in the field based on morphological features becausemany species are cryptic species sibling species or isomor-phic species with similar morphologies [13] Furthermore

HindawiBioMed Research InternationalVolume 2018 Article ID 6170502 10 pageshttpsdoiorg10115520186170502

2 BioMed Research International

it is more difficult to obtain a mosquito sample in the fieldthan in the laboratory because the external characteristicsare often damaged during trapping and transportation [14]One solution to this is the use of high-efficiency moleculartechniques for mosquito identification However these havevery high associated costs [14 15] making them unsuitablefor use in the field where many samples are collected Thus anew technique for identifying mosquito vectors in the field isrequired

Geometric morphometrics (GM) is a fast inexpensivetechnique [14 16] used to analyze the size and shape of indi-viduals based on a suite of traits GM has been appliedwidely in a number of fields including entomology (egmos-quito [17] blow fly [18] bee [19] and eggs of Triatominae[20]) Furthermore several studies have used theGMmethodto classify species and to examine variation among medi-cally important mosquitoes that are morphologically sim-ilar or sibling species [14 15 21ndash24] GM analyses can beconducted using landmark- or outline-based methods [13]each of which has different advantages depending on thecharacteristics and specificity of the sample The landmark-based approach uses the coordinates of landmarks to analyzethe morphology [23] and is popular in the field of medicalentomology with many studies having shown it can distin-guish different species of mosquito vectors successfully suchas Anopheles spp and Aedes spp [14 21 23] By contrastthe outline-based approach uses contour data [25] and hasalso been shown to be useful for identifying some mosquitospecies such as Aedes scutellaris [14] However surprisinglythe specificity with which mosquito vectors can be identifieddiffers between the two techniques [14 23] and thereforea comparison of their performance in the field is requiredbefore they can be used

Thus the aim of this study was to compare the efficiencyof landmark- and outline-based GM techniques to separatespecies of mosquito vectors in Huay Nam Nak village inRatchaburi Province Thailand an endemic area for mosqui-to-borne diseases The result from this research will serve asa guideline for choosing the best GM technique for the iden-tification of vectors in the field in Thailand to facilitate thecontrol of vector-borne diseases

2 Materials and Methods

21 Study Site and Mosquito Collection Mosquitoes werecollected from Huay Nam Nak village (13∘22101584036010158401015840N99∘16101584034910158401015840E) in the Suan Phueng District of Ratchaburi Pro-vince Thailand during June to August 2015 (Figure 1) Thestudy site consistedmainly of woodenhouses and agriculturalfields with streams flowing through and mountains and hillyforests surrounding the village

Three Mosquito Magnet Independence Mosquito Traps(Woodstream Corporation USA) were placed approximately5 meters away from houses in the village for 24 hours (600am to 600 am of the following day) per week over the 3-month study period to collect nocturnal and diurnal mos-quito vectors Mosquito samples were collected from eachtrap in the morning (600 am) and transported to thelaboratory of the College of Allied Health Science Suan

Sunandha Rajabhat University Samut Songkhram Provincefor morphological identification Female mosquitoes werethen morphologically identified to species using the illus-trated keys to the mosquitoes of Thailand [26]

22 Mosquito Wing Preparation Only the right wings offemale mosquitoes were analyzed by GM The right wingof each individual was dissected under a Nikon AZ 100Mstereomicroscope (Nikon Corp Tokyo Japan) and placedbetween a glass microscope slide and coverslip using Hoyerrsquossolution as amountingmediumThen all of thewing samplesfor each species were photographed using a Nikon DS-Ri1 SIGHT digital camera connected to a Nikon EclipseE600 microscope (Nikon Corp Tokyo Japan) under 40timesmagnification alongside a 1-mm scale bar and were analyzedusing the Collecting Landmarks for Identification and Char-acterization (CLIC) Program Since the aim of this study wasto compare the effectiveness of landmark- and outline-basedGM for species discrimination the same set of wing pictureswas used for each species in the two analyses outlined below

23 Landmark-Based GM Analysis A total of 17 landmarkswere selected based on the ease with which they could beplotted across all mosquito species and their low likelihood ofbeing damaged eg by wing vein intersection The positionsof these landmarks on the wing of each individual weredigitized (Figure 2) The wing size was then computed as thecentroid size (CS) which is the square root of the sum of thesquared distances between each individual landmark and thecenter of the landmark configuration [14 27]Thewing shapevariables (partial warps [PW]) were computed as principalcomponents (PCs) of the PW(known as relative warps [RW])after Generalised Procrustes Analysis (GPA) (the statisticalprocedure of superimposition) and discriminant analysis(DA) (or Canonical Variate Analysis [CVA]) was then used toanalyze the shape variables for the separation of each species

