Stephens Juliano 12 JME Wing Shape

8/11/2019 Stephens Juliano 12 JME Wing Shape

1/12

VECTOR-BORNEDISEASES, SURVEILLANCE, PREVENTION

Wing Shape as an Indicator of Larval Rearing Conditions forAedesalbopictusandAedes aegypti(Diptera: Culicidae)

C. R. STEPHENS ANDS. A. JULIANO1

School of Biological Sciences, Illinois State University, Normal, IL 61761

J. Med. Entomol. 49(4): 927938 (2012); DOI: http://dx.doi.org/10.1603/ME12012

ABSTRACT Estimating a mosquitos vector competence, or likelihood of transmitting disease, if ittakes an infectious bloodmeal, is an important aspect of predicting when and where outbreaks ofinfectious diseases will occur. Vector competence can be affected by rearing temperature and inter-and intraspecic competition experienced by the individual mosquito during its larval development.This research investigates whether a new morphological indicator of larval rearing conditions, wingshape, can be used to distinguish reliably temperature and competitive conditions experienced duringlarval stages.Aedes albopictus (Skuse) andAedes aegypti (L.) (Diptera: Culicidae) larvae were rearedin lowintraspecic, highintraspecic, or highinterspecic competition treatments at either 22 or 32C.The right wing of each dried female was removed and photographed. Nineteen landmarks and 20semilandmarks were digitized on each wing. Shape variables were calculated using geometric mor-phometric software. Canonical variate analysis, randomization multivariate analysis of variance, andvisualization of landmark movement using deformation grids provided evidence that althoughsemilandmark position was signicantly affected by larval competition and temperature for bothspecies, the differences in position did not translate into differences in wing shape, as shown indeformation grids. Twoclassicationproceduresyielded successrates of 26 49%. Accounting forwingsize produced no increase in classication success. There seemed to be a signicant relationshipbetween shape and size. These results, particularly the low success rate of classication based on wingshape, show that shape is unlikely to be a reliable indicator of larval rearing competition and

temperature conditions forAe.albopictusandAe.aegypti.

KEY WORDS geometric morphometrics, classication, vector competence, larval competition

The likelihood that a mosquito, or any vector, willshow disseminated infection after taking an infectiousbloodmeal, and thus be able to transmit disease, is itsvector competence (Hardy et al. 1983, Eldridge andEdman 2000). A mosquitosvector competencecan beinuenced by its environment, including the larval

rearing environment. For example, inter- and intras-pecic competition among larval Aedes albopictus(Skuse) andAedes aegypti(L.) (Diptera: Culicidae)produced adultAe. albopictus with signicantly higherrates of infection and dissemination with Sindbis virus(i.e., higher vector competence; Alto et al. 2005). Ef-fects onAe. aegyptiwere similar, although not signif-icant. Alto et al. (2008) found similar effects of larvalcompetition on vector competence fordenguevirus in

Ae. albopictus and Ae. aegypti. Greater competitivestress during larval stages produced adultAe. albopic-tusthat were signicantly more competent as dengue

vectors, with both a higher proportion of infectedindividuals as well as a higher proportion with dis-seminated infections compared with those from lowcompetition conditions, with no difference between

inter- and intraspecic competition treatments. ForAe. aegypti,the results were not signicant, but theyagain showed a trend similar to that ofAe. albopictus,with high competition treatments leading to increaseddengue infection and dissemination rates. Increasedinterspecic larval competition with Ae. albopictusalso has been shown to produce adultAedes triseriatus(Say) that show increased rates of infection and dis-semination for LaCrosse virus (Bevins 2008).

Vector competence is one important component ofa mosquito populations vectorial capacity; vectorcompetence is the average rate at which infected bitesoccur after an infected host has been introduced (An-derson and Rico-Hesse 2006). It would therefore bedesirable to have a way of sampling a natural popu-lation of adult mosquitoes and accurately estimatingeach individuals vector competence as a step toward

estimating vectorial capacity. Some indicator of thelarval competition conditions under which an adultmosquito developed would be one contributor to es-timates of vector competence. Such a measure couldhelp to predict when and where outbreaks of vector-borne illness might occur.1 Corresponding author, e-mail: [email protected].

0022-2585/12/09270938$04.00/0 2012 Entomological Society of America


2/12

The most straightforward index of an adult mosqui-tos larval competition environment would be adultsize; the negative relationship between larval compe-tition and adult size is well known (Livdahl 1982;Agnew et al. 2000; Gimnig et al. 2002; Alto et al. 2005,2008; Jirakanjanakit et al. 2007). However, other fac-tors in the larval rearing environment, most notably

temperature, also affect the relationship between sizeand competition. Larval rearing temperature has aninverse relationship with size for various mosquitospecies (Rueda 1990, Lyimo et al. 1992, Tun-Lin et al.2000, Alto and Juliano 2001). Furthermore, larval rear-ing temperature is inversely related to adult infectionrate for both Rift Valley Fever and Venezuelan equineencephalitis (Turell 1993) and inversely related toadult body titer of Chikungunya virus (Westbrook2010). This means that an adult mosquito may be largebecause of low competition (yielding lower vectorcompetence) or low temperature (yielding greater

vector competence), or it may be small because ofhigh competition (yielding greater vector compe-tence) or high temperature (yielding lower vectorcompetence). Other stressors acting on larvae (e.g.,pesticide exposure) also can alter adult size and mayaffect vector competence (Muturi et al. 2011). There-fore, a method that can distinguish both an adult mos-quitos larval competition environment as well as itslarval rearing temperature would be desirable. Here,we evaluate geometric morphometric analysis of wingshape and size as a way to classify adult mosquitoesinto their larval rearing conditions, testing specically

