morphometric inferences on sibling species and …€¦ · syst. biol. 46(l):180-194/ 1997...

Syst. Biol. 46(l):180-194/ 1997

MORPHOMETRIC INFERENCES ON SIBLING SPECIES AND SEXUALDIMORPHISM IN NEOCHLAMISUS BEBBIANAE LEAF BEETLES:MULTIVARIATE APPLICATIONS OF THE THIN-PLATE SPLINE

DEAN C. ADAMS AND DANIEL J. FUNK

Department of Ecology and Evolution, State University of New York, Stony Brook, New York 11794-5245, USA;E-mail: [email protected] (D.C.A.)

Abstract.—Nominally polyphagous species of herbivorous insects sometimes are comprised ofmultiple morphologically similar biological species with more specialized appetites. When meris-tic morphological traits cannot be found to distinguish such suspected sibling species, moleculardata are increasingly sought as a source of evidence. A role for morphology in distinguishingsuch taxa might be reclaimed, however, by recent advances in geometric morphometric methods,such as the statistical analysis of partial-warp scores from the thin-plate spline. We employed thismethod to detect and characterize subtle shape differences among populations and between sexesof the nominal leaf beetle species Neochlamisus bebbianae. Using the thin-plate spline, the shapesof specimens from seven beetle populations collected from five host plants in five eastern NorthAmerican localities were calculated. These shapes were analyzed by MANOVA, revealing signif-icant variation in both uniform and nonuniform components of shape among test populations.Significant sexual dimorphism in size, shape, and allometric relationships were also documentedacross these populations. More interestingly, our study provided evidence of sibling species wheretraditional taxonomic approaches have failed. Individual MANOVAs revealed significant shapevariation between sympatric populations from different host plants in each of three localities.Because these sympatric shape differences were significant when adjusted for size, they cannotbe attributed to allometric consequences of size variation among test populations. Because certainbeetle populations differed significantly in size and shape when reared in a common environment,these morphometric traits may have a genetic basis. Together, these results are consistent with anearlier suggestion that N. bebbianae represents a complex of host-specific races or sibling species,a hypothesis that has received additional support from recent studies on host use traits, sex ratios,and mitochondrial DNA. In sum, these analyses demonstrate the power and utility of the thin-plate spline as a morphological means of discriminating among closely related and anatomicallyhomogeneous taxa. [Chrysomelidae; host races; morphometrics; Neochlamisus; relative warps; sex-ual dimorphism; sibling species; thin-plate spline.]

When separate biological species are ex- acters, he inferred that interbreeding didtremely similar in morphology, document- not occur between such sympatric popu-ing their existence is often difficult and lations, which he accordingly diagnosed asmany complexes of such cryptic or sibling separate "phytophagic species." Walsh'sspecies go undetected, languishing under approach has since been used to ferret outsingle nominal species names (Mayr, many host-specific biological species from1942). This is especially true when muse- within single nominal species assumed toum specimens and traditional taxonomic have more generalized feeding habitsapproaches provide the only available ev- (Brown, 1958, 1959; Diehl and Bush, 1984).idence on biological variation. Observa- The recognition of cryptic biological spe-tions on behavioral or ecological polymor- ties is of taxonomic significance and is alsophisms, however, may hint at such sibling vitally important for evolutionary biolo-species and have proven especially infor- gists, ecologists, and applied biologistsmative in systematic studies of herbivo- who often assume the evolutionary androus insect taxa. In 1864, Benjamin Walsh genetic coherence of the nominal speciesdescribed slight but consistent morpholog- that they study. Data on "intraspecific"ical differences between nominally conspe- differentiation in host use, patterns of eco-cific populations of insects associated with logical specialization, or the efficacy of bi-different host plants at single localities, ological control agents, for example, mayFrom the strict segregation of these char- demand quite different interpretations if

180

Downloaded from https://academic.oup.com/sysbio/article-abstract/46/1/180/1685504/Morphometric-Inferences-on-Sibling-Species-andby gueston 16 September 2017

1997 ADAMS AND FUNK—LEAF BEETLE MORPHOMETRICS 181

nominally conspecific study populationsactually belong to separate, host-specificsibling species. However, although mor-phological data are commonly a valuablesource of phylogenetically informativecharacters (Hillis, 1987; Patterson et al,1993), even careful taxonomic scrutinymay fail to furnish morphological apomor-phies that discriminate among extremelysimilar taxa from different hosts. Fortu-nately, even morphologically identical spe-cies can sometimes be distinguished usingmolecular data such as allozymes or DNAsequences. Recently, molecular genetic ap-proaches have documented evolutionarydifferentiation among host-associated pop-ulations within a number of nominal her-bivore species (e.g., Guttman et al., 1981;Diehl and Bush, 1984; Berlocher, 1989;Menken, 1989; Waring et al., 1990; Roini-nen et al., 1993; Brown et al., 1996; Funk,1996). Must morphology, then, wholly cedeits diagnostic role to the molecules whenclosely related species have not diverged inmeristic morphological traits? In such in-stances, morphology might still play a rolethrough the quantitative analysis of con-tinuously varying shape characters, the do-main of morphometrics.

In traditional morphometric analyses,linear distances are measured betweenpairs of apparently homologous morpho-logical points, known as landmarks (Black-ith and Reyment, 1971; Marcus, 1990,1993). These interlandmark distances arethen analyzed using multivariate statistics.One limitation of the traditional approach,however, is its assumption that these dis-tances, too, are homologous across taxa,even though each distance reflects the(possibly independent) evolution of thestructures that it spans (Bookstein, 1982,1990). A second limitation is the partial re-dundancy of information about shape pro-vided by any two distances, yieldingweaker statistical tests of morphologicaldifferences (Strauss and Bookstein, 1982).Because of these shortcomings, morpho-metricians are now describing two- andthree-dimensional forms by capturing andanalyzing the Cartesian coordinates of ho-mologous landmarks (Bookstein, 1989,

1991; Rohlf and Marcus, 1993). Such geo-metric morphometric methods allow shapeto be described more comprehensively bypreserving the positional relationshipsamong landmarks, thus minimizing re-dundancy and allowing for more powerfulstatistical tests of shape differences. In thispaper, we characterize shape variation us-ing a recent advance in geometric morpho-metric methodology: the analysis of par-tial-warp scores using the thin-plate spline(Bookstein, 1989, 1991). These analyseswere performed using data from host-as-sociated populations of the nominal leafbeetle species, Neochlamisus bebbianae(Brown).

Neochlamisus (Karren, 1972) is a genus ofNorth American chrysomelid beetles thatfeed on the foliage of their host plants asboth larvae and adults. Although membersof this genus as a whole use host plantsfrom a taxonomically disparate array ofplant families (Karren, 1972), most speciesappear to have monophagous tendencies,primarily using a single host genus or evena single species. The nominal species N.bebbianae, however, appears to be moregeneralized in its feeding habits. Field col-lections and laboratory studies documentits use of six plant genera from four diver-gent families of angiosperms (Brown,1943, 1946, 1952; Funk, 1996, unpubl.data): Acer rubrum (red maple, Aceraceae),Alnus rugosa/A. serrulata (speckled al-der/hazel alder, Betulaceae), Betula nigra(river birch, Betulaceae), Corylus americana(American hazel, Betulaceae), Quercus spp.(oak, Fagaceae), and Salix bebbiana (Bebb'swillow, Salicaceae).