24 Outline-BasedGMAnalysis Theoutline used for analysiswas constructed using coordinates along the contour of thelower part of the mosquito wing (Figure 3) This analyticalcontour is the most complete part of the wing in all mosqui-toes and also lacks thick scales and is difficult to tear Thewing size was estimated as the length of the perimeter of thecontour which was separate from the set of shape variablesusedThe shape variables were constructed using the normal-ized elliptic Fourier coefficients (NEF)

25 Statistical Analysis Prior to conducting theGManalysesthis research assessed the quality and measurement errorof the digitized landmarks and digitized contour used forthe landmark-based and outline-based GM analyses respec-tively by calculating the repeatability index (R)The repeata-bility of each coordinate was tested for 70 randomly selectedwings (10 wings per species) In order to do this measure-ments of the coordinates used in the original wing pho-tographs were repeated to give a total of 140 images The re-peatability was then computed based on an ANOVA designPrior to analysis the error was reduced by averaging the twodigitizations [23 25]

BioMed Research International 3

N

Thailand

Huay Nam Nak Village

km

0 200100

Figure 1 Location of the study site

12

3

4

5

6

7

8

9

10

11

12

13

1415

16

17

m

Figure 2 Position of the 17 landmarks on the mosquito wing used for the landmark-based GM analysis

The variation in wing CS (in mm) of mosquito sampleseach species was illustrated by quantile plots between P 25and P 75 Differences in the wing CS (for landmark-basedGM) and the length of the perimeter (for outline-based GM)among species were analyzed using nonparametric permuta-tion tests (1000 runs) with Bonferroni correction test and asignificance level of p lt 005

Wing shape variation among species was also visualizedby superimposition the mean landmark configurations usingProcrustes superimposition in the landmark-based analysis

and using EFA in the outline-based analysis The shape vari-ables for each species were separated using DA and displayedas a factor map The Mahalanobis distance was calculatedfromDA to assess the degree of similarity betweenpopulationand differences in theMahalanobis distance (ie wing shape)among species were computed using nonparametric permu-tation tests (1000 runs) with Bonferroni correction test and asignificance level of p lt 005 Then validated reclassificationwhereby each individual was allocated to its closest groupaccording to the Mahalanobis distance without being used


m

Figure 3 Digitized contour of the lower section of a mosquito wing used for the outline-based GM analysis

to help determine the group center was performed [23] anda single linkage hierarchical classification tree was createdbased on the bootstrap technique on input data within spe-cies according to Morales Vargas etal [28] and visualizedusing XYOM

Finally since size and shape are not independent attri-butes this research examined the relationship between thetwo (the allometric effect) also using linear correlation [13]after the GM analysis of the size variables

26 Soware Both the landmark- and outline-based GManalyses were performed using the CLIC package version97 which was developed by Professor Jean-Pierre Dujardin[13 29] and is freely available at httpxyom-cliceu Variousmodules of the CLIC package were used including the COOmodule to digitize the landmarks or pseudolandmarks TheTET module was used to transform the data for analysisTheMOG and FOGmodules were used to construct the sizeand shape variables perform principle component analyses(PCA) and DA compute Procrustes distances and createquantile plots in the landmark- and outline-based analysesrespectively The VAR module was used to analyze allometryand the statistical significance of differences in the size vari-ables among species The PAD module was used to analyzethe statistical significance of differences in the shape variablesamong species The single linkage hierarchical classificationtree as built by the recent online morphometric packageXYOM (httpsxyomio) was compared to the UPGMA treeas computed by the R software (httpscranr-projectorg)

3 Results

This research collected 273 mosquitoes belonging to sevenspecies within three genera (Table 1)

31 Repeatability The two sets of repeated measurementsfrom the same images used in the landmark- and outline-based GM analyses showed good repeatability scores forboth size and shape In the landmark-based analysis the

Table 1 Species and numbers of mosquitoes used in the analyses

Species Number of mosquitoesAnopheles barbirostris 31Anopheles subpictus 61Culex quinquefasciatus 21Culex vishnui 70Culex whitmorei 24Aedes aegypti 45Aedes albopictus 21Total 273

repeatability of the CS was 098 while in the outline-basedanalysis the repeatability of the perimeter length was 097

32 Allometry This research found a weak relationship be-tween size and shape with the allometric residuals of the dis-criminant factors explaining 8 of the variation in thelandmark-based GM and 5 of the variation in the outline-based GM However since both of the components are im-portant for species identification neither was removed fromthe GM analyses