whether wing shape gives a reliable classication.Geometric morphometrics (Bookstein 1991, 1996a,b;Zelditch et al. 2004) is a method of analyzing shape thataccountsforthespatialrelationships among thevariablesin the analysis, which are landmark points on the struc-ture in question (Rohlf 1999). As such, geometric mor-phometrics retains the geometry of the morphologicalstructure(Adams et al. 2004). Landmarks are plotted onimages of the object, resulting in sets of coordinates foreach individual structure, that can then be used in avariety of statistical methods to determine whether be-tween-groupdifferencesexist. Using geometric morpho-metrics also allows for the visualization of shape change

between groups and can aid in the interpretation ofresults in ways that are not possible using traditionalmorphometrics (Rohlf 1999). In addition, this method isrelatively easy andinexpensiveandcould thus bereadilyused in the eld. Here, for two species ofAedes, wedetermine whether wing shape can be used to discrim-inate among adults that developedunder different larvalcompetition and temperature conditions. If this can bedone, wing shape could be used as an indicator of vectorcompetence for eld-collected individual adults.

StudySpecies.Ae. albopictus isnativetoAsia and theislands of the Western Pacic and Indian Oceans, but

it has recently been introducedinto Africa, the MiddleEast, Europe and North and South America (Gratz2004). Ae. albopictus can transmit at least 22 arbovi-ruses (Moore and Mitchell 1997) and is second only to

Ae. aegypti in its importance as a vector of dengue(Benedict et al. 2007).Ae. albopictusprobably acts as

a maintenance vector of dengue in rural areas ofSoutheast Asia, where dengue epidemics often occur(Gratz 2004).Ae. albopictusis also important becauseit is an aggressive diurnal biter and could serve as abridge vector for various viruses (Moore and Mitchell1997).

Ae. aegypti is native to Africa, but it is now found

throughout most tropical and subtropical regions(Tabachnick and Powell 1979, Kamgang et al. 2011).Its geographic distribution and population densitieshave increased because of uncontrolled urbanizationcreating more larval habitat in the form of articialcontainers (Gubler 1998). Ae. aegypti is an efcientvector because it bites during the day and spends itsentire life around human habitation (Morrison et al.2008). Furthermore,Ae. aegypti often feeds on severalpeople to acquire a full bloodmeal (Gubler 1998).Ae.aegypti is the most competent, and most important,vector of dengue virus (Rueda et al. 1990) and it can

transmit yellow fever, chikungunya virus, and otherarboviruses (Morrison et al. 2008).

Materials and Methods

Rearing Larvae and Wing Preparation.Ae. albopic-tusandAe. aegyptifrom Tampa, FL, colonies (threesix generations in the laboratory) were reared fromeggs in plastic containers containing 200 ml of water,1.0gofliveoak(Quercus virginianaMill.) leaves,0.01 gof dried crickets [Gryllodes sigillatus(Walker)], and1 l of tree hole water inoculum. Each container was

randomly assigned to one of ve competition treat-ments and one of two temperature treatments, chosenbased on a preliminary run of this experiment. Thecompetition treatments (described by initial numbersofAe. albopictus:Ae. aegypti rst-instar larvae) were15:0, 30:0, 15:15, 0:30, and 0:15, and the temperaturetreatments were 22 and 32C. There were 10 replicatesof each treatment combination. Containers were keptin temperature-controlled environmental chamberswith a photoperiod of 14:10 (L:D) h until all individ-uals eclosed as adults or died. Adults were identiedfor sex and species and then dried at 50C for 24 h.Work with mosquitoes was done under protocols 01-

2010 and 02-2007 of the Institutional Animal Care andUse Committee.

The right wings were gently removed from adultfemales, for 248 Ae. albopictus and 143 Ae. aegyptiwings in total. If the right wing was damaged, the leftwas used. If both wings were damaged, the female wasexcluded from the study. The wings were then pho-tographed with a Sony Power HAD image capturingsystem and loaded into TPSdig version 2.14 (Rohlf2009) for landmark digitization. Nineteen landmarksand 20 semilandmarks were plotted on images of themosquito wings. Landmarks were placed at intersec-

tions of wing veins or intersections of veins and thewing margin (Fig. 1). Semilandmarks were placedalong the posterior edge of the wing (Fig. 1).Semilandmarks are placed along curves that do nothave features that could otherwise be designated withtraditional landmarks (Zelditch et al. 2004). The po-

928 JOURNAL OFMEDICALENTOMOLOGY Vol. 49, no. 4


3/12

sition of an individual semilandmark is only informa-tive relative to the positions of the surroundingsemilandmarks, so the semilandmarks must be inter-

preted as a group. The arrangement of the group ofsemilandmarks can show changes in the bowing of thecurve and nothing more. This procedure resulted in aset of landmark coordinates for each wing. Landmarkcoordinate data were transformed into IntegratedMorphometrics Package (IMP) les by using Coord-Gen (Sheets 2006a), semilandmark positions were op-timized using TPSrelw version 1.46 (Rohlf 2008), andcentroid sizes (square root ofthe sum of the squareddistances between each landmark and its centroid, orcenter of the conguration of landmarks) for all in-dividuals were calculated. The combined landmarkdata then had the nonshape variation removed using

the generalized least squares Procrustes superimpo-sition method (Bookstein 1991) to translate, scale, androtate each individuals conguration against the ref-erence (average) conguration. The resulting co-ordinates were then used to calculate partial and uni-form warp scores, which describe to what extent, andin what direction, the reference conguration wouldhave to change to match exactly the shape of thatindividual (Zelditch et al. 2004). These partial warpscores were the measures of shape that were used asvariables in many of the subsequent analyses.