In his studies of these beetles in Ontario,Canada, Brown (1943, 1946, 1952) ob-served that local populations sometimesused one of these hosts and ignored oth-ers. He also noted subtle color differencesamong series of beetles associated withdifferent host plants. These observationsconvinced Brown that all species of Neo-chlamisus were highly specialized and thateach N. bebbianae host thus supported aseparate, host-specific sibling species. Hav-ing studied specimens from four hosts, Al-nus, Corylus, Salix, and Quercus, Brown de-


182 SYSTEMATIC BIOLOGY VOL. 46

scribed populations associated with thefirst three as separate sibling species andsuggested that Quercus populations likelyrepresented a fourth sibling (Brown, 1943,1946, 1952).

However, in his revision of Neochlamisus,Karren (1972) found no consistent mor-phological differences among specimensfrom these populations in relation to theirhost of origin despite his examination ofinternal male genitalia, which often iden-tify externally indistinguishable insect spe-cies (Eberhard, 1985). Karren thus synon-ymized Brown's sibling species under N.bebbianae, concluding that the host-associ-ated forms were simply populations of asingle, more ecologically generalized spe-cies. Karren also referred particular speci-mens collected from Acer and Betula (hostassociations unknown to Brown but sinceconfirmed by D.J.E) to N. bebbianae. Basedon the study of larval characters, LeSage(1984) resurrected the idea that Alnus- andSalix-using forms belong to separate sib-ling species. Clearly, additional sources ofdata are needed to address the evolution-ary status of host-associated populationspresently assigned to the nominal speciesN. bebbianae.

Another intriguing aspect of Neochlami-sus biology is the unusual patterns of sexallocation exhibited by various species, in-cluding N. bebbianae. Population sex ratiosare often strongly female biased (unpubl.data), and many species are apparentlysexually dimorphic in size, with femaleslarger than males (Karren, 1972). Qualita-tive observations on interspecific patternsof sexual size and shape dimorphism(D.J.E, pers. obs.) seem to parallel those re-vealed by recent quantitative investiga-tions of sexual dimorphism and its evolu-tionary causes in water striders (e.gvFairbairn, 1990,1992; Fairbairn and Prezio-si, 1994). These patterns suggest a furtheravenue of evolutionary study that might beadvanced by morphometric approaches.

In this study, we employed a recent ad-vance in geometric morphometric meth-odology to identify and describe shapevariation between sexes and among host-associated populations of Neochlamisus beb-

TABLE 1. Data on populations of Neochlamisus beb-bianae collected from five of its six well-documentedhost plants. * = plants believed by Brown (1943,1946,1952) to support host-specific sibling species withinN. bebbianae. No beetles were collected for study fromthe host of the fourth purported sibling, hazel. TheQuebec AZnus-associated population was not treatedin the original analyses.

Location

GeorgiaGeorgiaGeorgiaNew YorkNew YorkOklahomaOklahomaOklahomaOntarioOntarioOntarioQuebec

Host plant

Acer rubrumAcer rubrumBetula nigraAcer rubrumAcer rubrumQuercus spp.*Betula nigraBetula nigraSalix bebbiana*Salix bebbiana*Alnus incana*Alnus incana*

Beetle specimens

Sex

femalemalefemalefemalemalefemalefemalemalefemalemalefemalefemale

n

303030303030303030303030

bianae. Our goals were (1) to provide a fur-ther morphological test of the hypothesisthat N. bebbianae includes host-specific sib-ling species, (2) to assay patterns of sexualdimorphism that might point towards fur-ther research, and (3) to reclaim for mor-phology a role increasingly played by mo-lecular genetic analyses, that of discerningdifferences between closely related andmorphologically similar populations andtaxa.

MATERIALS AND METHODS

Collection of Specimens

The specimens treated in this studywere collected from several regions span-ning eastern North America: northeasternGeorgia; Suffolk County, New York; Lati-mer County, Oklahoma; Ottawa, Ontario;and Cowansville, Quebec (Table 1). Thesespecimens were collected from five of thesix well-documented host plants of N. beb-bianae, leaving Corylus as the only host notrepresented by a test population in thisstudy. Animals from each Georgia, Okla-homa, and New York study populationwere collected from a number of siteswithin each region, but those from Ontarioand Quebec were each collected from asingle site. The specific specimens from



each of the host-associated populations(Table 1) that were used for analysis wererandomly selected from beetles pooledacross collecting sites within a given re-gion.

In three cases, study specimens werecollected from two different hosts in a sin-gle region, allowing comparisons of sym-patric (or quasi-sympatric) host-associatedpopulations. In Ontario, Alnus- and Salix-associated beetles were collected fromplants that intermingled in the same fields.In Georgia, some Acer and Betula beetleswere collected from the same sites, wheretheir hosts intermingled along streams. InOklahoma, although both Quercus and Bet-ula beetles were not collected from inter-mingling host plants at any single collec-tion site, distances between collection sitesfor the two hosts were as little as a fewhundred meters, whereas distances be-tween collection sites from a single hostwere as great as 20 km. Thus, collectionsites for the two hosts were themselvesspatially intermingled at a larger geo-graphic scale.

From each test population, 30 animalsfrom a given sex were collected for anal-ysis. Owing to the rarity of males, only fe-males were examined for some popula-tions (Table 1). All test animals werecollected directly from their host plant, inmost cases as eggs or young larvae, whichwere subsequently reared to eclosion onfoliage from the host on which they werecollected. Ten of the 30 specimens of Okla-homa Betula males and females and 10 ofthe 30 Oklahoma Quercus females werecollected as adult beetles, however, andsome of the Alnus and Salix beetles fromOntario and Quebec were collected as lateinstar larvae or pupae.

Data Collection

Eleven landmarks were chosen for theircapacity to define major elements of shapeand for their reliability as homologous bi-ological structures (Fig. 1; Table 2). Basedon Bookstein's (1991) classification, five ofthese landmarks are of type I, representingmeeting points of two or more tissues, andsix are of type II, marking points of max-

FIGURE 1. Dorsal (a) and lateral (b) views of the 11landmarks used to define the shape of Neochlamisusbebbianae specimens (see description of landmarks inTable 2). For each view, the anterior of the animal isto the right. HC = head capsule; Th = thorax; Pr =pronotum; Ab = abdomen. Illustrations reproducedfrom Karren (1972).

imum curvature. Two-dimensional coor-dinates for these landmarks were collectedfrom all 330 specimens using a Wild dis-secting microscope, a CCD video camera,and the program MorphoSys (Meachamand Duncan, 1990). Specimens were scoredin random order with respect to popula-tion of origin.