33 Wing Size In the landmark-based analysis An barbi-rostris had the largest wings (mean = 354 mm) and Aeaegypti had the smallest (mean = 219 mm) (Table 2 and Fig-ure 4) In addition An barbirostris had significant intraspeci-fic variation in mean CS that was the highest among all spe-cies

In the outline-based analysis An barbirostris again hadthe largest wings (mean = 451 mm) while Cx whitmorei hadthe smallest (mean = 339mm) Both An barbirostris and Aealbopictus exhibited significant intraspecific variation in themean perimeter length of the contour

34 Wing Shape Superimposition of the mean wing shapesof each mosquito species using Procrustes superimposition


Aedes albopictusAedes aegyptiCulex whitmoreiCulex vishnuiCulex quinquefasciatusAnopheles subpictusAnopheles barbirostris

Minimum Median MaximumP 25 P 75

16 19 22 25 28 30 33 36 39

(a)


P 25 P 75

268237 298 328 358 388 418 449 479

Minimum Median Maximum

(b)

Figure 4 Variation in (a) the wing centroid size used in the landmark-based analysis and (b) the perimeter length of the contour used in theoutline-based approach in mosquito species The plots in each panel show the 25th and 75th quartiles and the median

Table 2 Mean wing centroid size and mean perimeter length of the contour in the seven mosquito species

Species Landmark-based approach Outline-based approachn Mean (Min - Max) (mm) SD Mean (Min - Max) (mm) SD

An barbirostris 31 354 (308 - 388)a 019 451(406 - 479)a 021An subpictus 61 274(225 - 356)b 034 358(292 - 452)bc 042Cx quinquefasciatus 21 271(213 - 319)b 033 346(237 - 422)b 053Cx vishnui 70 270(213 - 317)b 026 351(245 - 405)b 034Cx whitmorei 24 266(249 - 282)b 011 339(313 - 365)b 017Ae aegypti 45 219(163 - 256)c 021 371(275 - 437)c 035Ae albopictus 21 230(182 - 250)c 015 388(354 - 416)d 019Species with different superscripts letters had significantly different wing sizes at p lt 005 Min minimum Max maximum SD standard deviation

showed the positions of landmarks 1 7 8 12 13 and 17 variedamong species in the landmark-based analysis (Figure 5)and the outlines of Ae aegypti and Ae albopictus were dis-tinct from the other species in the outline-based analysis(Figure 6)

Comparison of the factor maps derived fromDA showedboth the landmark- and outline-based GM analyses gave verysimilar results particularly at the genus level (Figure 7) Basedon the Mahalanobis distances wing shape was significantlydifferent among all species in both the landmark- andoutline-basedGManalyses (nonparametric permutation test1000 cycles p lt 005 Table 3)The greatest Mahalanobisdistanceswere betweenAe aegypti andAn subpictus (886) inthe landmark-based analysis and between Ae albopictus andAn barbirostris (2267) in the outline-based analysis

The validated reclassification scores were high (gt80accuracy) for five species using the landmark-based approach(An barbirostris [96] An subpictus [81] Cx whitmorei[91] Ae aegypti [88] and Ae albopictus [90]) and fourspecies using the outline-based approach (An barbirostris[93] An subpictus [86] Cx vishnui [85] Ae aegypti[86] and Ae albopictus [95]) The validated reclassifica-tion scores were lowest for Cx vishnui with the landmark-based approach (65) and for Cx quinquefasciatus withthe outline-based approach (42) (Table 4) Single LinkageHierarchical classification trees separated each species of

mosquito (Figure 8) The trees that were produced using thetwo different approaches were very similar

4 Discussion

This research analyzed 273 wings across seven species ofmosquitoes using two different GM approaches Mosquitowings are almost bidimensional and relatively rigid whichreduces mistakes when digitizing them for GM analysis [16]as reflected by the good repeatability scores for both size andshape in the present study

41 Wing Size This result found few differences between thelandmark- and outline-based analyses in terms of the wingCS with both methods showing An barbirostris and Aedesspp had significantly larger and smaller wings respectivelythan the other species However these species also exhib-ited significant intraspecific differences in size likely dueto variation in factors such as temperature humidity andfood availability [15] Because of such variation it has beensuggested previously shape is more appropriate than size fordistinguishing among morphologically similar species and isalso more informative in terms of the genetics and evolutionof organisms [15 16 23] However wing size is also usefulin the initial identification of species particularly since some


An barbirostrisAn subpictusCx quinquefasciatusCx vishnui Cx whitmoreiAe aegyptiAe albopictus