Effects of Competition. First, a preliminary analysis

tested whether the competition treatments stressedthe larvae. If competition and temperature affect life-history traits such as survivorship, time to eclosion,and wing size, then it would be meaningful to test foreffectsofcompetition andtemperatureonwing shape.If the competition treatments were effective, one

would expect a decrease in survivorship, an increasein median days to eclosion, and a decrease in wing sizewith increased inter- or intraspecic density (Livdahl

1982, Russell 1986, Hard et al. 1989). For each species,treatment effects on these three variables were testedusing analysis of variance (ANOVA). Because com-petition signicantly affected these life-history traits(Table 1), analyses on wing shape and size were thenperformed.

Morphometric Differences between TreatmentGroups. Partial warp scores were used in canonicalvariate analyses (CVAs) to describe differences inshape among treatment groups.CVAs were performedusing CVAGen (Sheets 2006b), including a prelimi-nary analysis to determine whether shape differedbetween the two species regardless of treatment. Be-

cause of software limitations, all CVAs were donewithout the inclusion of centroid size as a dependentvariable. CVAGen also was used to visualize differ-ences in landmark and semilandmark position be-tween groups.

Because a CVA cannot test for statistically signi-cant differences between groups, a multivariateANOVA (MANOVA) also must be performed(Zelditch et al. 2004). For standard MANOVAs onshape data, the appropriate number of degrees offreedom must be calculated, but there is no bestmethod of doing so when semilandmarks are used

(Zelditch et al. 2004). To circumvent this problem,randomization MANOVAs were used. These do notrequire the calculation of new degrees of freedom.This process shufes the observations and randomlyassigns them to different temperature and competi-tion treatments. The shufing procedure is repeated

Fig. 1. Landmarks plotted on right wings fromAe. albopictus andAe. aegypti females. Landmarks 119 (white) aretraditional landmarks, and 2039 (black) are semilandmarks along the posterior edge.

July 2012 STEPHENS ANDJULIANO: WINGSHAPE ANDAedesREARINGCONDITIONS 929


4/12

many times, producing hundreds or thousands of ran-domized data sets (permutation). Then, a MANOVAis performed on each random permutation, as well as

on the original data. For each effect in the model(temperature, competition, and interaction, in thiscase), if 5% of the permutations yield a P valuesmaller than that obtained for the original data set, theeffect is deemed signicant. Even though this methoddoes not analyze every possible permutation of thedata set, it does take a large, random sample of allpossible permutations while remaining computation-ally feasible. The randomization MANOVA was per-formedusing SAS9.2(SAS Institute 2008). Themacrosfor randomization and analysis processes were mod-ied from a randomization wrapper for ANOVA (Cas-

sell 2002). Each randomization test created 1,000 per-mutations. A preliminary randomization MANOVAwas done to determine whether the species differed inshape overall, and subsequent analyses were run foreach species both with and without centroid size.

Classification. To determinewhether shape andsizevariables can be used to discriminate among individ-uals based on rearing conditions (i.e., to predict theconditions from which an individual came), two clas-sication procedures were performed. The rst was a

jackknife classication procedure that accompaniedeach CVA and was performed with CVAGen, whichwas run without centroid size. In this analysis, each

observation was held out of the data set to develop thediscriminant function, and then classied based on theresulting function. Percent correct assignment wasthen assessed. The second procedure used the dis-criminant function procedure (PROC DISCRIM) inSAS 9.2 (SAS Institute 2008) to create linear discrim-ination functions and to run cross-validation. Discrim-inant function analysis in SAS was run both with andwithout centroid size.

Within each procedure, individuals were classiedinto two sets of treatment conditions. The rst runclassied individuals into temperature and competi-

tion treatments. However, it is possible that only themagnitude of competition (i.e., total larval density) isimportant, rather than type of competition (i.e., inter-versus intraspecic competition). Therefore, the sec-ond run classied individuals into temperature anddensity treatments. In this second run, the high intra-

and high interspecic groups were termed the high-density treatment, and the low intraspecic treat-ment was thelow-densitytreatment.

Relationship of Shape to Centroid Size. Multivariateanalyses of covariance (MANCOVAs) were per-formed using SAS 9.2 (PROC GLM) for each speciesto determine the relationship between size and shapefor the treatment groups. First, we tested whether therelationship between size and multivariate shape wasthe same for all treatment groups. We then deter-mined whether size was signicantly related to mul-tivariate shape. We also compared the results of thedifferent classication procedures to determinewhether including size resulted in better discrimina-tion of individual adults for larval competition and

temperature conditions.

Results

Effects of Competition.The interaction of compe-tition level and temperature was marginally nonsig-nicant forAe. albopictussurvivorship (Table 1; Fig.2a). For both species, the effect of competition treat-ment was signicant (Table 1; Fig. 2a and b). ForAe.albopictus in both temperatures, low intraspeciccompetition resulted in the greatest survivorship, fol-lowed by the high interspecic and then high intras-

pecic treatments (Fig. 2a). InAe. aegypti,only com-petition affected survivorship (Table 1), with lowintraspecic competition yielding the greatest survi-vorship, and both high intra- and high interspecictreatments produced equally low levels of survivor-ship (Fig. 2b).

Median days to eclosion for Ae. albopictuswas sig-nicantly affected by competition and temperature,but not the interaction (Table 1). Increased compe-tition resulted in an increase in median days to eclo-sion (Fig. 3a). Median days decreased from 33.8 1.182 at 22C to 25.75 1.182 at 32C. ForAe. aegypti,

theinteraction between competition andtemperaturewas signicant (Table 1). In both temperatures, highinter- and high intraspecic competition resulted in alonger time to eclosion than the low intraspecictreatment, and this trend was signicant at 22C butnot at 32C (Fig. 3b).

