santiago-alarcon&parker2007

8/3/2019 Santiago-Alarcon&Parker2007

1/11

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofitpublishers, academic institutions, research libraries, and research funders in the common goal of maximizing access tocritical research.

SEXUAL SIZE DIMORPHISM AND MORPHOLOGICALEVIDENCE SUPPORTING THE RECOGNITION OF TWOSUBSPECIES IN THE GALPAGOS DOVE

Author(s): DIEGO SANTIAGO-ALARCON and PATRICIA G. PARKERSource: The Condor, 109(1):132-141. 2007.Published By: Cooper Ornithological Society

DOI: 10.1650/0010-5422(2007)109[132:SSDAME]2.0.CO;2URL:http://www.bioone.org/doi/full/10.1650/0010-5422%282007%29109%5B132%3ASSDAME%5D2.0.CO%3B2

BioOne ( www.bioone.org ) is an electronic aggregator of bioscience research content,and the online home to over 160 journals and books published by not-for-profit societies,

associations, museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associatedcontent ind icates your acceptance of BioOnes Terms of Use, available atwww.bioone.org/page/terms_of_use .

Usage of BioOne content is strictly limited to personal, educational, and non-commercialuse. Commercial inquiries or rights and permissions requests should be directed to theindividual publisher as copyright holder.
http://www.bioone.org/page/terms_of_usehttp://www.bioone.org/page/terms_of_usehttp://www.bioone.org/http://www.bioone.org/doi/full/10.1650/0010-5422%282007%29109%5B132%3ASSDAME%5D2.0.CO%3B2http://www.bioone.org/doi/full/10.1650/0010-5422%282007%29109%5B132%3ASSDAME%5D2.0.CO%3B2


2/11

SEXUAL SIZE DIMORPHISM AND MORPHOLOGICALEVIDENCE SUPPORTING THE RECOGNITION OF TWO

SUBSPECIES IN THE GALA PAGOS DOVE

D IEGO SANTIAGO -A LARCON 1 AND PATRICIA G. P ARKERDept. of Biology and Whitney R. Harris World Ecology Center, University of Missouri-St.Louis, One University

Blvd., St.Louis, MO 63121

Abstract. Sexual size dimorphism is a conspicuous trait of many wild bird species.Differences in body size between the sexes might reflect selective pressures and trade-offsto optimize performance. Here, we analyze the size dimorphism of the Galapagos Dove(Zenaida galapagoensis ) using principal component and discriminant analyses withsamples obtained from six islands: Santiago, Santa Fe, Santa Cruz, Espanola, Genovesa,and Wolf. We also reanalyze published morphological data but also including additionalsamples from Wolf Island to account for morphological differences among islands. Maleswere significantly larger than females. Discriminant analyses correctly classified 98% of

males and 100% of females, and cross-validation of the model correctly classified 97% of males and 98% of females. We created two sexual size dimorphism indices using wingchord and tarsus as body-size surrogates. Significant differences were found in the sexualsize dimorphism index for both measurements among islands. Significant differences insexual size dimorphism among islands might indicate the role of different selectivepressures acting on individual islands (e.g., competition, predation, resources, sexualselection), which might result in life history variation of the species among islands. For thefirst time, we provide significant morphological evidence supporting the classification of the Galapagos Dove into two subspecies: Z. g. galapagoensis and Z. g. exsul .

Key words: Galapagos, morphology, sexual size dimorphism, Zenaida galapagoensis .

Dimorfismo Sexual de Tamano y Evidencia Morfologica que Apoya la Separacion deZenaida galapagoensis en Dos Subespecies

Resumen. El dimorfismo sexual de tamano es una caracter stica comun de muchasespecies de aves. Las diferencias en el tamano corporal entre machos y hembras puedereflejar la accion de factores de seleccion y soluciones de costo-beneficio que actuan paraoptimizar el rendimiento de las aves en su ambiente. Analizamos el dimorfismo de tamanode Zenaida galapagoensis por medio de analisis de componentes principales y analisis dediscriminacion utilizando muestras colectadas en seis islas: Santiago, Santa Fe, SantaCruz, Espanola, Genovesa y Wolf. Tambien reanalizamos informacion morfologicapreviamente publicada, en donde incluimos muestras provenientes de la isla Wolf la cualno estuvo comprendida en el esudio anterior, para analizar diferencias morfologicas entrelas islas. Los machos fueron significativamente mas grandes que las hembras. El analisis dediscriminacion clasifico correctamente el 98% de los machos y el 100% de las hembras, elanalisis de validacion del modelo de discriminacion clasifico correctamente el 97% de losmachos y el 98% de las hembras. Creamos un ndice de dimorfismo sexual de tamano

utilizando la cuerda alar y el tarso como medidas de tamano corporal. Se encontrarondiferencias significativas en el dimorfismo sexual de tamano entre algunos pares de islaspara ambas medidas. Diferencias significativas en el dimorfismo sexual de tamano puedenindicar la accion de diferentes presiones selectivas en diferentes islas (e.g., competicion,depredacion, recursos, seleccion sexual), lo cual puede crear variaciones en la biolog a dela especie entre las distintas islas. Por primera vez, proveemos resultados morfologicossignificativos que apoyan la separacion de Z. galapagoensis en dos subespecies: Z. g. galapagoensis y Z. g. exsul .