1

23

4

5

678 9

10

1112

13 14

15

16

17

Figure 5 Superimposition of the mean landmark configurations of the wings for the seven mosquito species


Figure 6 Superimposition of the outlines of the wings for the seven mosquito species

Table 3 Mahalanobis distances of wing shape among the seven mosquito species

Mahalanobis distancesAB AS CQ CW CV AAE AAL

Landmark-based approachAn barbirostris 000An subpictus 410lowast 000

Cx quinquefasciatus 783lowast 833lowast 000

Cx vishnui 755lowast 770lowast 274lowast 000

Cx whitmorei 791lowast 812lowast 392lowast 310lowast 000

Ae aegypti 811lowast 886lowast 520lowast 596lowast 668lowast 000

Ae albopictus 785lowast 875lowast 649lowast 738lowast 819lowast 447lowast 000Outline-based approachAn barbirostris 000An subpictus 741lowast 000





Ae albopictus 2267lowast 2155lowast 2138lowast 2160lowast 2171lowast 604lowast 000lowast indicates statistical significance at plt 005


An barbirostris Cx quinquefasciatusAn subpictus Cx vishnui

Canonical variable 1

Cano

nica

l var

iabl

e 2

568

713

minus436minus49

2

(a)

Cx whitmorei Ae albopictusAe aegypti

Cano

nica

l var

iabl

e 2


175

5

013 1287

minus65

1

(b)

Figure 7 Factor maps from (a) landmark- and (b) outline-based discriminant analysis

Table 4 Validated reclassification accuracies for the seven mosquito species

Species Percentage of reclassificationLandmark-based approach Outline-based approach

An barbirostris 96 (3031) 93 (2931)An subpictus 81 (5061) 86 (5361)Cx quinquefasciatus 66 (1421) 42 (921)Cx vishnui 65 (4670) 85 (6070)Cx whitmorei 91 (2224) 79 (1924)Ae aegypti 88 (4045) 86 (3945)Ae albopictus 90 (1921) 95 (2021)

species of mosquitoes are larger than many other species forexampleAn barbirostriswas clearly larger thanCulex speciesin this study Similarly Stanford et al [29] found wing beatfrequencies which mediate assortative mating are related tosize and consequently are unique for each species

42 Wing Shape The shape and venation of mosquito wingsare unique characteristics that can be used to separatedifferent genera and species [14] This research found littledifference between the landmark- and outline-based meth-ods when considering the wing shape of mosquitoes Forboth methods factor maps derived from DA showed therewas no overlap between genera but some overlap betweenspecies (Figure 7) However wing shape was found to besignificantly different among all seven species using bothmethods Similarly Wilke et al [30] found GM was a goodtool (100 accuracy) for species identification at the genus

level for Aedes Anopheles and Culex but was also efficientat the subgenus and species levels In both GM analysesthe reclassification scores were higher for Anopheles spp andAedes spp (gt80 accuracy in all species) than for Culexspp with the exception of Cx whitmorei for the landmark-based analysis (91) and Cx vishnui for the outline-basedanalysis (85) Consistent with this the single linkage hier-archical classification trees clearly separated the three generain the landmark- and outline-based analysis (Figure 8(a))Superimposition of the mean landmark configurations ofeach mosquito species showed landmarks 2 3 5 6 7 9and 10 were in similar places for Culex spp (particularlyCx quinquefasciatus and Cx vishnui) and the other species(Figure 5)

These findings show GM can be used successfully toclassify Anopheles spp which are often implicated in malariatransmission An barbirostris is a suspected vector of Plas-modium spp in Thailand [31] while An subpictus is not


Aedes spp Culex spp Anopheles spp

6

5

4

3

2

1

0Ae aegypti Ae albopictus Cx quinquefasciatus Cx vishnui Cx whitmorei An barbirostris An subpictus

(a)


6

5

4

3

2

1

0Ae albopictus Ae aegypti Cx whitmorei Cx quinquefasciatus Cx vishnui An subpictus An barbirostris

(b)

Figure 8 Single linkage hierarchical classification tree for the (a) landmark and (b) outline-based geometric morphometric analyses

considered a malaria vector in Thailand but is a vector inSri Lanka [32] GM is also useful for identifying Aedes sppincluding Ae aegypti and Ae albopictus which are dengueand chikungunya vectors respectively [14] However GMhad low classification rates for Culex spp including Cxquinquefasciatus Cx vishnui and Cx whitmorei which havebeen incriminated as Japanese encephalitis and filariasisvectors [26]