Choosing a Geometric Morphometric Method

A variety of geometric morphometricmethods have been developed to analyzetwo-dimensional landmark data such asthose collected from our study specimens.With Euclidean distance matrix analysis(Lele and Richtsmeier, 1991; Richtsmeier etal., 1992), the shapes of two specimens arecompared by calculating ratios of their in-



TABLE 2. Morphological landmarks used in thisstudy, as classified by Bookstein (1991). See Karren(1966) for definitions of morphological terms.

Land-mark

1

2

3

4

5

6

7

8

9

1011

Description

junction of pronotum and left ely-tron

junction of pronotum and left mar-gin of head capsule

junction of pronotum and right mar-gin of head capsule

junction of pronotum and right ely-tron

apex of most posterolateral tubercle(su-3a) on right elytron

apex of most posterolateral tubercle(su-3a) on left elytron

apex of left pronotal gibbosity(=hump)

apex of right pronotal gibbosity(=hump)

junction of mesoscutellum, right ely-tron, and left elytron

apex of su-2 tubercle on left elytronapex of su-2 tubercle on right ely-

tron

Type

I

I

I

I

II

II

II

II

I

IIII

terlandmark distances for all possible dis-tances (Lele and Richtsmeier, 1991). Shapedifferences increase as these ratios departfrom unity. Although this method allowsfor the statistical description of shape, onlytwo specimens can be compared at a timeand visualization of shape variation is notpossible (see Rohlf et al., 1996). Finite-ele-ment scaling analysis (Cheverud et al.,1983) adopts an engineering perspective,decomposing the form of each specimeninto a composite of small geometric ele-ments and calculating the "strain" neces-sary to deform the elements of one speci-men into those of the other (Cheverud andRichtsmeier, 1986). Unfortunately, resultsfrom this method are sensitive to the (ar-bitrary) choice of the elements employed(Bookstein, 1991). The analysis of super-imposed specimens (e.g., Walker, 1993) de-scribes shape variation by comparing in-dividual specimens with a consensusconfiguration, representing the averageCartesian coordinates for each landmarkacross all specimens (Bookstein, 1991;Rohlf, 1993b). In this approach, specimensare first superimposed on one another so

that the deviations of landmarks are min-imized based on some criterion (Rohlf andSlice, 1990). The residuals from this super-imposition are then treated as shape dataand used directly for group comparisons(Walker, 1996). Both median-based (Siegeland Benson, 1982) and least squares orProcrustes (Sneath, 1967; Gower, 1971)methods have been proposed. Althoughthe Procrustes method is the most mathe-matically convenient criterion, empiricalstudies have indicated that both methodsgenerate nearly identical results whensample sizes and the number of landmarksare large (Slice, 1993). Although the anal-ysis of superimposed specimens allows fora description of total shape variation, itdoes not allow for the partitioning ofshape into uniform and nonuniform com-ponents. Such partitioning of shape varia-tion is only possible with the thin-platespline.

Using the thin-plate spline (Bookstein,1989, 1991), variation among specimens isexpressed as variation in the parameters ofan interpolation function that compares agiven specimen to the consensus configu-ration. These parameters are treated as de-scriptors of shape, much as the parametersfrom a Fourier analysis are used to de-scribe the outline of a curve (see Rohlf,1990). Specimens are first superimposedusing the Procrustes method because thisdistance measure is assumed by Kendall's(1984) shape space, which is used by thethin-plate spline (Bookstein, 1996a). A con-sensus configuration is determined, andthe thin-plate spline function is calculated.This function is then decomposed into anumber of geometrically orthogonal ele-ments called principal warps. These warpsare a set of multidimensional shape axesdescribing a set of deformations of variousmagnitudes that together account for theentire shape of the consensus configura-tion. Superimposed specimens are project-ed onto these principal warps to describetheir deviations from the consensus config-uration (Bookstein, 1989; Rohlf, 1993b).These projections, or partial-warp scores,are used as data in multivariate analyses(Rohlf, 1993b; Rohlf et al., 1996).



Shape deformations can also be mathe-matically partitioned into uniform andnonuniform components. Uniform com-ponents represent simple overall linearstretching or compression of specimens,whereas nonuniform components charac-terize more complex changes in shape,such as nonlinear patterns of deformationwhere changes can be localized to smallregions on specimens. The thin-platespline is particularly appealing because itallows the graphical depiction of shapevariation as deformations in a manner sim-ilar to D'Arcy Thompson's (1917) transfor-mation grids, where dramatic and subtlevariations alike can be appreciated visual-ly. For these reasons, we have adopted thethin-plate spline to analyze shape variationamong populations of N. bebbianae.

Morphometric and Statistical Analyses

Centroid size, the square root of the sumof the squared distances from each land-mark to the centroid (i.e., the center ofgravity), was calculated for each specimen,and Procrustes superimposition of speci-mens was provided by the computer pro-gram GRF-ND (Slice, 1994), allowing theconsensus configuration to be determined.Bookstein's (1993, 1996b) uniform shapecomponents and nonuniform shape vari-ables (partial-warp scores) were then cal-culated using the thin-plate spline pro-gram in NTSYS-pc (TPSWTS; Rohlf,1993a), generating a set of 16 partial-warpscores for each specimen. To identify anddescribe major trends in nonlinear shapevariation, we performed a principal com-ponents analysis on the matrix of partial-warp scores (the W matrix) (Rohlf, 1993b).In such an analysis, the first principal com-ponent axis, i.e., the first relative warp(Bookstein, 1991), represents the directionof maximal nonuniform shape variationamong all specimens (Rohlf, 1993b). Whencalculating the relative warps, we set thescaling parameter, a, to 0, as recommend-ed by Rohlf (1993b) for taxonomic studies.A value of 0 weights all partial-warpscores equally so that neither large-scalenor small-scale shape differences are em-phasized. Although different values of a

affect the description of shape variation byrelative warps, they do not affect statisticalinferences based on a MANOVA and ca-nonical variates analysis because these sta-tistical methods are scale free.

To determine if shape varied significant-ly among test populations and between thesexes, we performed a two-way MANOVAon both the uniform (wl and u2) and non-uniform (W matrix) components of shape.Because no evidence of interactions be-tween test population and sex were found,we were able to perform a canonical vari-ates analysis to identify the patterns ofshape similarity among populations. Fol-lowing these analyses, we specificallycompared pairs of populations from boththe Georgia and Oklahoma sites using aMANOVA to determine whether sympatricpopulations from different hosts differedin shape. We also compared the allometricslopes of sympatric populations using aMANCOVA to determine whether shapedifferences among populations were sim-ply the allometric consequence of size dif-ferences among populations. Both analyseswere performed using females only be-cause one of the two populations in eachcomparison was not represented by malespecimens. Following the completion ofthe original analyses, A/nws-associatedspecimens were obtained from Ontario, al-lowing a third sympatric comparison (withOntario Salix beetles). The MANOVA andMANCOVA were likewise applied to thesepopulations. Although the sympatric com-parisons had been conceived a priori, theremaining 19 population comparisonswere made a posteriori after transformingthe generalized distances from the canon-ical variates analysis into Hotelling T2 val-ues (Rohlf, 1993a). Using the Bonferronimethod (Sokal and Rohlf, 1995), an exper-iment-wise error rate of 0.05 was main-tained by dividing by the number of un-planned comparisons (19), to obtain acritical a of 0.0026.