INTRODUCTION

Animal life-history traits are strongly influ-

enced by body size. Sexual dimorphism within

species (e.g., plumage color or body size) isa widespread phenomenon throughout theanimal kingdom (Butler et al. 2000). Differentbody sizes reflect trade-offs and selectivepressures (e.g., environmental conditions, sexu-al selection) acting on different populations andbetween the sexes (Wikelski and Trillmich 1997,

Manuscript received 17 January 2006; accepted 23October 2006.

1 E-mail: [email protected]

The Condor 109:132141# The Cooper Ornithological Society 2007

[132]


3/11

Owens and Hartley 1998). In general, there isa strong positive association between sexual sizedimorphism and individual body size (Fair-bairn 1997, Wikelski and Trillmich 1997).Apparently, this relationship results from theadvantage of being large, with the sexesbenefiting in different ways as size increases.Males may acquire priority access to foodresources, better territories, and increasedmating success; females may acquire higherfertility (Fairbairn 1997, Wikelski and Trillmich1997).

Geographic isolation of populations is ex-pected to influence differentiation of bothmorphological and genetic characters, due to

drift or different selective regimes (Slatkin 1985,Hutchison and Templeton 1999, Coleman andAbbott 2003). Santiago-Alarcon et al. (2006)showed that males and females of the Galapa-gos Dove ( Zenaida galapagoensis ) are morpho-logically similar among five islands of theGalapagos archipelago (Santa Fe, Santa Cruz,Santiago, Espanola, and Genovesa). High geneflow is apparent among these islands (Santiago-Alarcon et al. 2006); hence, significant differ-ences in sexual size dimorphism among islands

would not be expected. However, as Wikelskiand Trillmich (1997) showed in lava lizards of the archipelago, sexual size dimorphism mayvary significantly among populations locatedwithin the same island group and in closegeographical proximity, possibly due to highvariation in microclimatic conditions on differ-ent islands (Wikelski and Trillmich 1997).

The Galapagos Dove is an endemic specieswhose biology is poorly known. Our knowledgeabout this species is restricted to taxonomicrelationships (Goodwin 1977, Johnson andClayton 2000), morphological records (Ridg-way 1897, Gifford 1913, Prestwich 1959), andsome aspects of its breeding and feedingecology on Genovesa Island (Grant and Grant1979). Morphological and ecological studies of birds in the Galapagos archipelago have beenconfined to few species, including Darwinsfinches (Bowman 1961, Boag 1981, 1983, Grantet al. 1985, Grant 2001), Galapagos mocking-birds ( Nesomimus spp.; Curry 1988, 1989,Curry and Grant 1989), and the GalapagosHawk ( Buteo galapagoensis ; Bollmer et al. 2003,2005). Measurements and general descriptionsof Galapagos Doves are given by Ridgway(1897), Gifford (1913), and Swarth (1931).

These authors provide measurements for a lim-ited number of individuals and do not presentany rigorous statistical analysis in support of their claims for two different subspecies of thedove. Color patterns in this species are the samefor both sexes, although females tend to beduller than males. However, differences in colorintensity are not obvious and can vary in-dividually, making sex difficult to determine.Juveniles have a different color pattern thanadults and are easily distinguished. Someaspects of the biology of sexually monochro-matic bird species (i.e., the same or very similarplumage coloration between the sexes) aredifficult to study in the wild because sexes are

identical (Winker et al. 1996).We analyzed the sexual size dimorphism of

the Galapagos Dove using principal componentand discriminant analyses, created two sexualsize dimorphism indices, and evaluated whetherthere were significant differences across islandsby reanalyzing the morphological data inSantiago-Alarcon et al. (2006) but includingsamples from Wolf Island, individuals fromwhich represent what is considered a differentsubspecies of the Galapagos Dove (Swarth

1931, Grant and Grant 1979). The specificquestions addressed in this s tudy are:(1) whether there is sexual dimorphism inbody size of the endemic Galapagos Dove;(2) whether there are differences among islandpopulations in the degree of sexual dimor-phism; and (3) whether there are morphologicaldifferences that separate the two subspecies.