Our findings demonstrate both GM approaches can beused to identify mosquitoes in the study area particularly tothe genus level Currently the landmark-based GM approachis the most popular for helping to identify species and inves-tigating variation among vectors [14 21 23] as it is less time-consuming requiring the definition of only a few analyticalpoints for analysis and is a powerful approach By con-trast few studies use the outline-based GM approach be-cause it requires a lot of time and samples than the landmark-based approach However this approach has the advantage ofnot requiring the use of specific locations [25] Furthermorethis research found that the outline-based GM approach wasbetter andmore effective at discriminating some species suchas Cx vishnui

5 Conclusions

Both the landmark- and outline-based GM methods arepractical and effective for discriminating between species ofmosquitoes in the study area However since each species ofmosquito has a unique wing identity the best method needsto be selected to suit the species occurring in a particularvector control area GM has the advantages of being easyto use low cost and quick as well as not requiring advancedentomological skills and these advantages make it particu-larly attractive for use in the field to facilitate the control ofmosquito vectors

Data Availability

The data supporting the conclusions of this article are pro-vided within the article The datasets generated and analyzedduring the current study are available from the correspondingauthor upon reasonable request

Conflicts of Interest

The author declares that there are no conflicts of interest


Acknowledgments

The author would like to thank College of Allied HealthScience Suan Sunandha Rajabhat University Thailand fortheir kind support of our research This work was supportedby Suan Sunandha Rajabhat University Bangkok Thailand

References

[1] R Killick-KendrickMedical Entomology for Students 1996[2] M ServiceMedical Entomology for Students 4th edition 2008[3] M A Tolle ldquoMosquito-borne diseasesrdquo Current Problems in

Pediatric and Adolescent Health Care vol 39 no 4 pp 97ndash1402009

[4] World Health Organization Mosquito Born Diseases WHO2016

[5] Ministry of Public Health (MOPH) ldquoAnnual Reportrdquo Bureau ofEpidemiology MOPHThailand 2014

[6] T Chaiphongpachara P Bunyuen andKK Chansukh ldquoDevel-opment of a more effective mosquito trapping box for vectorcontrolrdquo e Scientific World Journal vol 2018 Article ID6241703 8 pages 2018

[7] T Chaiphongpachara S Laojun and C KunphichayadechaldquoEffect of the CDC light trap on control of nocturnal mosqui-toes in coastal Samut Songkhram ProvinceThailandrdquoBiodiver-sitas Journal of Biological Diversity vol 19 no 5 pp 1750ndash17542018

[8] T Chaiphongpachara O Padidpoo K K Chansukh and SSumruayphol ldquoEfficacies of five edible mushroom extracts asodor baits for resting boxes to attract mosquito vectors A fieldstudy in Samut Songkhram Province Thailandrdquo Tropical Bio-medicine vol 35 no 3 pp 653ndash663 2018

[9] C Tananchai R Tisgratog W Juntarajumnong et al ldquoSpeciesdiversity and biting activity of Anopheles dirus and Anophelesbaimaii (Diptera Culicidae) in a malaria prone area of westernThailandrdquo Parasites amp Vectors vol 5 no 1 article 211 2012

[10] T Chaiphongpachara and S Sumruayphol ldquoSpecies diversityand distribution of mosquito vectors in coastal habitats ofSamut Songkhram province Thailandrdquo Tropical Biomedicinevol 34 no 3 pp 524ndash532 2017

[11] K Tainchum M Kongmee S Manguin M J Bangs and TChareonviriyaphap ldquoAnopheles species diversity and distribu-tion of the malaria vectors of Thailandrdquo Trends in Parasitologyvol 31 no 3 pp 109ndash119 2015

[12] T Chaiphongpachara S Pimsuka W S Na Ayudhaya and WWassanasompong ldquoThe application of geographic informationsystem in dengue haemorrhagic fever risk assessment in samutSongkhram province Thailandrdquo International Journal of GEO-MATE vol 12 no 30 pp 53ndash60 2017

[13] J-P Dujardin ldquoMorphometrics applied to medical entomol-ogyrdquo Infection Genetics and Evolution vol 8 no 6 pp 875ndash8902008

[14] S Sumruayphol C Apiwathnasorn J Ruangsittichai et alldquoDNA barcoding and wing morphometrics to distinguish threeAedes vectors inThailandrdquo Acta Tropica vol 159 pp 1ndash10 2016

[15] J Ruangsittichai C Apiwathnasorn and J-P Dujardin ldquoInter-specific and sexual shape variation in the filariasis vectorsMansonia dives and Ma bonneaerdquo Infection Genetics andEvolution vol 11 no 8 pp 2089ndash2094 2011