Table 1. ANOVA results for effects of competition and temperature on survivorship, median days to eclosion, and centroid size

Effect df Survivorship Median days Centroid size

F P F P F P

Ae. albopictusCompetition 2 47.14


5/12

Centroid size for Ae. albopictus was signicantlyaffected only by temperature (22C, 728.34 7.412;32C, 695.38 7.412). There was a signicant inter-action forAe. aegypti (Table 1). At 22C, both the lowintraspecic and high interspecic treatments pro-duced larger females than the high intraspecic treat-ment (Fig. 4), although this trend was not signicant.Even though increased competition (increased larval

density) is expected to produce smaller adults, thehigh inter- and intraspecic treatments at 32C pro-duced larger females than the low intraspecic treat-ment (Fig. 4), although the trend was again not sig-nicant. This result further illustrates that size cannotbe used to determine larval rearing conditions. Therelationship between size and rearing conditions dif-fered between the two species, and the relationshipwas either counter-intuitive or nonexistent. Overall,the competition treatments were successful at stress-ing the larvae, butthe phenotypic responses to densityand temperature were complex and multifaceted.

Morphometric Differences Between TreatmentGroups. The two species were well separated viashape regardless of treatment (Fig. 5). The CVAyielded a distinct separation of two clouds of pointsalong the rst CVA axis, suggesting that some aspectof wing shape strongly separatesAe. albopictus andAe.

aegypti.The difference in shape between the averageAe. albopictusand the average Ae. aegyptiwing (Fig.6) indicates that most of the difference occurs in thepositions of the semilandmarks along the posterioredge of the wing, with some less pronounced differ-ences in the center of the wing. The accompanyingrandomization MANOVA yielded a highly signicantspecies effect (P 0.001). A jackknife classication

was able to assign 96.16% of the individuals to thecorrect species. Because the consistent differences inlandmark position indicate the two species have dif-ferent wing shapes, subsequent morphometric analy-ses were done for each species separately.Ae. albopictus. The canonical variate analysis for

competition and temperature (Fig. 7) resulted in onesignicant axis (2370 591.81, P 0.0001), alongwhich the two temperature groups were reasonablywell separated. The difference in landmark positionsbetween the average high and average low tempera-ture wings(data not shown) comes primarily from the

semilandmarks shifting noticeably along the posterioredge. Randomization MANOVA conrmed a highlysignicant temperature effect (P 0.001), a margin-ally nonsignicant competition effect (P 0.064), anda nonsignicant interaction (P 0.572) for shapevariables.

Fig. 2. Percentage of survival (average SE) for each competition treatment forAe.albopictusby temperature (a) andAe. aegypti (across both temperatures; b). Filled squares in panel a represent the 22C treatment, and open squares representthe 32C treatment. Means with different letters are signicantly different.



6/12

Contrasts were performed within eachtemperatureto determine whether shape could be used to distin-guish between high inter- and high intraspecic com-petition independent of temperature (sequential Bon-ferroni e 0.05). Within both temperatures, the twotreatments were not signicantly different (22C:F74,169 1.04,P 0.4076 and 32C:F74,169 1.08,P0.3337). This result agreed with the visualizations of

shape differences between the corresponding com-

petition groupsin eachtemperature(data notshown).In both cases, the only difference between the groupswas a slight shift of the semilandmarks along theircurve.

Because shape did not differ between the high in-ter- and high intraspecic competition treatments,contrasts were used to determine whether shape dif-fered solely because of density (i.e., magnitude of

competition, regardless of species composition) by

Fig. 3. Median days to eclosion (average SE) for each competition treatment for Ae. albopictus (across bothtemperatures; a) andAe.aegyptiby temperature (b). Filled circles in panel b represent the 22C treatment, and open circlesrepresent the 32C treatment. Means with different letters are signicantly different.

Fig. 4. Centroidsize(average SE) forAe. aegypti for all competition treatments by temperature.Filled circles representthe 22C treatment, and open circles represent the 32C treatment.



7/12

comparing high- and low-density groups within eachtemperature. For both temperatures, with correctionfor multiple tests, the high and low densities did notdiffer in shape (22C:F74,169 0.96, P 0.5662 and32C: F74,169 1.37, P 0.0492). Again, the onlychange between the average treatment shapes werethe shifts of semilandmarks along the curve (data notshown).

Randomization MANOVA, including centroid sizeamong the dependent variables, tested whether somecombination of size and shape differed among treat-ments. Inclusion of centroid size produced resultssimilar to those found excluding centroid size. Tem-peratureremained highly signicant (P0.001),com-petition marginally signicant (P 0.046), and theinteraction remained nonsignicant (P 0.5750). Asecond set of contrasts including centroid size yieldedthe same conclusions as the rst set. The high inter-andintraspecic competition treatments didnot differ

at either temperature (22C:F75,167 1.03,P 0.4250and 32C:F75,167 1.08, P 0.3445). There was alsono difference between high and low density at eithertemperature (22C: F75,167 1.04, P 0.4061and 32C:F75,167 1.34, P 0.0612).Ae.aegypti.Canonical variate analysis yielded two

signicant axes (2370 514.39, P0.0001 and2292

350.93,P 0.010). The rst corresponded to temper-ature and the second roughly to competition treat-ment (Fig. 8). There seemed to be an interaction,because the score on the competition axis for eachtreatment depended on the temperature treatment(Fig. 8). The randomization MANOVA conrmed asignicant temperature by competition interaction(P 0.046), so contrasts were performed to seewhichgroups differed in shape. Contrasts comparing highinter- and high intraspecic competition treatmentswithin each temperature were not signicant (22C:F74,64 1.15, P 0.2881 and 32C:F74,64 1.28, P0.1576). Only the semilandmarks slid along the pos-terior edge when transitioning between the two com-

petition treatments for both temperatures (data notshown), although the movement was very slight. Con-trasts between high and low density for each temper-ature indicated that shape was signicantly differentbetween the two densities in the 32C treatment(F74,64 1.85;P 0.0065), but not for the 22C treat-ment (F74,64 1.01, P 0.4906). Despite being sta-tistically signicant, the movement at 32C was con-ned to slight movement of the semilandmarks alongthe curve; semilandmark shifts were even smaller at22C (data not shown).