METHODS

FIELD METHODS

We conducted this study in the Galapagosarchipelago from May through July 2002, fromJune through July 2004, and during July 2005.We sampled individuals using hand nets andmist nets following the guidelines in Ralph et al.(1996). We took blood samples (50 ml) byvenipuncture from 25 birds each from SantaCruz (15 males, 10 females), Santa Fe (15males, 10 females), and Espanola Islands (11males, 14 females), 30 each from Santiago (18males, 12 females) and Genovesa Islands (20males, 10 females), and 29 from Wolf Island (13males, 16 females; Fig. 1). Samples were mixedwith 500700 ml of Longmires lysis buffer(Longmire et al. 1988). We used a PCR-based

SIZE DIMORPHISM IN THE GALA PAGOS DOVE 133


4/11

technique for sexing individuals (Fridolfssonand Ellegren 1999). We visited San Cristobalduring 2002 and Darwin during 2005, but dueto small sample sizes ( n 5 2 and 4, respectively)these islands were not included in our analysis.To quantify intersexual differences in morphol-ogy, we took the following measurements to thenearest 0.1 mm from the right side of eachindividual: (1) tarsus (from the joint betweenthe tibiotarsus and the tarsometatarsus to thebent joint between the tarsometatarsus andmetatarsals); (2) tail (posterior base of uropy-gial gland to tip of central rectrices); (3) exposedculmen (from the tip of the feathering to thebills tip); (4) bill width (calipers were orientedat a 90 u angle to the axis of the bill and mea-surement was taken at the tip of the feathering);(5) bill depth (at the tip of the feathering and at90 u angle to the axis of the bill); and (6) wingchord, to the nearest 0.5 mm, using a ruler witha brass perpendicular stop (unflattened, fromcarpal joint to the tip of the longest primary).Mass was measured to the nearest 0.1 g usingPesola scales (100 g and 300 g). Raw morpho-logical measurements can be obtained by re-quest from DS-A. Measurements were taken byDS-A on all the islands but Santa Fe, where

Jennifer Bollmer (JB) conducted the sampling.Although JB worked with DS-A on otherislands and was instructed in dove measurementby DS-A, measurement error due to observervariability could create false differences amongislands. Hence, both DS-A and JB measuredRock Pigeon ( Columba livia ) and MourningDove ( Zenaida macroura ) individuals at theSaint Louis Zoo to assess observer variability.There were no significant differences betweenthese observers in any of the measured variables(same variables as measured in this study; Z $0.3, P $ 0.44). Wild birds were released within40 m of their capture location.

STATISTICAL ANALYSES

We used principal component analysis (PCA)to describe morphological variation betweenthe sexes. We used SPSS v. 11.0.1 for Windows(SPSS 2001) in all analyses. Although allvariables were normally distributed (Kolmo-gorov-Smirnov test, P $ 0.09) and had thesame scale and dimension (except mass), theywere log-transformed to examine the propor-tional contributions of large and small mea-surements equally. Principal component (PC)scores were normally distributed (Kolmogorov-

FIGURE 1. Map of the Galapagos archipelago, with islands on which Galapagos Doves were measuredshaded in gray. Wolf and Darwin islands are located 186 km and 239 km, respectively, northwest of thenorthern tip of Isabela Island.

134 DIEGO SANTIAGO-ALARCON AND PATRICIA G. PARKER


5/11

Smirnov test, P $ 0.86). All components witheigenvalues $ 1 were retained for subsequentanalyses. Eigenvectors were rotated using var-imax rotation. We retained rotated eigenvectorswhen the explained variance was higher thanthat of unrotated components or when theinterpretation of principal components waseasier. We used t-tests for group comparisonsof PC scores (males vs. females); t-tests wereindependent and two-tailed. Variances of PCscores were homogeneous among groups (Le-venes test, P . 0.32). In addition to thedimorphism analysis, we conducted two PCAs,one for males and one for females, to examinemorphological variation among islands. These

analyses used previously published data (San-tiago-Alarcon et al. 2006) with the addition of samples from Wolf Island, which were notavailable for analysis in the previous study.Individuals from Wolf Island represent what isconsidered a different subspecies of the Gala-pagos Dove ( Z. g. exsul ), which is restricted tothe northern islands of Darwin and Wolf. Weconducted the analyses describing morpholog-ical variation among islands separately by sexto prevent any variance due to sexual dimor-

phism from masking variation among popula-tions. Bill depth was excluded from the PCA forfemales because only one individual was mea-sured for this variable on one of the islands(Santiago). We used ANOVAs to test for differ-ences among islands. We used Tukey post-hoctests any time an ANOVA was significant.