[16] C Garros and J P Dujardin ldquoGenetic and PheneticApproachesto Anopheles Systematicsrdquo in Anopheles Mosquitoes - NewInsights into Malaria Vectors pp 81ndash105 2013

[17] C Lorenz F Almeida F Almeida-Lopes et al ldquoGeometricmor-phometrics in mosquitoes What has been measuredrdquo Infec-tion Genetics and Evolution vol 54 pp 205ndash215 2017

[18] N Sontigun K L Sukontason B K Zajac et al ldquoWing mor-phometrics as a tool in species identification of forensicallyimportant blow flies of Thailandrdquo Parasites amp Vectors vol 10no 1 2017

[19] D A De Souza Y Wang O Kaftanoglu et al ldquoMorphometricidentification of queens workers intermediates in in vitroreared honey bees (Apis mellifera)rdquo PLoS ONE vol 10 no 42015

[20] S Santillan-Guayasamın A G Villacıs M J Grijalva and J-P Dujardin ldquoThe modern morphometric approach to identifyeggs of Triatominaerdquo Parasites amp Vectors vol 10 no 1 pp 1ndash102017

[21] G F Gomez E J Marquez L A Gutierrez J E Conn and MM Correa ldquoGeometric morphometric analysis of ColombianAnopheles albimanus (Diptera Culicidae) reveals significanteffect of environmental factors on wing traits and presence ofa metapopulationrdquoActa Tropica vol 135 no 1 pp 75ndash85 2014

[22] M T Motoki L Suesdek E S Bergo and M A M SallumldquoWing geometry of Anopheles darlingi Root (Diptera Culici-dae) in five major Brazilian ecoregionsrdquo Infection Genetics andEvolution vol 12 no 6 pp 1246ndash1252 2012

[23] J-PDujardin ldquoModernmorphometrics of medically importantinsectsrdquo in Genetics and Evolution of Infectious Diseases pp473ndash501 2011

[24] N Jaramillo-O J-P Dujardin D Calle-Londono and I Fon-seca-Gonzalez ldquoGeometricmorphometrics for the taxonomyof11 species of Anopheles (Nyssorhynchus) mosquitoesrdquoMedicaland Veterinary Entomology vol 29 no 1 pp 26ndash36 2015

[25] J-PDujardin D Kaba P Solano M Dupraz K D McCoy andN Jaramillo-O ldquoOutline-based morphometrics an overlookedmethod in arthropod studiesrdquo Infection Genetics and Evolu-tion vol 28 pp 704ndash714 2014

[26] R Rattanarithikul B A Harrison P Panthusiri and R E Cole-man ldquoIllustrated keys to the mosquitoes of Thailand I Back-ground geographic distribution lists of genera subgenera andspecies and a key to the generardquo Southeast Asian Journal ofTropical Medicine and Public Health vol 36 no 1 pp 1ndash802005

[27] J-P Dujardin et al ldquoThe exchangeability of shaperdquo BMCResearch Notes vol 3 no 1 2010

[28] R E Morales Vargas N Phumala-Morales T Tsunoda C Api-wathnasorn and J-P Dujardin ldquoThe phenetic structure ofAedes albopictusrdquo Infection Genetics and Evolution vol 13 no1 pp 242ndash251 2013

[29] M R Sanford B Demirci C D Marsden Y Lee A J Corneland G C Lanzaro ldquoMorphological differentiationmaymediatemate-choice between incipient species of Anopheles gambiaessrdquo PLoS ONE vol 6 no 11 2011

[30] A B BWilke R De Oliveira Christe L CMultini et al ldquoMor-phometric wing characters as a tool for mosquito identifica-tionrdquo PLoS ONE vol 11 no 8 2016

[31] P Sriwichai Y Samung S Sumruayphol et al ldquoNatural humanPlasmodium infections in major Anopheles mosquitoes in


western Thailandrdquo Parasites amp Vectors vol 9 no 1 article 172016

[32] S N Surendran P J Jude and R Ramasamy ldquoVariations insalinity tolerance of malaria vectors of the Anopheles subpictuscomplex in Sri Lanka and the implications formalaria transmis-sionrdquo Parasites amp Vectors vol 4 no 1 article no 117 2011