MANOVA and contrasts including centroid size

with shape variables yielded conclusions identical tothe analysis of shape alone. Temperature (P 0.001),competition (P 0.0110), and the interaction (P0.0330) were all signicant. High inter- and intraspe-cic competition treatments were not signicantlydifferent at either temperature (22C:F75,63 1.12,

Fig. 5. CVA plot of species effect on shape. One axis was found to be signicant (P 0.0001). Filled circles representAe.albopictusand open circles representAe.aegypti.

Fig. 6. Deformation gridof shape differencesbetweenanaverage Ae. albopictus and Ae. aegypti wing. The changeshave been exaggerated 250 for easier viewing.



8/12

P 0.3263 and 32C:F75,63 1.33, P 0.1214) andhigh- and low-density treatments were signicantlydifferent at 32C (F75,63 1.83;P 0.0072) but not at22C (F75,63 0.99; P 0.5265).

Classification. The results of the jackknife classi-

cation procedure (Table 2) show that frequency ofcorrect classication was consistently low for bothspecies, although there is a moderate increase foreach when individuals are classied into tempera-

ture and density conditions instead of temperatureand competition conditions. The results of the dis-crimination and cross-validation approach (Table2) are very similar to the jackknife procedure. Forboth species, success increased when classifying

into temperature and density conditions. However,the effect of adding centroid size was different forthe two species. Including centroid size slightlylowered the success of the procedure for Ae. al-

Fig. 7. CVA plot forAe.albopictus, with individuals grouped by temperature and competition treatments. One axis wasfound to be signicant (P 0.0001). Filled shapes represent 22C treatments and open shapes represent 32C treatments.Circles represent low intra-specic competition treatments (15:0 [albopictus:aegypti]), squares represent high interspecictreatments (15:15), and triangles represent high intraspecic treatments (30:0).

Fig. 8. CVA plot forAe.aegypti, with individuals grouped by temperature and competition treatments. Two axes weresignicant (P 0.0001 andP 0.010). Filled shapes represent 22C treatments and open shapes represent 32C treatments.Circles represent low intraspecic competition treatments (0:15 [albopictus:aegypti]), squares represent high interspecictreatments (15:15), and triangles represent high intraspecic treatments (0:30).



9/12

bopictus,but it produced a small increase in successfor Ae. aegypti (Table 2).

Relationship of Shape to Centroid Size. For bothspecies multivariate analysis of shape showed nonsig-nicant interactions of temperature, competition, ortemperature competition with centroid size (P0.05; data not shown), indicating homogeneous slopesso that MANCOVA is appropriate.MANCOVA (Table3) showed a signicant relationship between centroidsize and shape. However, these results should be in-terpreted with caution. The df for these tests have notbeen adjusted to account for the inclusion ofsemilandmarks in the analysis (Zelditch et al. 2004).This means that these tests are more likely to nd therelationship between size and shape signicant than atest with the appropriate df. MANCOVA showed sig-

nicant effects of temperature, competition, and theirinteraction on wing shape inAe. aegypti, whereas onlytemperature was signicant forAe. albopictus(Table3). Even though the relationship was signicant, itdoes not seem that accounting for size is benecialwhen classifying adults into their larval rearing con-ditions. Including centroid size did not produce ameaningful increase in successful classication rate(Table 2), and because successful classication is themain goal of this study, we do not look further into therelationship between size and shape.

Discussion

We observed signicant differences in shape vari-ables between the two temperature treatments for

both species, and between the high- and low-densitytreatments at 32C forAe. aegypti.However, the clas-sication success remained very low regardless of theprocedure used. To understand these results, it is im-portant to remember what each part of the analysiscan tell us. The randomization MANOVA found dif-ferences in shape variables that arise because of sig-nicant shifts in landmark position. The randomiza-tion MANOVA cannot determine which landmarks orsemilandmarks moved or the manner in which theymoved. To understand the nature of the movement, it

is necessary to visualize changes in landmark positionusing deformation grids. In each grid, the majority,if not all, of the movement of landmarks andsemilandmarks was among the semilandmarks alongthe posterior edge of the wing.

Semilandmarks as a set can only provide informa-tion about the bowing of the curve that they delineate(Zelditch et al. 2004). If the bowing changes, that is,if the curve bends more or less, between groups, thenthe shape of the conguration has changed in thatarea. If we look more closely at the way in which thesemilandmarks in the current study move, we see thatthey shift along the curve and not perpendicular to it

in either direction. This means that the movement ofthe semilandmarks is not describing a change in theshape of the curve, a problem arising from the com-mon practice of equidistant spacing of semilandmarks(Gunz et al. 2005). Consider two hypothetical andidentical wings shown in Fig. 9. The semilandmarksalong the posterior edge are merely arranged differ-ently along the same curve in both wings. This is thekind of movement that was deemed signicant usingthe randomization MANOVA. Although the positionsof the semilandmarks changed going from high to lowtemperature, or from high to low density, the actual

shape of the wing did not. Thus, competition andtemperature seem to have no effect on actual wingshape, leading to very low classication success. Withan error rate from 50 to 75%, this method would notgive investigators condence that larval rearing con-ditions have been correctly identied.

Table 2. Results of the CVA jackknife and discriminant func-

tion classification procedures

Grouping Ae. albopictus Ae. aegypti

Jackknife procedureTemp, competition 27.82 26.57Temp, density 46.77 39.86

Discriminant analysisCentroid size excluded

Temp, competition 31.05 27.97Temp, density 49.60 39.16

Centroid size includedTemp, competition 30.36 32.17Temp, density 47.77 46.85

Values are the percentage of individuals correctly assigned to theirlarval rearing conditions (temperature and competition or density).