We used discriminant analysis (DA) to createa discriminant function that complemented thePCA analysis for sexual dimorphism. Becausedoves from Wolf Island are significantly larger(see results) than those of the southern islands,we conducted separate discriminant analysesfor the southern subspecies ( Z. g. galapagoensis )and the northern subspecies ( Z. g. exsul ).Samples from Genovesa Island were takentwo years (2004) after the rest of the samples(2002), thus we decided to compute two dis-criminant analyses for the southern subspecies.One analysis excluded Genovesa samples,which subsequently were used to validate themodel; the other analysis used all individualsfrom the five sampled islands. This secondmodel was validated using the U -method (leave-one-out; McGarigal et al. 2000). In addition tothe normality and variance assumptions ful-filled for PCA and detection of outliers, we

checked for equality of variance-covariancematrices using Boxs test for the two DAanalyses ( M 5 5.5, df 5 6, P 5 0.51 and M 5 7.1, df 5 6, P 5 0.33, respectively). Multi-collinearity was checked by conducting pairwisecorrelations (culmen vs. tarsus: r 5 0.82, P 50.01; culmen vs. wing chord: r 5 0.77, P 5 0.01;and wing chord vs. tarsus: r 5 0.82, P 5 0.01;only correlations $ 0.7 are shown). We retainedwing chord and tarsus and excluded culmenfrom the analysis because wing chord andtarsus explained most of the variance betweengroups ( F 1,5 5 304.8, P , 0.001, and F 1,5 5448.2, P , 0.001, respectively). Even thoughtarsus and wing chord were highly correlated,they presented high tolerance during the vari-able selection process. In addition, we checkedfor agreement between standardized canonicalcoefficients and structure coefficients; variableweights were in the same rank order and hadequal signs, thus proving the adequacy of thegenerated models. Prior probabilities wereassumed to be equal because our sampleincluded similar numbers of males and females,both when the discriminant analysis modelincluded all samples and when Genovesa

individuals were excluded. We conducted step-wise backward discriminant function analysisusing Wilks lambda as the optimizationcriterion.

We validated the discriminant analysis fromWolf Island using the U -method (leave-one-out) and we also used the four individuals (twomales and two females) from Darwin Island,although we recognize the need for a muchlarger and independent data set to validate thismodel. We checked for equality of variance-

covariance matrices using Boxs test ( M 5 0.8,df 5 1, P 5 0.35). Multicollinearity waschecked by conducting pairwise correlations(only mass vs. tail had a correlation . 0.7 andboth variables were excluded from the anlysis).Prior probabilities were assumed to be equalbecause our sample included a similar numberof males and females.

We calculated a sexual size dimorphism index(Smith 1999) as:

male size =female size { 1:

Wing chord and tarsus were used as surrogatesfor body size to test whether there weresignificant differences in degree of sexual size



6/11

dimorphism among islands. Wing chord andtarsus proved to be the two variables that bestpredicted body size according to the principalcomponent analysis (high PC score and highextracted variance). Additionally, they were thevariables that most reduced Wilks lambdaduring discriminant analyses (DA). To increasesample sizes among islands for the calculationof the sexual size dimorphism index we used thePopTools 2.5.4 (Hood 2003) add-in for Micro-soft Excel to generate 100 randomizations foreach population before the index was calculat-ed. We took all individual measurements madeon males and females from each island and usedthem as source data to generate 100 random-ized measurements for each sex. We nextcalculated the sexual size dimorphism index asshown by the above formula for each of the 100measurement pairs and finally calculated theaverage of the 100 pairs. Because variances

were not homogeneous across populations, weconducted a Kruskal-Wallis test to evaluatedifferences among islands in the degree of sexual size dimorphism. Values are reported asmeans 6 SD.

RESULTS

We retained the first component from the PCAconducted to compare the sexes for statisticalanalysis. PC1 explained 73% of the variancewhereas PC2 explained 11%. PC1 describedoverall body size and the variance extractedfrom each variable was . 79% (Table 1). Maleswere significantly larger than females (PC1: t1555 2 9.3, P , 0.001; Fig. 2).

For statistical analysis of females, we alsoretained the first component of the PCA. PC1explained 66% of the variance and describedoverall body size. PC2 explained 14% of thevariance and was a compound size component

defined mainly by bill width, tail, and mass(Table 2). The variance extracted from eachvariable was . 76% (Table 2). We foundsignificant differences among islands ( F 5,61 520.4, P , 0.001). Genovesa Island females weresignificantly smaller than females on SantaCruz and Santiago Islands (Tukey test, HSD5 0.9 and 1.1, respectively, P # 0.01 for bothcomparisons); however, there was overlapamong individuals of these islands (Fig. 3A).Wolf Island females were significantly largerthan females from the other islands, clearlyforming a separate group in ordination spacefrom the rest of the female doves (Tukey test,2.2 . HSD . 1.2, P , 0.001 for allcomparisons; Fig. 3A).

For males, we retained the first componentfrom the PCA for statistical analysis. PC1explained 60% of the variance and PC2explained 12%. PC1 described overall body sizeand PC2 was a compound size componentdefined mainly by bill width, tail, and mass(Table 2). The variance extracted from eachvariable was . 63% (Table 2). We foundsignificant differences among islands ( F 5,73 59.4, P , 0.001). Males from Genovesa Islandwere significantly larger than doves on Santa Fe

FIGURE 2. Morphological ordination space be-

tween males and females of the Galapagos Dove.PC1 is an axis of overall body size and PC2 is a vectorof bill and tail size and mass. Males are larger thanfemales, and birds of both sexes are larger on Wolf Island (northern subspecies) than on the other islands(southern subspecies).