We investigated sexual dimorphism by(1) comparing male and female centroidsize to determine if the sexes were dimor-phic in size, (2) determining from the over-all MANCOVA if the sexes were dimorphic


186 SYSTEMATIC BIOLOGY VOL. 4 6

(a) + deviation

FIGURE 2. Generalized least squares (Procrustes)superimposition of 330 specimens of N. bebbianaeshowing the range of variation at each landmark aftertranslation, rotation, and scaling of data from individ-ual specimens. The outline formed by connecting themean coordinates of each landmark represents theconsensus configuration for an "average" N. bebbianaespecimen.

in shape, (3) performing a multivariate re-gression of shape on size to determine ifoverall shape variation had an allometriccomponent, and (4) performing a MAN-COVA with sex as the categorical variableand centroid size as the covariate to deter-mine if sexual dimorphism in shape wasan allometric consequence of dimorphismin size. We also generated representationsof male and female shapes by regressingthe partial-warp scores onto a dummyvariable with a separate state for each sex.All statistical analyses were performed us-ing NTSYS-pc (Rohlf, 1993a) and SAS.

RESULTS

General Description of Shape Variation

Figure 2 depicts variation among testspecimens at each landmark and illustratesan average N. bebbianae study specimen byconnecting the mean values for each land-mark, which together define the consensusconfiguration. The first relative warp(RW1) explained 29% of shape variationamong specimens, and the first three rel-ative warps combined to explain 58% ofthis variation. Shape variation along RW1illustrates sources of deformation from theconsensus configuration (Fig. 3). At thepositive extreme of RW1, internal land-

(b) mean

***

• • • • *

\

/

/

\

\

\

\ /

\

/

V

•

(c) - deviation

FIGURE 3. Overall deformations in shape along thefirst relative warp for all 330 beetle specimens. Thisfigure represents the direction of maximal nonuni-form shape variation obtained from a principal com-ponents analysis of the partial-warp scores (W ma-trix). The range of observed shape variation along thisaxis is presented. The broader, squatter shapes illus-trate positive deformations (a) and the narrower, moreelongate shapes show negative deformations (c) fromthe consensus configuration (b).

marks (7-11 in Fig. 1) are closely arrangedand outline landmarks (1-6) define a rela-tively squat shape (Fig. 3a) in which thepronotum (defined by landmarks 1-4 and9) occupies a large proportion of the totalsurface area. At the negative extreme ofRW1, internal landmarks are more highlydispersed, and outline landmarks define amore elongate shape (Fig. 3b) with a rela-tively smaller pronotum.



TABLE 3. Two-way MANOVA for uniform and nonuniform (W matrix) components of shape. Analyses treatall 330 beetle specimens, using two variables for uniform components and 2p — 6 (=16) variables for nonuni-form components.

Source

UniformHost plantSex

Host X Sex

NonuniformHost plantSex

Host x Sex

df

414

414

Wilks A

0.90220.98910.9962

0.60680.86890.9116

parms

2, 4, 3202,1, 3202, 4, 320

16, 4, 32016,1, 32016, 4, 320

F

4.241.770.31

2.562.890.91

df

8,6422,3214,642

64,120416, 30732, 614

P

0.0001***0.17200.8727

0.0001***0.0002***0.6136

Shape Differentiation among Test Populations

Two-way MANOVAs on both uniformand nonuniform components of shape re-vealed significant variation among testpopulations (Table 3), as did canonical var-iates analysis (Wilks A = 0.47884, P = 2 X10 ~13). A minimal spanning tree based ongeneralized distances illustrated the rela-tive similarity of populations separated inshape space by ordination along the firsttwo canonical axes (Fig. 4; Sneath and So-kal, 1973). Projecting specimens onto thecanonical vectors and performing a regres-

FIGURE 4. Canonical variates ordination of beetletest populations, presented as a minimal spanningtree. Beetle shapes are thin-plate spline representa-tions of the deformations implied by variation alongthe first and second canonical axes. State or provinceabbreviations for collection localities are followed byabbreviations for host plant: Ac = Acer; Al = Alnus; B— Betula; Q = Quercus; S = Salix.

sion of partial-warp scores on each canon-ical axis illustrated implied deformationsin shape along these axes (Fig. 4).

For all three regions in which beetleswere collected from two host plants, high-ly significant shape differences were de-tected between these sympatric, host-as-sociated populations (Georgia: Wilks A =0.6154, P = 0.0012; Oklahoma: Wilks A =0.5788, P = 0.0002; Ontario: Wilks A =0.1431, P = 0.0001). Additionally, 6 of the19 possible unplanned comparisonsamong allopatric populations showed sig-nificant shape differentiation, even at acritical a of 0.0026: Georgia Acer versusQuebec Alnus, Ontario Salix, and Oklaho-ma Quercus; Oklahoma Betula versus Geor-gia Betula and Ontario Salix; and OntarioSalix versus Oklahoma Quercus (Table 4).

For the sympatric Georgia and Oklaho-ma comparisons, no differences werefound between the allometric slopes of thesympatric populations (Georgia: Wilks A= 0.7065, P = 0.4167; Oklahoma: Wilks A= 0.8508, P = 0.8512). However, in eachcase, population shapes were significantlydifferent when size was held constant al-lometrically (Georgia: Wilks A = 0.5346, P= 0.0164; Oklahoma: Wilks A = 0.5464, P= 0.0222). These populations thus havedifferent intercepts, so population shapedifferences are not simply allometric pro-jections of population size differences. Thesympatric Ontario comparison did showsignificantly different allometric slopes be-tween populations (Wilks A = 0.5345, P =0.0200), again revealing separate allometrictrajectories.



TABLE 4. Multiple comparisons of nonuniform shape components among test populations, t = significantplanned comparisons using MANOVA; * = significant unplanned comparisons based on Hotelling's T2 valueswith Bonferroni adjustment for critical a. The Ontario Alnus test population is not treated in these comparisonsbecause its specimens were collected after these analyses were completed.

GA AcerGA BetulaNY AcerOK QuercusOK BetulaONT SalixQUE Alnus

GA Acer

039.8t23.864.8*39.853.1*59.2*

GA Betula

030.844.957.8*28.831.5

NYA:er

033.815.038.329.8

OK Quercus

048.7t71.4*55.3

OK Betula

060.5*45.0

ONT Salix

044.4

QUE Alnus

0

Shape Variation and Sexual Dimorphism

Comparison of centroid sizes showedthat females were significantly larger thanmales (F = 17.09, P = 5.2 X 10~15), sup-porting Karren's (1972) observations. HOW-

FIGURE 5. Predicted female and male shapes rep-resented as thin-plate splines. Females are more elon-gate, with more anterior pronotal gibbosities and pro-notum/elytrum junctions. Males are squatter, with thepronotal gibbosities and pronotum/elytrum junctionspositioned more posteriorly.

ever, the two-way MANOVA revealed thatthe sexes differed significantly in nonuni-form (although not in uniform) compo-nents of shape (Table 3). Predicted malesare squatter with more posteriorly posi-tioned pronotal gibbosities and prono-tal /elytral junctions, and females are moreelongate with more anteriorly positionedgibbosities and junctions (Fig. 5). Femalesalso appear to allocate a much larger pro-portion of the body to the posterior seg-ments of the thorax and the abdomen thando males.