Table 3. MANCOVA results for the effects of temperature,

competition, and centroid size on shape

Effect Pillais trace F df P

Ae. albopictusTemp 0.716 5.68 74,167


10/12

If all of the analyses were redone excluding thesemilandmarks, the resultsare unlikely to change. Theinclusion of semilandmarks does not obscure move-ment of traditional landmarks (Zelditch et al. 2004).They serve as an additional way to capture shapechange separate from landmarks, particularly for mor-phological areas where landmarks are absent. If there

was no movement of traditional landmarksin the pres-ence of semilandmarks, there should be no movementof traditional landmarks excluding semilandmarks.

The low classication success was not improved bythe inclusion of centroid size in the analysis. For bothprocedures where centroid size was included, suc-cessful classication rate increased only slightly oreven decreased compared with excluding centroidsize. Even though the relationship between centroidsize and shape is signicant, reliable classication isthe main goal of this study and accounting for size wasultimately not benecial.

Although our ndings were not useful for discrim-inating among larval rearing conditions, they are con-sistent with previous work on effectsof environmentalvariables on insect wing shape. For Ae. aegypti, Mo-rales Vargas et al. (2010) found that altering larvaltemperature produced no signicant changes in adultwing shape. In Drosophila birchii Dobzhansky &Mather, the latitude at which populations were foundalso had no effect on wing shape (Grifths et al. 2005).However, the results from the study are inconsistentwith other work on the effect of larval rearing tem-perature on adult wing shape for both mosquitoes and

Drosophila. In Anopheles superpictus Grassi, differ-ences in wing shape caused by larval temperaturewere quite large (Aytekin et al. 2009), whereas thedifferences in our study were only in semilandmarkposition and not in overall shape. Similarly, variationin larval temperatures has produced signicant differ-ences inDrosophila simulansYutaka wing shape (De-bat et al. 2003).

Geometric morphometrics has already proven to bea valuable tool forentomological problems.It hasbeenusedto investigate physiologicalchanges between dif-ferent populations of triatomines (Jaramillo et al.2002,Schachter-Broide et al. 2004) and tsetse ies (Bouyer

et al. 2007). This method also is frequently used tosolve taxonomic problems, where it can provide sup-port for proposed changes in taxonomic status or callinto question previouslyaccepted classications (Bay-lac et al. 2003, Mutanen 2005). It also can serve as aninexpensive, reliable, one-step way to identify mem-bersof otherwisemorphologically similar species(Vil-lemant et al. 2007, Henry et al. 2010).

The results from this experiment show that wingshape by itself is not a useful means of discriminatinglarval competition and temperature conditions forAe.albopictusorAe. aegyptifrom Florida, despite nding

signicant effects of competition and temperature onsemilandmark position. However, this approach couldbe successfully used in future studies under differentcircumstances. For example, there are many moremosquito species across different genera that can actas vectors for other diseases (Jupp et al. 1981, Rosen

et al. 1985, Turell et al. 2001), and a similar morpho-metric shape analysis on them might prove produc-tive. Indeed, Aytekin et al. (2009) has shown thatdifferences in wing shape ofAn. superpictus caused bylarval temperature are readily detected by morpho-metric analysis. If larval rearing conditions of thisspeciesinuence vector competence, then wing shape

may be a useful indicator of vector competence inAn.superpictus.Alternative approaches may involve mor-phometricanalyses on the shape of a mosquitos entirebody, not just the wings. Analyzing shape in threedimensions by using geometric morphometrics pres-ents some difculties (Spencer and Spencer 1995,Bookstein 1996b, Zelditch et al. 2004), but it has beendone (Dean et al. 1996, OHiggins and Jones 1998,Hildebrandet al.1999). Theproblem of uninformativesemilandmarks also requires a solution, because mos-quito wing shape is probably inuenced by the prox-imal, posterior curve of the wing (Fig. 1). Gunz et al.

(2005) suggest a statistical solution to the problem ofsemilandmark placement on curves,and this approachmay be useful in future work examining environmen-tal effects on wing shape.

Acknowledgments

We thank Miriam Zelditch, Thom DeWitt, and Mel Jonasfor help with geometric morphometric analysis and inter-pretation. We also thank Phil Lounibos (Florida MedicalEntomology Laboratory) for providing access to facilities forwork in Florida. This work was funded in part by grant

R01-(AI)-44793 from the National Institutes of Health Na-tional Institute of Allergy and Infectious Diseases.

References Cited

Adams, D. C., F. J. Rohlf, and D. E. Slice. 2004. Geometricmorphometrics: ten years of progress following the rev-olution. Ital. J. Zool. 71: 516.

Agnew, P., C. Haussy, and Y. Michalakis. 2000. Effects ofdensity and larval competition on selected life historytraits ofCulex pipiens quinquefasciatus (Diptera: Culici-dae). J. Med. Entomol. 37: 732735.

Alto, B. W., and S. A. Juliano. 2001. Precipitation and tem-

perature effects on populations ofAedes albopictus (Dip-tera: Culicidae): implicationsfor range expansion.J. Med.Entomol. 38: 646656.

Alto, B. W., L. P. Lounibos, S. Higgs, and S. A. Juliano. 2005.Larval Competition differentially affects infection inAedesmosquitoes. Ecology 86: 32793288.

Alto, B. W., L. P. Lounibos, C. N. Mores, and M. H. Reiskind.2008. Larval competition alters susceptibility of adultAedes mosquitoes to dengue infection. Proc. R. Soc. B 275:463471.