TABLE 1. Principal component (PC) scores andproportion of variance extracted (communalities)from each variable for the analysis of morpholo-gical differences between male and female GalapagosDoves. PC scores represent the correlation of eachvariable with the principal component andcommunalities are the sums of squares of thecorrelation coefficients on the first two componentsor the proportion of variance extracted from eachvariable.

Variable PC1 PC2 Communalities

Culmen 0.88 0.19 0.81Bill width 0.75 0.48 0.79Tarsus 0.89 0.18 0.84Tail 0.82 2 0.44 0.87Wing chord 0.93 2 0.02 0.87

Mass 0.852

0.37 0.86



7/11

Island (Tukey test, HSD 5 0.9, P 5 0.02);however, as in the case of females, there wasoverlap among individuals of these islands(Fig. 3B). Male doves on Wolf Island weresignificantly larger than males on the otherislands (Tukey test, 1.9 . HSD . 1.1, P #0.006 for all comparisons; Fig. 3B). As in thecase of females, Wolf males also formeda separate group in ordination space (Fig. 3B).

The discriminant analysis that excluded

Genovesa individuals retained wing chord(W), tarsus (T), and tail (Ta) as significantvariables discriminating between males andfemales ( l 5 0.17, x 23 5 105.9, P , 0.001).Tarsus had the highest discriminating power ( l5 0.22), followed by wing chord ( l 5 0.19), andtail ( l 5 0.17). The discriminant function was:

D ~ { 41:08 z T 0 :689

z W 0 :123 z Ta 0 :08 :

Individuals that scored zero or positive valueswere classified as males and those that scorednegative values were classed as females. Canon-ical correlation was high ( r 5 0.90). The modelcorrectly classified 94% of males and 100% of females (Fig. 4). Cross-validation by the U -method yielded 94% correct classification formales and 100% for females. When discrimi-nant scores for Genovesa individuals werecomputed with the above function, we obtained100% correct classification for both males andfemales.

The discriminant model including all samplesfrom the southern islands retained wing chord(W), tarsus (T), and bill width (BW) as signi-ficant variables discriminating between groups

FIGURE 3. (A) Morphological ordination spacebetween islands for adult female Galapagos Doves.PC1 is an axis of overall body size and PC2 is a vectorreflecting bill size and tarsus length. (B) Morpholog-ical ordination space between islands for maleGalapagos Doves. PC1 is an axis of overall bodysize and PC2 is a composite vector reflecting bill sizeand tarsus length. Note that the y axes of A and Bhave different scales. Birds from Wolf Island(northern subspecies, Z. g. galapagoensis ) are signif-icantly larger than birds on the southern islands(southern subspecies, Z. g. exsul ). There is broadoverlap among birds of the southern islands.

TABLE 2. Principal component (PC) scores and communalities (proportion of variance extracted) from theanalysis of morphological differences of male and female Galapagos Doves among six islands of theGalapagos archipelago. PC scores represent the correlation of each variable with the principal component andcommunalities represent the sums of squares of the correlation coefficients on the first two principalcomponents or the proportion of variance extracted from each variable.

Variable

Males Females

PC1 PC2 Communalities PC1 PC2 Communalities

Culmen 0.79 0.07 0.64 0.84 2 0.26 0.78Bill width 0.59 0.67 0.81 0.74 2 0.45 0.76Bill depth 0.80 0.37 0.77 Not included Not included Not includedTarsus 0.79 2 0.04 0.63 0.84 2 0.26 0.78Tail 0.77 2 0.36 0.73 0.72 0.54 0.82Wing chord 0.85 2 0.21 0.77 0.89 0.02 0.79Mass 0.77 2 0.33 0.71 0.80 0.45 0.85



8/11

(l 5 0.15, x 2 3 5 161.4, P , 0.001). As in theformer model, tarsus was the variable that bestdiscriminated between groups ( l 5 0.19),followed by wing chord ( l 5 0.16), and billwidth ( l 5 0.15). The discriminant functionwas:

D ~ { 45:24 z T 0:826 z W 0 :14 z BW 0 :722 :

Individuals that scored zero or positive valueswere classified as males and those that scorednegative values were categorized as females.Canonical correlation was high ( r 5 0.92). Themodel correctly classified 98% of males and100% of females (Fig. 4). Cross-validation bythe U -method yielded 97% correct classificationfor males and 98% for females.