A statistically significant relationship be-tween shape and size across all test spec-imens (R2 = 0.2971, F = 8.27, P < 0.0001)illustrated significant allometry. Althoughallometric slopes did not differ significant-ly between the sexes (Wilks A = 0.9695, P= 0.8720), shape was significantly sexuallydimorphic when size was held constant al-lometrically, meaning that the intercept formales and females differed. This suggeststhat males and females vary along sepa-rate, although parallel, allometric trajecto-ries.

DISCUSSION

Morphometric Evidence of Sibling Species

Our multivariate analyses of partial-warp scores suggest that certain host-as-sociated populations currently assigned tothe nominal leaf beetle species Neochlami-sus bebbianae may represent separate evo-lutionarily differentiated taxa. In each ofthree cases where sympatric test popula-tions were collected from separate hostplants, shape differences between these



populations were highly significant. Suchlocal shape differentiation might be ex-plained by reproductive isolating mecha-nisms that restrict the homogenizing ef-fects of gene flow between sympatricpopulations. This study thus demonstratesthe capacity of thin-plate spline analysis todetect subtle morphological evidence of bi-ological species that may have eluded thetaxonomist relying on meristic characters(Karren, 1972). More specifically, shapedifferentiation between Alnus- and Salix-associated populations in Ontario supportsthe contention of Brown (1943) and LeSage(1984) that these host forms belong to sep-arate sibling species, differentiation be-tween Quercus and Betula beetles in Okla-homa is consistent with Brown's (1952)suspicion of a Quercus-spedfic sibling, anddifferentiation between Betula and Acerbeetles in Georgia suggests that thesehosts, unexamined by Brown, may alsosupport separate sibling species.

Before sympatric shape differences areinterpreted as evidence for reproductiveisolation and genetic differentiation, how-ever, the possibility that they are environ-mentally induced must be considered. Thehost-associated populations currently as-signed to N. bebbianae might belong to asingle biological species with generalizedfeeding habits that develops different adultshapes depending on its larval host. Suchtrophically induced polymorphisms havebeen documented, for example, in richlidfishes (Meyer, 1987). Host plant-inducedshape variation, however, has rarely beenfound in herbivorous insects; one excep-tion is Gillham and Claridge's (1994) studyof Alnetoidia alneti leafhoppers. Adult leaf-hoppers collected from two host plant spe-cies as adults differed significantly inshape, but those collected from these twohost plants as young nymphs and thenreared to maturity on a single host speciesdid not exhibit different shapes.

Such direct induction of adult shape byjuvenile feeding experience is, however,more plausible in hemimetabolous leaf-hoppers than in holometabolous beetles.Whereas hatchling leafhoppers are essen-tially miniature adults that grow and

sprout wings during development, Neo-chlamisus larvae are grublike creatures thatmetamorphose into adults with a striking-ly different appearance. During metamor-phosis, larval tissues are almost entirelybroken down before adult structures arebuilt anew. At this time, any host plant-induced changes in larval shape are likelyto be lost, just as host plant-inducedchanges in feeding preferences generallyare (Jaenike, 1990:249). Because adults donot grow and because we selected land-marks from rigid elytral and pronotalstructures, it is also unlikely that observedshape differences were influenced by adultfeeding experience. Our rearing of Betula-associated larvae from Georgia and Okla-homa on the same host material underidentical laboratory conditions provided apost facto common garden experimentshowing that N. bebbianae populations mayexhibit significant shape variation evenwhen larval environment is held constant(Table 4). These results suggest that shapedifferences may have a genetic basis.

Although herbivore shape may rarely bedirectly induced by larval host plant, her-bivore size may vary according to the suit-ability of the host for larval development.If beetle shape varies as an allometric con-sequence of such size variation, larval hostassociation might thus indirectly inducepopulation shape differences. Such an al-lometric component of shape variation wasdocumented within N. bebbianae test pop-ulations. However, because allometric tra-jectories differed significantly betweensympatric populations in every case, theshape differences between these popula-tions cannot be attributed to differences insize. Furthermore, our post facto commongarden experiments also suggest that sizevariation among populations may itself begenetically based rather than host induced.In both cases where two test populationswere collected from the same host andreared on its foliage in a common labora-tory environment, adults of the two pop-ulations differed significantly in size(Georgia and New York Acer: F = 23.34, P< 0.0001; Georgia and Oklahoma Betula: F= 15.57, P < 0.0001).



Our morphometric inferences on the ex-istence of host-associated taxa within N.bebbianae receive further support from oth-er data sources in each case: Ontario Alnus-and Sa/zx-associated populations exhibitstrikingly different sex ratios, and al-though each will feed on the other's host,each prefers its native host foliage in feed-ing trials (unpubl. data); Oklahoma Betula-and Qwercws-associated populations entire-ly reject one another's hosts in no-choicefeeding trials (unpubl. data); and Betula-and Acer-associated populations fromGeorgia largely reject one another's hostsacross a number of host preference andperformance assays (Funk, 1996). Further-more, Betula beetles (and to a lesser extentAlnus beetles) lack the variable degree ofbluish tint that characterizes Acer and Salixbeetles, supporting Brown's belief in thediagnostic value of color. These resultssuggest the genetic differentiation of thesesympatric population pairs. Insights intothe history of gene flow between them areprovided by other recent studies. Matingexperiments revealed significant sexualisolation between allopatric Acer- and Bet-w/a-associated populations. MitochondrialDNA (mtDNA) sequences collected fromsympatric Acer and Betula beetles in Okla-homa sort monophyletically according tohost plant, suggesting a lack of recent geneflow between them, consistent with the hy-pothesis that they represent completely re-productively isolated sibling species(Funk, 1996). However, certain CanadianAlnus- and Sa/ix-associated beetles sharemtDNA haplotypes, suggesting a more re-cent history of gene flow between popu-lations that may represent incompletelyisolated host races rather than completelyisolated sibling species (for a discussion ofterms, see Jaenike, 1981; Diehl and Bush,1984). The limited number of comparisonsin the present study are insufficient for di-agnosing the actual number and host spec-ificity of evolutionarily differentiated hostforms within N. bebbianae. They also offerno evidence on whether the reproductiveisolation between these host forms origi-nally evolved in allopatry or sympatry.Addressing these issues will require con-

siderable genetic and experimental analy-sis.