Anderson, J. R., and R. Rico-Hesse. 2006. Aedes aegyptivec-torial capacityis determined by the infecting genotype ofdengue virus. Am. J. Trop. Med. Hyg. 75: 88689.

Aytekin, S., A. M. Aytekin, and B. Alten. 2009. Effect ofdifferent larval rearing temperatures on the productivity(R0) and morphology of the malaria vector Anophelessuperpictus Grassi (Diptera: Culicidae) using geometricmorphometrics. J. Vector. Ecol. 34: 3242.

Baylac, M., C. Villemant, and G. Simbolotti. 2003. Combin-ing geometric morphometrics with pattern recognition



11/12

for the investigation of species complexes. Biol. J. Linn.Soc. 80: 8998.

Benedict, M. Q., R. S. Levine, W. A. Hawley, and L. P.Lounibos. 2007. Spread of the tiger: global risk of inva-sion by the mosquitoAedes albopictus. Vector Borne Zoo-notic Dis. 7: 7685.

Bevins, S. N. 2008. Invasive mosquitoes, larval competition,

and indirect effects on the vector competence of nativemosquito species (Diptera: Culicidae). Biol. Invasions 10:11091117.

Bookstein, F. L. 1991. Morphometric tools for landmark da-ta: geometry and biology. Cambridge University Press,New York.

Bookstein, F. L. 1996a. Biometrics, biomathematics and themorphometric synthesis. Bull. Math. Biol. 58: 313365.

Bookstein, F. L. 1996b. Combining the tools of geometricmorphometrics, pp. 131152. In L. F. Marcus, M. Corti, A.Loy, G.J.P. Naylor, and D. E. Slice (eds.), Advances inmorphometrics. Plenum, New York.

Bouyer, J. S. Ravel, J. P. DuJardin, T. De Meeu s, L. Vial, S.Thevenon, L. Guerrini, I. Sidibe, and P. Solano. 2007.

Population structuring of Glossina palpalis gambiensis(Diptera: Glossinidae) according to landscape fragmen-tation in the Mouhoun River, Burkina Faso. J. Med. En-tomol. 44: 788795.

Cassell, D. L. 2002. A randomization-test wrapper for SASPROCs. (http://www2.sas.com/proceedings/sugi27/p251-27.pdf).

Dean, D., P. Buckley, F. Bookstein, J. Kamath, D. Kwon, L.Friedman, and C. Lys. 1996. Three dimensional MR-based morphometric comparison of schizophrenic andnormal cerebral ventricles. Lect. Notes Comput. Sci.1131: 361372.

Debat, V., M. Begin, H. Legout, and J. R. David. 2003. Al-lometric and nonallometric components of Drosophilawing shape respond differently to developmental tem-perature. Evolution 57: 27732784.

Eldridge, B. F.,and J. D.Edman. 2000. Medicalentomology.Kluwer Academic Publishers, Dordrecht, The Nether-lands.

Gimnig, J. E., M. Ombok, S. Otieno, M. G. Kaufman, J. M.Vulule, and E. D. Walker. 2002. Density-dependent de-velopment of Anopheles gambiae (Diptera: Culicidae)larvae in articial habitats. J. Med. Entomol. 39: 162172.

Gratz, N. G. 2004. Critical review of the vector status ofAedes albopictus.Med. Vet. Entomol. 18: 215227.

Griffiths,J. A.,M. Schiffer, and A. A. Hoffmann. 2005. Clinalvariation and laboratory adaptation in the rainforest spe-

cies Drosophila birchii for stress resistance, wing size,wing shape and development time. J. Evol. Biol. 18: 213222.

Gubler, D. J. 1998. Dengue and dengue hemorrhagic fever.Clin. Microbiol. Rev. 11: 480496.

Gunz, P., P. Mitteroecker, and F. L. Bookstein. 2005.Semilandmarks in three dimensions, pp. 7398. In D. E.Slice (ed.), Modern morphometrics in physical anthro-pology.Kluwer Academic/PlenumPublishers, NewYork,

Hard, J. J., W. E. Bradshaw, and D. J. Malarkey. 1989. Re-source- and density-dependent development in tree-holemosquitoes. Oikos 54: 137144.

Hardy, J. L., E. J. Houk, L. D. Kramer, and W. C. Reeves.1983. Intrinsic factors affecting vector competence ofmosquitoes for arboviruses. Annu. Rev. Entomol.28: 229262.

Henry, A., P. Thongsripong, I. Fonseca-Gonzalez, N. Jara-millo-Ocampo, and J. Dujardin. 2010. Wing shape ofdengue vectors from around the world. Infect. Genet.Evol. 10: 207214.

Hildebrand, T., A. Laib, R. Mu ller, J. Dequeker, and P.Ru egsegger. 1999. Direct three-dimensional morpho-metric analysis of human cancellous bone: microstruc-tural data from spine, femur, iliac crest, and calcaneus.J. Bone Miner. Res. 14: 11671174.

Jaramillow, N. O., D. Castillo, and M. E. Wolff. 2002. Geo-metric morphometric differences between Panstrongylus

geniculatus fromeldand laboratory. Mem.Inst. OswaldoCruz 97: 667673.Jirakanjanakit, N., S. Leemingsawat, S. Thongrungkiat, C.

Apiwathnasorn, S. Singhaniyom, C. Bellec, and J. P. Du-jardin. 2007. Inuence of larvaldensityor food variationon the geometryof the wing ofAedes (Stegomyia) aegypti.Trop Med. Int. Health 12: 13541360.

Jupp, P. G., B. M. McIntosh, I. dos Santos, and P. de Moor.1981. Laboratory vector studies on six mosquito and onetick species with chikungunya virus. Trans. R. Soc. Trop.Med. Health 75: 1519.