The discriminant analysis for individualssampled on Wolf Island retained only tarsus(T) as a significant variable discriminating

between males and females ( l 5 0.14, x 2 1 552.6, P , 0.001). The discriminant functionwas:

D ~ { 141 :27 z T 94 :77 :

Individuals with scores of zero or positivevalues were classified as males and thosescoring negative values were classed as females.Canonical correlation was high ( r 5 0.93). Themodel correctly classified 100% of males and100% of females. Cross-validation by the U -method yielded 100% correct classification formales and females. When discriminant scoresfor Darwin individuals were computed with theabove function, we obtained 100% correctclassification for both males and females. Afterconducting a discriminant analysis that includ-ed the samples from all six islands, the correctclassification of males and females by the modeldecreased to around 80%. This result was dueto the inclusion of individuals from Wolf Island, which are significantly larger thanindividuals from other islands, as shown bythe PCA.

We found significant differences amongisland populations in the degree of sexual sizedimorphism for both wing chord (Kruskal-Wallis test: x 2 4 5 37.8, P , 0.001) and tarsus(Kruskal-Wallis test: x 2 4 5 29.3, P , 0.001)indices. Differences for the wing chord index

FIGURE 4. Scatter plot showing the two variables with the highest discriminating power (tarsus length andwing chord) for both males and females of the Galapagos Dove by island. Ellipses indicate individuals fromWolf Island. Inset is the box-plot of discriminant (DA) scores for males and females for the model thatincluded all samples of the southern subspecies ( Z. g. galapagoensis ) of the Galapagos Dove. Birds from Wolf Island are larger than birds on the southern islands for both males and females. This result is supported by thediscriminant analyses, which suggests that tarsus length and wing chord measurements are enough toconfidently separate males from females.



9/11

were identified between Genovesa and all theother islands (Genovesa 5 0.116 6 0.036,Santiago 5 0.083 6 0.034, Santa Cruz 50.095 6 0.030, Espanola 5 0.090 6 0.041,Santa Fe 5 0.098 6 0.042, Wolf 5 0.088 60.053). In the case of the tarsus index,differences were found between Santiago andWolf Islands, and between Santa Cruz vs. SantaFe, Genovesa, and Wolf Islands (Genovesa 50.120 6 0.028, Santiago 5 0.108 6 0.029, SantaCruz 5 0.100 6 0.044, Espanola 5 0.115 60.042, Santa Fe 5 0.120 6 0.060, Wolf 5 0.1296 0.040). We plotted sexual size dimorphism vs.body size for both males and females by island(Fig. 5) and found no positive correlationsbetween the two measurements. The samplefrom Wolf Island, in particular, had low sexualsize dimorphism relative to individual size.

DISCUSSION

Our finding of significant morphological differ-ences between the sexes is in agreement withmorphological records for other Zenaida spe-cies, in which males are generally larger thanfemales (Baptista et al. 1997). The threediscriminant models developed in this studysupported the results obtained by the principalcomponent analysis. Importantly, the efficiencyof the discriminant model was reduced whensamples from Wolf Island (northern subspecies)were included in the analysis, which supportsobserved morphological differences between

individuals from Wolf and from the otherislands.

This is the first study to provide statisticallysignificant morphological evidence that sup-ports the separation of the Galapagos Doveinto two subspecies, as has been previouslysuggested by other authors (Ridgway 1897,Gifford 1913, Swarth 1931). This result suggeststhat populations on Wolf and Darwin islandsmight be somewhat isolated from the otherislands, which is supported by significantpositive F ST values (based on five microsatelliteloci; see Santiago-Alarcon et al. [2006] for anexplanation of the molecular and geneticanalyses used) and lower gene flow between

northern and southern islands as estimated withprogram MIGRATE (DS-A and PGP, unpubl.data). In contrast, there is evidence of high geneflow among populations of the southern sub-species based on microsatellite markers (San-tiago-Alarcon et al. 2006).

The sexual size dimorphism wing chord indexcalculated for Genovesa doves was significantlyhigher than indices calculated for the otherislands. Because of high gene flow amongsouthern island populations of the Galapagos

Dove (Santiago-Alarcon et. al. 2006), morpho-logical differences between the sexes on thisisland might be the result of phenotypicplasticity in response to different local condi-tions (Wikelski and Trillmich 1997). GenovesaIsland is unique in the Galapagos archipelagoin that the top predator, the Galapagos Hawk,has been replaced by the Short-eared Owl ( Asio flammeus galapagoensis ). Thus, variation inpredation regimes on Genovesa due to differenthunting strategies and other life history traits of the Short-eared Owl compared to the Galapa-gos Hawk, coupled with abiotic conditions,might be generating different pressures on thedove population of Genovesa.