Hindawiwwwhindawicom

International Journal of

Volume 2018

Zoology

Hindawiwwwhindawicom Volume 2018

Anatomy Research International

PeptidesInternational Journal of



Journal of Parasitology Research

GenomicsInternational Journal of


Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018


BioinformaticsAdvances in

Marine BiologyJournal of



Neuroscience Journal


BioMed Research International

Cell BiologyInternational Journal of



Biochemistry Research International

ArchaeaHindawiwwwhindawicom Volume 2018


Genetics Research International


Advances in

Virolog y Stem Cells International



Enzyme Research



MicrobiologyHindawiwwwhindawicom

Nucleic AcidsJournal of

Volume 2018

Submit your manuscripts atwwwhindawicom


it is more difficult to obtain a mosquito sample in the fieldthan in the laboratory because the external characteristicsare often damaged during trapping and transportation [14]One solution to this is the use of high-efficiency moleculartechniques for mosquito identification However these havevery high associated costs [14 15] making them unsuitablefor use in the field where many samples are collected Thus anew technique for identifying mosquito vectors in the field isrequired

Geometric morphometrics (GM) is a fast inexpensivetechnique [14 16] used to analyze the size and shape of indi-viduals based on a suite of traits GM has been appliedwidely in a number of fields including entomology (egmos-quito [17] blow fly [18] bee [19] and eggs of Triatominae[20]) Furthermore several studies have used theGMmethodto classify species and to examine variation among medi-cally important mosquitoes that are morphologically sim-ilar or sibling species [14 15 21ndash24] GM analyses can beconducted using landmark- or outline-based methods [13]each of which has different advantages depending on thecharacteristics and specificity of the sample The landmark-based approach uses the coordinates of landmarks to analyzethe morphology [23] and is popular in the field of medicalentomology with many studies having shown it can distin-guish different species of mosquito vectors successfully suchas Anopheles spp and Aedes spp [14 21 23] By contrastthe outline-based approach uses contour data [25] and hasalso been shown to be useful for identifying some mosquitospecies such as Aedes scutellaris [14] However surprisinglythe specificity with which mosquito vectors can be identifieddiffers between the two techniques [14 23] and thereforea comparison of their performance in the field is requiredbefore they can be used

Thus the aim of this study was to compare the efficiencyof landmark- and outline-based GM techniques to separatespecies of mosquito vectors in Huay Nam Nak village inRatchaburi Province Thailand an endemic area for mosqui-to-borne diseases The result from this research will serve asa guideline for choosing the best GM technique for the iden-tification of vectors in the field in Thailand to facilitate thecontrol of vector-borne diseases

2 Materials and Methods

21 Study Site and Mosquito Collection Mosquitoes werecollected from Huay Nam Nak village (13∘22101584036010158401015840N99∘16101584034910158401015840E) in the Suan Phueng District of Ratchaburi Pro-vince Thailand during June to August 2015 (Figure 1) Thestudy site consistedmainly of woodenhouses and agriculturalfields with streams flowing through and mountains and hillyforests surrounding the village

Three Mosquito Magnet Independence Mosquito Traps(Woodstream Corporation USA) were placed approximately5 meters away from houses in the village for 24 hours (600am to 600 am of the following day) per week over the 3-month study period to collect nocturnal and diurnal mos-quito vectors Mosquito samples were collected from eachtrap in the morning (600 am) and transported to thelaboratory of the College of Allied Health Science Suan

Sunandha Rajabhat University Samut Songkhram Provincefor morphological identification Female mosquitoes werethen morphologically identified to species using the illus-trated keys to the mosquitoes of Thailand [26]

22 Mosquito Wing Preparation Only the right wings offemale mosquitoes were analyzed by GM The right wingof each individual was dissected under a Nikon AZ 100Mstereomicroscope (Nikon Corp Tokyo Japan) and placedbetween a glass microscope slide and coverslip using Hoyerrsquossolution as amountingmediumThen all of thewing samplesfor each species were photographed using a Nikon DS-Ri1 SIGHT digital camera connected to a Nikon EclipseE600 microscope (Nikon Corp Tokyo Japan) under 40timesmagnification alongside a 1-mm scale bar and were analyzedusing the Collecting Landmarks for Identification and Char-acterization (CLIC) Program Since the aim of this study wasto compare the effectiveness of landmark- and outline-basedGM for species discrimination the same set of wing pictureswas used for each species in the two analyses outlined below

23 Landmark-Based GM Analysis A total of 17 landmarkswere selected based on the ease with which they could beplotted across all mosquito species and their low likelihood ofbeing damaged eg by wing vein intersection The positionsof these landmarks on the wing of each individual weredigitized (Figure 2) The wing size was then computed as thecentroid size (CS) which is the square root of the sum of thesquared distances between each individual landmark and thecenter of the landmark configuration [14 27]Thewing shapevariables (partial warps [PW]) were computed as principalcomponents (PCs) of the PW(known as relative warps [RW])after Generalised Procrustes Analysis (GPA) (the statisticalprocedure of superimposition) and discriminant analysis(DA) (or Canonical Variate Analysis [CVA]) was then used toanalyze the shape variables for the separation of each species