Kamgang, B., S. Marcombe, F. Chandre, E. Nchoutpouen, P.Nwane, J. Etang, V. Corbel, and C. Paupy. 2011. Insec-ticide susceptibility ofAedes aegypti andAedes albopictusin Central Africa. Parasite Vector 4: 79.

Livdahl, T. P. 1982. Competitionwithinand betweenhatch-ing cohorts of a treehole mosquito. Ecology 63: 17511760.

Lyimo, E. O., W. Takken, and J. C. Koella. 1992. Effect ofrearing temperature and larval density on larval survival,age at pupation and adult size of Anopheles gambiae.Entomol. Exp. Appl. 63: 265271.

Moore, C. G., and C. J. Mitchell. 1997. Aedes albopictusinthe United States: ten-year presence and public healthimplications. Emerg. Infect. Dis. 3: 329334.

MoralesVargas,R. E., P. Ya-umphan, N. Phumala-Morales, N.Komalamisra, and J. Dujardin. 2010. Climate associated

size and shape changes inAedes aegypti(Diptera: Culic-idae) populations from Thailand. Infect. Genet. Evol. 10:580585.

Morrison, A. C., E. Zielinski-Gutierrez, T. W. Scott, and R.Rosenberg. 2008. Dening challenges and proposing so-lutions for control of the virus vectorAedes aegypti. PLoSMed. 5: 68.

Mutanen, M. 2005. Delimitation difculties in species splits:a morphometric case study on theEuxoa triticicomplex(Lepidoptera, Noctuidae). Syst. Entomol. 30: 632643.

Muturi, E. J., C.-H. Kim, B. W. Alto, M. Berenbaum, and M.Schuler. 2011. Larval environmental stress alters adultmosquito tness and competence for arboviruses. Trop.

Med. Int. Health 16: 955964.OHiggins, P., and N. Jones. 1998. Facial growth in Cerco-cebus torquatus:an application of three-dimensional geo-metric morphometric techniques to the study of mor-phological variation. J. Anat. 193: 251272.

Rohlf, F. J. 1999. Shape statistics: procrustes superimposi-tions and tangent spaces. J. Classif. 16: 197223.

Rohlf, F. J. 2008. TPSRelw version 1.46. Department ofEcology and Evolution, State University of New York.(http://morph.bio.sunysb.edu./morph/index/html).

Rohlf, F. J. 2009. TpsDig version 2.14. Department of Ecol-ogyandEvolution,State University of New York. (http://morph.bio.sunysb.edu./morph/index/html).

Rosen, L., L. E. Roseboom, D. J. Gubler, J. C. Lien, and B. N.

Chaniotis. 1985. Comparative susceptibility of mosquitospecies and strains to oral and parenteral infection withdengue and Japanese encephalitis viruses. Am. J. Trop.Med. Hyg. 34: 603615.

Rueda, L. M., J. K. Patel, R. C. Axtell, and R. E. Stinner. 1990.Temperature-dependent development and survival rates



12/12

of Culex quinquefasciatus and Aedes aegypti (Diptera:Culicidae). J. Med. Entomol. 27: 892898.

Russell, R. C. 1986. Larval competition between the intro-ducedvectorofdenguefeverinAustralia,Aedes aegypti(L),and a native container-breeding mosquito,Aedes notoscrip-tus (Skuse) (Diptera, Culicidae). Aust. J. Zool. 34: 527534.

SAS Institute. 2008. SAS/STAT 9.2 users guide. SAS In-

stitute, Cary, NC.Schachter-Broide, J., J. P. Dujardin, U. Kitron, and R. E.Gu rtler. 2004. Spatial structuring ofTriatoma infestans(Hemiptera, Reduviidae) populations from northwest-ern Argentina using wing geometric morphometry.J. Med. Entomol. 41: 643649.

Sheets, H. D. 2006a. CoordGen. (http://www3.canisius.edu/sheets/moremorph.html).

Sheets, H. D. 2006b. CVAGen. (http://www3.canisius.edu/sheets/moremorph.html).

Spencer, M. A., and G. S. Spencer. 1995. Video-based three-dimensional morphometrics. Am. J. Phys. Anthropol. 96:443453.

Tabachnick, W. J., and J. R. Powell. 1979. A world-wide

survey of genetic variation in the yellow fever mosquito,Aedes aegypti.Genet. Res. 34: 215229.

Tun-Lin, W., T. R. Burkot, and B. H. Kay. 2000. Effects oftemperature and larval diet on development rates and

survival of the dengue vector Aedes aegypti in northQueensland, Australia. Med. Vet. Entomol. 14: 3137.

Turell, M. 1993. Effect of environmental temperature onthe vector competence ofAedes taeniorhynchusfor RiftValley fever and Venezuelan equine encephalitis viruses.Am. J. Trop. Med. Hyg. 49: 672676.

Turell, M., M. L. OGuinn,D. J. Dohm, and J. W. Jones. 2001.

Vector competence of North Americanmosquitoes (Dip-tera: Culicidae) for West Nile virus. J. Med. Entomol. 38:130134.

Villemant, C., G. Simbolotti, and M. Kenis. 2007. Discrim-ination ofEubazus (Hymenoptera, Braconidae) siblingspecies using geometric morphometrics analysis of wingvenation. Syst. Entomol. 32: 625634.

Westbrook, C. J. 2010. Larval ecologyand adult vector com-petence of invasive mosquitoes Aedes albopictus andAedes aegyptifor Chikungunya virus. Ph.D. dissertation,University of Florida, Gainesville.

Zelditch,M. L.,D. L. Swiderski, H. D. Sheets,and W. L. Fink.2004. Geometric morphometrics for biologists. Aca-

demic, New York.

Received 19 January 2012; accepted 26 April 2012.


Documents

Stephens Juliano 12 JME Wing Shape