In addition, we identified significant differ-ences between some island pairs for the sexualsize dimorphism tarsus index. It has been sug-gested that tarsus size depends on the foragingstrategy of a species (Miles and Ricklefs 1984).If the two sexes of a species have differentecological roles, some of their morphologicaltraits might vary. Thus, differences in thissexual size dimorphism index between someislands might be due to differences in habitatstructure and ecological interactions. For ex-ample, inhabited islands such as Santa Cruz

FIGURE 5. Scatter plot showing the sexual sizedimorphism (SSD) wing index against the averagewing chord (body size surrogate) for both male andfemale Galapagos Doves. The lack of a positiverelationship between SSD and body size might reflectdifferences in selection pressures for both males andfemales on the different islands.



10/11

have been altered and are composed of habitatsnot found on other uninhabited islands. Thesedifferences among islands might represent areasof strong selection pressure for endemic doves.Nonetheless, because of high gene flow amongsouthern island populations, observed differ-ences are more likely the result of phenotypicplasticity.

The ecological causes of sexual size dimor-phism in the Galapagos Dove are difficult toidentify because there are limited data for thisspecies (Grant and Grant 1979). However,morphological differences are likely not theresult of a single factor, but rather the result of multiple interacting factors (Moore 1990,

Owens and Hartley 1998, Clegg and Owens2002).

ACKNOWLEDGMENTSWe thank all the people who provided help duringthe different stages of the field season, particularly N.Whiteman, J. Bollmer, G. Jimenez, J. Merkel, J.Rabenold, and N. Gottdenker. We are grateful to thestaff of the Charles Darwin Research Station fortheir invaluable help and logistical support during thecourse of this study, especially P. Robayo. We thankN. Cuevas and J. Freire who helped with dovesampling at Tortuga Bay, Santa Cruz. Permits forsample collection were provided by the GalapagosNational Park authorities. B. Loiselle, R. Ricklefs,and two anonymous reviewers made helpful com-ments and suggestions on earlier versions of thismanuscript. S. Siers helped in the preparation of thefigures. Financial support was provided by TheWhitney R. Harris World Ecology Center, The SaintLouis Zoo FRC # 05-2, and E. Des Lee Collabora-tive Vision in Zoological Research.

LITERATURE CITEDBAPTISTA , L. F., P. W. T RAIL , AND H. M. H ORBLIT .

1997. Family Columbidae (pigeons and doves),p. 60243. In J. del Hoyo, A. Elliot, and J.Sargatal [ EDS. ], Handbook of the birds of theworld. Vol. 4. Sandgrouse to cuckoos. LynxEdicions, Barcelona.

BOAG , P. T. 1981. Morphological variation in theDarwins finches (Geospizinae) of Daphne Ma- jor Island, Galapagos. Ph.D. dissertation,McGill University, Montreal.

BOAG , P. T. 1983. The heritability of externalmorphology in Darwins ground finches ( Geo-spiza ) on Isla Daphne Major, Galapagos. Evo-lution 37:877894.

BOLLMER , J. L., T. S ANCHEZ , M. D. C ANNON , D.SANCHEZ , B. C ANNON , J. C. B EDNARZ , T. DEVRIES , M. S. S TRUVE , AND P. G. P ARKER . 2003.Variation in morphology and mating systemamong island populations of Galapagos Hawks.Condor 105:428438.

BOLLMER , J. L., N. K. W HITEMAN , M. D. C ANNON ,J. C. B EDNARZ , T. DE VRIES , AND P. G.P ARKER . 2005. Population genetics of theGalapagos Hawk: genetic monomorphism with-in isolated populations. Auk 122:12101224.

BOWMAN , R. L. 1961. Morphological differentiationand adaptation in the Galapagos finches. Uni-versity of California Publications in Zoology58:1302.

BUTLER , M. A., T. W. S CHOENER , AND J. B. L OSOS .2000. The relationship between sexual size di-morphism and habitat use in Greater AntilleanAnolis lizards. Evolution 54:259272.

CLEGG , S. M., AND I. P. F. O WENS . 2002. The islandrule in birds: medium body size and itsecological explanation. Proceedings of the RoyalSociety of London Series B 269:13591365.

COLEMAN , M., AND R. J. A BBOTT . 2003. Possible

causes of morphological variation in an endemicMoroccan groundsel ( Senecio leucanthemifoliusvar. casablancae ): evidence from chloroplastDNA and random amplified polymorphicDNA markers. Molecular Ecology 12:423434.

CURRY , R. L. 1988. Group structure, within groupconflict and reproductive tactics in cooperativelybreeding Galapagos Mockingbirds, Nesomimus parvulus . Animal Behaviour 36:17081728.

CURRY , R. L. 1989. Geographic variation in socialorganization of Galapagos mockingbirds: eco-logical correlates of group territoriality andcooperative breeding. Behavioral Ecology andSociobiology 25:147160.

CURRY , R. L., AND P. R. G RANT . 1989. Demographyof the cooperatively breeding Galapagos Mock-ingbird, Nesomimus parvulus , in a climaticallyvariable environment. Journal of Animal Ecol-ogy 58:441463.