Morphometric Evidence of SexualDimorphism

Our analyses also allowed us to char-acterize morphological variation betweenthe sexes. Comparison of centroid sizesshowed that N. bebbianae females are sig-nificantly larger than males, statisticallysupporting Karren's (1972) observation.Such female-biased sexual size dimor-phism is commonly observed in inverte-brate taxa. Using more traditional morpho-metric measures, three recent studies ofinvertebrates (Wickman and Karlsson,1989 [butterflies]; Fairbairn, 1992 [waterstriders]; Prenter et al., 1995 [orb-weavingspiders]) have documented another inter-esting pattern: sexual dimorphism inshape attributable to relatively larger fe-male abdomens. Such shape dimorphismcould be the result of selection on in-creased female fecundity, a hypothesissupported by the correlation of fecunditywith female abdomen size in the commonwater strider Aquarius remigis (Preziosi etal., 1996). If fecundity typically increaseswith female abdomen size, Wickman andKarlsson's (1989) finding that insect abdo-men size increases allometrically withbody size suggests that the pattern of sex-ual size dimorphism described above mayalso be due partly to fecundity selection.Our findings of female-biased size dimor-phism and of sexual dimorphism in shapethat might be attributable to the relativelylarger abdomens of females (Fig. 5) sug-gest that N. bebbianae may offer a usefulsystems for studying these issues.

Studies of sexual dimorphism in taxawith larger males have commonly revealedan interspecific pattern of hyperallometry:as size increases, the degree of sexual sizedimorphism increases. In a taxon withlarger females (water striders), Fairbairn(1990; Fairbairn and Preziosi, 1994) docu-mented a complementary pattern of hy-poallometry: as size increases, the degreeof sexual size dimorphism decreases.These patterns suggest a role for sexual se-lection on male size. Likewise, within Neo-



chlamisus (1) females are the larger sex, (2)the largest known species is clearly theleast dimorphic in size, whereas smallerspecies seem to exhibit stronger sexual sizedimorphism (Funk, pers. obs), (3) signifi-cantly different allometric trajectories formale and female N. bebbianae suggest thatthese interspecific patterns of shape di-morphism could have an allometric com-ponent, and (4) a role for sexual selectionis hinted at by no-choice matings studiesshowing N. bebbianae copulations to besomewhat more likely, more rapidly initi-ated, and of longer duration on averagewhen male: female size ratios of paired an-imals are greater (Funk, 1996).

Studying Shape Differentiation with theThin-Plate Spline

Despite its advantages, the use of thethin-plate spline in biological studies hasthus far been limited. Formulations of themethod by Bookstein (1989,1991) includedits application to studies of rat skull de-velopment, fossil Foraminifera, and humanpatients with Aperr's syndrome. Rohlf(1993b) compared mosquito wings to dem-onstrate the merits of including a scalingparamenter, a. Other applications includestudies of squirrel scapulas (Swiderski,1993), the ontogeny of piranha (Fink andZelditch, 1995), rat skulls (Zelditch et al.,1992), and mole skulls (Loy et al., 1993;Rohlf et al., 1996). Using a somewhat con-troversial approach (F. L. Bookstein and F.J. Rohlf, pers. comm.), Zelditch et al. (1995)recently applied thin-plate splines to iden-tify synapormorphies supporting a partic-ular phylogenetic hypothesis of piranhaevolution. Systematic applications of geo-metric morphometrics have also been pro-posed elsewhere (Rohlf and Marcus, 1993).A few workers have statistically comparedgroups of taxa using these methods. Rohlfand Marcus (1993), Rohlf et al. (1996), andZelditch and Fink (1995) regressed partial-warp scores onto centroid size, finding sig-nificant allometric relationships. Rohlf etal. (1996) compared groups of mole taxausing canonical variates analysis andMANOVAs.

Multivariate statistical analysis of par-

tial-warp scores offers a potentially pow-erful and underexploited means of docu-menting shape differences among taxa.Our study represents one of the first suchstatistical applications of these methods.Morphometric analyses have been used bytaxonomists to justify synonymy and torecognize new species-level taxa (Daly,1985). Here, we document the capacity ofthe thin-plate spline to identify subtle butinformative variations in shape that are notapparent to the naked eye. These analysesthus endorse the recognition of siblingspecies within the nominal species N. beb-bianae that traditional taxonomic approach-es could not discern (Karren, 1972). Wefurther document significant shape differ-entiation in 6 of 19 comparisons betweenallopatric beetle populations, despite usinga very conservative critical a (Table 4). Thesuccessful use of the thin-plate spline inuncovering shape differences betweeneven closely related populations (e.g., Al-nus and Salix beetles) indicates that thismethod would also be useful in futurestudies of morphologically homogeneousspecies. Although molecular genetic ap-proaches have recently gained favor as ameans of distinguishing such taxa, thisstudy suggests that the shape charactersdiscernible by the spline may often be suf-ficiently evolutionarily plastic to allow dif-ferentiation and discrimination of evenclosely related populations and taxa; rela-tive warps clearly distinguished Alnus andSalix beetle populations, but rapidly evolv-ing mtDNA sequences did not (Funk,1996).

Two advantages of the thin-plate splineare its capacity to discern shape differ-ences and the potential to graphically il-lustrate shape variation. Visual depictionsof positional variation at landmarks (Fig.2) and of the major axes of shape defor-mation (Figs. 3-5) allow variable regions tobe identified for further study. Patterns ofvariation might suggest the taxonomic util-ity of particular regions within a particulargroup, whereas nonrandom patterns ofvariation may offer clues to the functionalimportance of particular structures. In ourstudy, such graphical depictions have illus-



trated the previously unrecognized taxo-nomic value of tubercle position (Fig. 4)and pointed to relative abdomen size as ameans of understanding the (possiblyadaptive) nature of sexual dimorphism inboth size and shape in Neochlamisus.

ACKNOWLEDGMENTS

We thank F. James Rohlf, Dennis Slice, Michael Bell,Leo Shapiro, and Douglas Futuyma for their sugges-tions and comments on versions of the manuscript.We also appreciate the assistance of Karl Stephan andLaurent LeSage in the location and collection of studypopulations. This work was supported in part by NSFgrants DEB-9317572 (to F. James Rohlf) and DEB-9229286 (to DJ.F.) and is contribution number 974from the Program in Ecology and Evolution at theState University of New York-Stony Brook.

REFERENCES

BERLOCHER, S. H. 1989. The complexities of host racesand some suggestions for their identification by en-zyme electrophoresis. Pages 51-68 in Electrophoret-ic studies on agricultural pests (H. D. Loxdale andJ. Den Hollander, eds.). Oxford Univ. Press, Oxford,England.

BLACKITH, R. E., AND R. A. REYMENT. 1971. Multivar-iate morphometrics. Blaisdell, New York.

BOOKSTEIN, F. L. 1982. On the cephalometrics of skel-etal change. Am. J. Orthod. 82:177-198.

BOOKSTEIN, F. L. 1989. Principal warps: Thin-platesplines and the decomposition of deformations.I.E.E.E. Trans. Pattern Anal. Mach. Intell. 11:567-585.