24 Outline-BasedGMAnalysis Theoutline used for analysiswas constructed using coordinates along the contour of thelower part of the mosquito wing (Figure 3) This analyticalcontour is the most complete part of the wing in all mosqui-toes and also lacks thick scales and is difficult to tear Thewing size was estimated as the length of the perimeter of thecontour which was separate from the set of shape variablesusedThe shape variables were constructed using the normal-ized elliptic Fourier coefficients (NEF)

25 Statistical Analysis Prior to conducting theGManalysesthis research assessed the quality and measurement errorof the digitized landmarks and digitized contour used forthe landmark-based and outline-based GM analyses respec-tively by calculating the repeatability index (R)The repeata-bility of each coordinate was tested for 70 randomly selectedwings (10 wings per species) In order to do this measure-ments of the coordinates used in the original wing pho-tographs were repeated to give a total of 140 images The re-peatability was then computed based on an ANOVA designPrior to analysis the error was reduced by averaging the twodigitizations [23 25]


N

Thailand


km

0 200100


12

3

4

5

6

7

8

9

10

11

12

13

1415

16

17

m






m





3 Results













16 19 22 25 28 30 33 36 39

(a)


P 25 P 75

268237 298 328 358 388 418 449 479


(b)









4 Discussion





1

23

4

5

678 9

10

1112

13 14

15

16

17




















Cano

nica

l var

iabl

e 2

568

713

minus436minus49

2

(a)


Cano

nica

l var

iabl

e 2


175

5

013 1287

minus65

1

(b)











6

5

4

3

2

1


(a)


6

5

4

3

2

1


(b)




5 Conclusions


Data Availability





Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018



N

Thailand


km

0 200100


12

3

4

5

6

7

8

9

10

11

12

13

1415

16

17

m






m





3 Results













16 19 22 25 28 30 33 36 39

(a)


P 25 P 75

268237 298 328 358 388 418 449 479


(b)









4 Discussion





1

23

4

5

678 9

10

1112

13 14

15

16

17




















Cano

nica

l var

iabl

e 2

568

713

minus436minus49

2

(a)


Cano

nica

l var

iabl

e 2


175

5

013 1287

minus65

1

(b)











6

5

4

3

2

1


(a)


6

5

4

3

2

1


(b)




5 Conclusions


Data Availability





Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018



m





3 Results













16 19 22 25 28 30 33 36 39

(a)


P 25 P 75

268237 298 328 358 388 418 449 479


(b)









4 Discussion





1

23

4

5

678 9

10

1112

13 14

15

16

17




















Cano

nica

l var

iabl

e 2

568

713

minus436minus49

2

(a)


Cano

nica

l var

iabl

e 2


175

5

013 1287

minus65

1

(b)











6

5

4

3

2

1


(a)


6

5

4

3

2

1


(b)




5 Conclusions


Data Availability





Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018





16 19 22 25 28 30 33 36 39

(a)


P 25 P 75

268237 298 328 358 388 418 449 479


(b)









4 Discussion





1

23

4

5

678 9

10

1112

13 14

15

16

17




















Cano

nica

l var

iabl

e 2

568

713

minus436minus49

2

(a)


Cano

nica

l var

iabl

e 2


175

5

013 1287

minus65

1

(b)











6

5

4

3

2

1


(a)


6

5

4

3

2

1


(b)




5 Conclusions


Data Availability





Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018




1

23

4

5

678 9

10

1112

13 14

15

16

17




















Cano

nica

l var

iabl

e 2

568

713

minus436minus49

2

(a)


Cano

nica

l var

iabl

e 2


175

5

013 1287

minus65

1

(b)











6

5

4

3

2

1


(a)


6

5

4

3

2

1


(b)




5 Conclusions


Data Availability





Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018





Cano

nica

l var

iabl

e 2

568

713

minus436minus49

2

(a)


Cano

nica

l var

iabl

e 2


175

5

013 1287

minus65

1

(b)











6

5

4

3

2

1


(a)


6

5

4

3

2

1


(b)




5 Conclusions


Data Availability





Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018




6

5

4

3

2

1


(a)


6

5

4

3

2

1


(b)




5 Conclusions


Data Availability





Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018



Acknowledgments


References




































Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018







Volume 2018

Zoology











Volume 2018

















Advances in




Enzyme Research





Volume 2018


Comparison of Landmark- and Outline-Based Geometric...

Documents

Transcript of Comparison of Landmark- and Outline-Based Geometric...