F AIRBAIRN , D. J. 1997. Allometry for sexual sizedimorphism: pattern and process in the co-evolution of body size in males and females.Annual Review of Ecology and Systematics28:659687.

F RIDOLFSSON , A. K., AND H. E LLEGREN . 1999. Asimple and universal method for molecularsexing of non-ratite birds. Journal of AvianBiology 30:116121.

G IFFORD , E. W. 1913. Expedition of the CaliforniaAcademy of Sciences to the Galapagos Islands,19051906. Proceedings of the California Acad-emy of Sciences 2:1132.

G OODWIN , D. 1977. Pigeons and doves of the world.Comstock, Cornell University Press, Ithaca, NY.

G RANT , P. R. 2001. Reconstructing the evolution of birds on islands: 100 years of reseach. Oikos92:385403.

G RANT , P. R . , I . A BBOTT , D . S CHLUTER , R . L .CURRY , AND L. K. A BBOTT . 1985. Variation inthe size and shape of Darwins finches. Bi-ological Journal of the Linnean Society 25:139.

G RANT , P. R., AND K. T. G RANT . 1979. Breedingand feeding ecology of the Galapagos Dove.Condor 81:397403.

H OOD , G . [ONLINE ]. 2003. PopTools. 5 www.cse.csiro.au/poptools/ 4 (30 May 2006).



11/11

H UTCHISON , D. W., AND A. R. T EMPLETON . 1999.Correlation of pairwise genetic and geographicdistance measures: inferring the relative influ-ences of gene flow and drift on the distributionof genetic variability. Evolution 53:18981914.

JOHNSON , K. P., AND D. H. C LAYTON . 2000. Amolecular phylogeny of the dove genus Zenaida :mitochondrial and nuclear DNA sequences.Condor 102:864870.

LONGMIRE , J. L., A. K. L EWIS , N. C. B ROWN , J. M.BUCKINGHAM , L. M. C LARK , M. D. J ONES , L. J.M EINCKE , J. M EYNE , R. L. R ATLIFF , F. A. R AY ,R. P. W AGNER , AND R. K. M OYZIS . 1988.Isolation and characterization of a highly poly-morphic centromeric tandem repeat in the familyFalconidae. Genomics 2:1424.

M CG ARIGAL , K., S. C USHMAN , AND S. STAFFORD .2000. Multivariate statistics for wildlife andecology research. Springer-Verlag, New York.

M ILES , D. B., AND R. E . R ICKLEFS . 1984. Thecorrelation between ecology and morphology indeciduous forest passerine birds. Ecology65:16291640.

M OORE , A. J. 1990. The evolution of sexual di-morphism by sexual selection: the separateeffects of intrasexual selection and intersexualselection. Evolution 44:315331.

OWENS , I. P. F., AND I. R. H ARTLEY . 1998. Sexualdimorphism in birds: why are there so manydifferent forms of dimorphism? Proceedings of the Royal Society of London Series B265:397407.

PRESTWICH , A. A. 1959. The Galapagos Dove infreedom and captivity. Avicultural Magazine65:6676.

R ALPH , C. J., G. R. G EUPEL , P. P YLE , T. E. M ARTIN ,D. F. D ESANTE , AND B. M ILA . 1996. Manual demetodos de campo para el monitoreo de avesterrestres. USDA Forest Service General Tech-nical Report PSW-GTR-159.

R IDGWAY , R. 1897. Birds of the Galapagos archi-pelago. Proceedings of the United States Na-tional Museum 19:614618.

SANTIAGO -A LARCON , D . , S . M. T ANKSLEY , ANDP. G. P ARKER . 2006. Morphological variationand genetic structure of Galapagos Dove ( Ze-naida galapagoensis ) populations: issues in con-servation for the Galapagos bird fauna. WilsonJournal of Ornithology 118:194207.

SLATKIN , M. 1985. Gene flow in natural populations.Annual Review of Ecology and Systematics16:393430.

SMITH , R. J. 1999. Statistics of sexual size dimor-

phism. Journal of Human Evolution 36:423 459.SPSS. 2001. SPSS release 11.0.1. SPSS, Inc., Chicago.SWARTH , H. S. 1931. The avifauna of the Galapagos

Islands. Occasional Papers of the CaliforniaAcademy of Sciences 18:1299.

W IKELSKI , M., AND F. T RILLMICH . 1997. Body sizeand sexual size dimorphism in marine iguanasfluctuate as a result of opposing natural andsexual selection: an island comparison. Evolu-tion 51:922936.

W INKER , K., J. T. K LICKA , AND G. V OELKER . 1996.Sexual size dimorphism in birds from southernVeracruz, Mexico. II. Thryothorus maculipectusand Henicorhina [leucosticta ] prostheleuca . Jour-nal of Field Ornithology 67:236251.


santiago-alarcon&parker2007

Documents