BOOKSTEIN, F. L. 1990. Introduction to methods forlandmark data. Pages 215-225 in Proceedings of theMichigan morphometrics workshop (F. J. Rohlf andF. L. Bookstein, eds.). Univ. Mich. Mus. Zool. Spec.Publ. 2.

BOOKSTEIN, F. L. 1991. Morphometric tools for land-mark data. Cambridge Univ. Press, New York.

BOOKSTEIN, F. L. 1993. A brief history of the morpho-metric synthesis. Pages 15-40 in Contributions tomorphometrics, Volume 8 (L .F. Marcus, E. Bello,and A. Garcia-Valdecasas, eds.). Monografias delMuseo National de Ciencias Naturales (CSIC), Ma-drid.

BOOKSTEIN, F. L. 1996a. Combining the tools of geo-metric morphometrics. Pages 131-151 in Advancesin morphometrics (L. F. Marcus, M. Corti, A. Loy,G. Naylor, and D. Slice, eds.). Plenum, New York.

BOOKSTEIN, F. L. 1996b. A standard formula for theuniform shape component. Pages 153-168 in Ad-vances in morphometrics (L. F. Marcus, M. Corti, A.Loy, G. Naylor, and D. Slice, eds.). Plenum, NewYork.

BROWN, J. M., W. G. ABRAHAMSON, AND P. A. WAY.1996. Mitochondrial DNA phylogeography of hostraces of the goldenrod ball gallmaker, Eurosta soli-daginis (Diptera: Tephritidae). Evolution 50:777-786.

BROWN, W. J. 1943. The Canadian species of Exema

and Arthrochlamys (Coleoptera, Chrysomelidae).Can. Entomol. 75:119-131.

BROWN, W. J. 1946. Some new Chrysomelidae, withnotes on other species (Coleoptera). Can. Entomol.78:47-54.

BROWN, W. J. 1952. Some species of Phytophaga (Co-leoptera). Can. Entomol. 84:336-342.

BROWN, W. J. 1958. Sibling species in the Chrysomel-idae. Proc. Tenth Int. Congr. Entomol. 1:103-109.

BROWN, W. J. 1959. Taxonomic problems with closelyrelated species. Annu. Rev. Entomol. 4:77-98.

CHEVERUD, J. M., J. L. LEWIS, W. BACHRACH, AND W.B. LEW. 1983. The measurement of form and vari-ation in form: An application of three-dimensionalquantitative morphology by finite-element methods.Am. J. Phys. Anthropol. 62:151-165.

CHEVERUD, J. M., AND J. T. RICHTSMEIER. 1986. Finite-element scaling applied to sexual dimorphism inrhesus macaque (Macaca mulatta) facial growth.Syst. Zool. 35:381-399.

DALY, H. V. 1985. Insect morphometrics. Annu. Rev.Entomol. 30:415^438.

DIEHL, S. R., AND G. L. BUSH. 1984. An evolutionaryand applied perspective of insect biotypes. Annu.Rev. Entomol. 29:471-504.

EBERHARD, W. G. 1985. Sexual selection and animalgenitalia. Harvard Univ. Press, Cambridge, Massa-chusetts.

FAIRBAIRN, D. J. 1990. Factors influencing sexual sizedimorphism in temperate water striders. Am. Nat.136:61-86.

FAIRBAIRN, D. J. 1992. The origins of allometry: Sizeand shape polymorphism in the common water-strider, Gerris remigis Say (Heteroptera, Gerridae).Biol. J. Linn. Soc. 45:167-186.

FAIRBARIN, D. J., AND R. F. PREZIOSI. 1994. Sexual se-lection and the evolution of allometry for sexual sizedimorphism in the water strider, Aquarius remigis.Am. Nat. 144:101-118.

FINK, W. L., AND M. L. ZELDITCH. 1995. Phylogeneticanalysis of ontogenetic shape transformations: Areassessment of the piranha genus Pygocentrus (Te-leostei). Syst. Biol. 44:344-361.

FUNK, D. J. 1996. The evolution of reproductive isola-tion in host-associated Neochlamisus leaf beetles: Arole for selection. Ph.D. Dissertation, Department ofEcology and Evolution, State Univ. New York, StonyBrook.

GILLHAM, M. C, AND M. F. CLARIDGE. 1994. A mul-tivariate approach to host plant associated morpho-logical variation in the polyphagous leafhopper, Al-netoidia alneti (Dahlbom). Biol. J. Linn. Soc. 53:127-151.

GOWER, J. C. 1971. Statistical methods of comparingdifferent multivariate analyses of the same data.Pages 138-149 in Mathematics in the archaeologicaland historical sciences (D. G. Kendall and P. Tautu,eds.). Edinburgh Univ. Press, Edinburgh.

GUTTMAN, S. I., T. K. WOOD, AND A. A. KARLIN. 1981.Genetic differentiation along host plant lines in thesympatric Enchenopa binotata Say complex. Evolution35:205-217.

HILLIS, D. M. 1987. Molecular versus morphological



approaches to systematics. Annu. Rev. Ecol. Syst. 18:23^2.

JAENIKE, J. J. 1981. Criteria for ascertaining the exis-tence of host races. Am. Nat. 117:830-834.

JAENIKE, J. J. 1990. Host specialization in phytopha-gous insects. Annu. Rev. Ecol. Syst. 21:243-273.

KARREN, J. B. 1966. A revision of the genus Exetna ofAmerica, north of Mexico (Chrysomelidae, Coleop-tera). Univ. Kans. Sci. Bull. 46:647-695.

KARREN, J. B. 1972. A revision of the subfamilyChlamisinae of America north of Mexico (Coleop-tera: Chrysomelidae). Univ. Kans. Sci. Bull. 49:875-988.

KENDALL, D. G. 1984. Shape manifolds, Procrusteanmetrics, and complex projective spaces. Bull. Lond.Math. Soc. 16:81-121.

LELE, S., AND J. T. RICHTSMEIER. 1991. Euclidean dis-tance matrix analysis: A coordinate free approachfor comparing biological shapes using landmarkdata. Am. J. Phys. Anthropol. 86:415-428.

LESAGE, L. 1984. Immature stages of Canadian Neo-chlamisus Karren (Coleoptera: Chrysomelidae). Can.Entomol. 116:383^09.

LOY, A., M. CORTI, AND L. F. MARCUS. 1993. Land-mark data: Size and shape analysis in systematics.A case study on Old World Talpidae (Mammalia,Insectivora). Pages 215-240 in Contributions to mor-phometrics, Volume 8 (L. F. Marcus, E. Bello, and A.Garcia-Valdecasas, eds.). Monografias del MuseoNacional de Ciencias Naturales (CSIC), Madrid.

MARCUS, L. F. 1990. Traditional morphometrics.Pages 95-130 in Proceedings of the Michigan mor-phometrics workshop (F. J. Rohlf and F. L. Book-stein, eds.). Univ. Mich. Mus. Zool. Spec. Publ. 2.

MARCUS, L. F. 1993. Some aspects of multivariate sta-tistics for morphometrics. Pages 77-122 in Contri-butions to morphometrics. Volume 8 (L. F. Marcus,E. Bello, and A. Garcia-Valdecasas, eds.). Monogra-fias del Museo Nacional de Ciencias Naturales(CSIC), Madrid.

MAYR, E. 1942. Systematics and the origin of species.Columbia Univ. Press, New York.

MEACHAM, C, AND T. DUNCAN. 1990. MorphoSys,version 1.29. Univ. California, Berkeley.

MENKEN, S. B. J. 1989. Electrophoretic studies on geo-graphic populations, host races, and sibling speciesof insect pests. Pages 181-202 in Electrophoreticstudies on agricultural pests (H. D. Loxdale and J.Den Hollander, eds.). Oxford Univ. Press, Oxford,England.

MEYER, A. 1987. Phenotypic plasticity and hetero-chrony in Cichlasoma managuense (Pisces, Cichlidae)and their implications for speciation in cichlid fish-es. Evolution 41:1357-1369.

PATTERSON, C, D. M. WILLIAMS, AND C. J. HUMPHRIES.1993. Congruence between molecular and morpho-logical phylogenies. Annu. Rev. Ecol. Syst. 24:153-188.

PRENTER, J., W. I. MONTGOMERY, AND R. W. ELVVOOD.1995. Multivariate morphometrics and sexual di-morphism in the orb-web spider Metillina segmentata(Clerck, 1757) (Araneae, Metidae). Biol. J. Linn. Soc.55:345-354.

PREZIOSI, R. E, D. J. FAIRBAIRN, D. A. ROFF, AND J. M.BRENNAN. 1996. Body size and fecundity in the wa-terstrider Aquarius remigis: A test of Darwin's fecun-dity advantage model. Oecologia 108:424-431.

RICHTSMEIER, J. T, J. M. CHEVERUD, AND S. LELE. 1992.Advances in anthropological morphometrics. Annu.Rev. Anthropol. 21:283-305.

ROHLF, F. J. 1990. Fitting curves to outlines. Pages167-177 in Proceedings of the Michigan morpho-metrics workshop (F. J. Rohlf and F. L. Bookstein,eds.). Univ. Mich. Mus. Zool. Spec. Publ. 2.

ROHLF, F. J. 1993a. NTSYS-pc, version 1.80. ExeterSoftware, Setauket, New York.

ROHLF, F. J. 1993b. Relative warp analysis and an ex-ample of its application to mosquito wings. Pages131-159 in Contributions to morphometrics, Volume8 (L. F. Marcus, E. Bello, and A. Garcia-Valdecasas,eds.). Monografias del Museo Nacional de CienciasNaturales (CSIC), Madrid.

ROHLF, F. J., A. LOY, AND M. CORTI. 1996. Morpho-metric analysis of Old World Talpidae (Mammalia,Insectivora) using partial-warp scores. Syst. Biol. 45:344-362.

ROHLF, F. J., AND L. F. MARCUS. 1993. A revolution inmorphometrics. Trends Ecol. Evol. 8:129-132.

ROHLF, F. J., AND D. E. SLICE. 1990. Extensions of theProcrustes method for the optimal superimpositionof landmarks. Syst. Zool. 39:40-59.

ROININEN, H., J. VUORINEN, J. TAHVANAINEN, AND R.

JULKUNEN-TIITTO. 1993. Host-preference and allo-zyme differentiation in shoot galling sawfly, Euuraatra. Evolution 47:300-308.

SIEGEL, A. E, AND R. H. BENSON. 1982. A robust com-parison of biological shapes. Biometrics 38:341-350.

SLICE, D. E. 1993. Extensions, comparisons, and ap-plications of superimposition methods for morpho-metric analysis. Ph.D. Dissertation, Department ofEcology and Evolution, State Univ. New York, StonyBrook.

SLICE, D. E. 1994. GRF-ND. Generalized rotational fit-ting of n-dimensional landmark data. Departmentof Ecology and Evolution, State Univ. New York,Stony Brook.

SNEATH, P. H. A. 1967. Trend-surface analysis oftransformation grids. J. Zool. Lond. 151:65-122.

SNEATH, P. H. A. AND R. R. SOKAL. 1973. Numericaltaxonomy. W. H. Freeman, New York.

SOKAL, R. RV AND F. J. ROHLF. 1995. Biometry, 3rdedition. W. H. Freeman, New York.

STRAUSS, R. E., AND F. L. BOOKSTEIN. 1982. The truss:Body form reconstruction in morphometrics. Syst.Zool. 31:169-173.

SWIDERSKI, D. L. 1993. Morphological evolution of thescapula in tree squirrels, chipmunks, and groundsquirrels (Sciuridae): An analysis using thin-platesplines. Evolution 47:1854-1873.

THOMPSON, D. W. 1917. On growth and form. Cam-bridge Univ. Press, Cambridge, England.

WALKER, J. A. 1993. Ontogenetic allometry of three-spine stickleback body form using landmark-basedmorphometrics. Pages 193-214 in Contributions tomorphometrics, Volume 8 (L. F. Marcus, E. Bello,and A. Garcia-Valdecasas, eds.). Monografias del



Museo National de Ciencias Naturales (CSIC), Ma-drid.

WALKER, J. A. 1996. Principal components of bodyshape variation within an endemic radiation ofthreespine stickleback. Pages 321-334 in Advancesin morphometrics (L. F. Marcus, M. Corti, A. Loy,G. J. P. Naylor, and D. E. Slice, eds.). Plenum, NewYork.

WALSH, B. D. 1864. On phytophagic varieties andphytophagic species. Proc. Entomol. Soc. Phila. 3:403-430.

WARING, G. L., W. G. ABRAHAMSON, AND D. J. HOW-ARD. 1990. Genetic differentiation among host-as-sociated populations of the gallmaker, Eurosta soli-daginis (Diptera: Tephritidae). Evolution 44:1648-1655.

WICKMAN, P.-Q, AND B. KARLSSON. 1989. Abdomen

size, body size and the reproductive effort of in-sects. Oikos 56:209-214.

ZELDITCH, M. L., F. L. BOOKSTEIN, AND B. L. LUNDRI-GAN. 1992. Ontogeny of integrated skull growth inthe cotton rat Sigmodon fulviventer. Evolution 46:1164-1180.

ZELDITCH, M. L., AND W. L. FINK. 1995. Allometryand developmental integration of body growth in apiranha, Pygocentrus nattereri (Teleostei: Ostariophy-si). J. Morphol. 223:341-355.

ZELDITCH, M. L., W. L. FINK, AND D. L. SWIDERSKI.1995. Morphometrics, homology, and phylogenet-ics: Quantified characters as synapomorphies. Syst.Biol. 44:179-189.

Received 11 December 1995; accepted 15 September 1996Associate Editor: Paula Mabee


morphometric inferences on sibling species and …€¦ · syst. biol. 46(l):180-194/ 1997...

Documents