Vanda Pribacic Ambrozic, Lajos Szirovicza, Josko Sindik, Pavao Rudan: Structures of morphological characteristics oof the population of the island of Rab Structures of morphological characteristics of the population of the island of Rab Vanda Pribacic Ambrozic Institute for Anthropological Research, vpribacic@inantro.hr Lajos Szirovicza University of Zagreb, lajos.zgb@gmail.com Josko Sindik Institute for Anthropological Research, jsindik@inantro.hr Pavao Rudan Croatian Academy of Sciences and Arts, Institute for Anthropological Research, rudan@hazu.hr Abstract The focus of this article is the part of a holistic anthropological interdisciplinary study of the population structure of the inhabitants of the Island of Rab, focused on multivariate analyses of latent morphological variables, to contribute to the explanation of its microevolution. The goals of the study were directed to determine population differences in a geographically limited area of Rab, based on the latent morphological variables of the head and body. A representative sample of 601 adult persons (aged 18-75 years) from the settlements of Banjol, Barbat, Lopar, Rab and Supetarska Draga was examined. The analysis included 30 anthropometric variables of the body, as well as 14 anthropometric variables of the head. Principal Component Analysis with Varimax rotation revealed five latent variables of the body and five latent variables of the head (in both sexes), while quasi-confirmatory factor analysis proved that there are no significant differences in component structures among men and women. Analysing the latent morphological variables enabled the clear insight that the examined population is heterogeneous, in the space of the investigated body and head variables for both sexes in certain sets of variables. KEYWORDS: anthropometry, population structure, Rab ANTHROPOLOGICAL NOTEBOOKS 21 (2): 41-54. ISSN 1408-032X © Slovene Anthropological Society 2015 41 Anthropological Notebooks, XXI/2, 2015 Introduction The concept of population structure is an attempt to describe the model of population changes or evolution more realistically. The scope of this study is to explore complex phenotypic morphological traits, as well as to determine the differences among people who live in a quite small geographic area of the northern Adriatic island of Rab. Exploring the population structure can be also viewed as the study of genetic differentiation within the network of small or limited populations, including small deviations from the Hardy-Weinberg balance caused by size, non-random recruitment or subdivision (Pribacic Ambrozic et al. 2009). Over the course of evolution, various human populations have developed in various parts of the world, exposed to the various pressures of environmental factors. Through the activity of the evolutionary forces of natural selection, genetic drift, gene flow and mutations, these populations have formed their distinctive characteristics (Pribacic Ambrozic et al. 2009). Having in mind the widespread migration pressures to which European populations were exposed during the 20th century, it is hardly impossible to find populations that already remain in reproductive isolation. Therefore, evolutional processes are the best studied in small, reproductively isolated communities (Wright 1931, 1943, 1951). In small local communities, the inbreeding phenomena and gene drift become evident (Rudan et al. 1987a, 1987b, 1987c, 1990a, 1990b), so the effective reproductive size of a population substantially affects the changes in gene frequencies. These island populations' isolates are very convenient for investigating theoretical hypotheses about microevolution (Rudan et al. 1998). Because of their features, the Adriatic islands' isolates have been the issues of numerous anthropological and genetic research projects over the previous 40 years (Rudan 1978; Rudan et al. 1979). The investigations of the inhabitants in the whole region of Middle Dalmatia started in 1972, with special emphasis on the particular populations of the islands of Hvar, Korcula, Peljesac, Brac, Pag, Silba, Krk, and Vis (Rudan 1980, 1982; Rudan et al. 1982, 1986, 1988, 1999; Pavicic, 2004). The research focused mainly on morphological characteristics gave valuable data on the biological structure of the Middle Dalmatian population, as well as of other Adriatic isolates, and are providing information on the course of their biological microevolution and their social and cultural features (Rudan et al. 1992, 1994; Smolej-Narancic et al. 1987, 1994). Morphological variations are confirmed in accordance with currently known ethnological and demographic data on migration movements of the inhabitants on this territory in the past and present (Pribacic Ambrozic et al. 2009). The studies of morphological (anthropometric) characteristics of certain populations in Middle Dalmatia are also useful in order to examine the degree and type of morphological variations among populations, for example on the principle of geographical division among the western and eastern parts of the island (on larger islands). The second principle of the analysis of differences is given on the level of populations placed on separate locations (one settlement on the island: Silba and Olib) (Pribacic Ambrozic et al. 2009). Previous research conducted using the same data was performed using univariate analysis on certain morphological variables of the body and head among populations on 42 Vanda Pribacic Ambrozic, Lajos Szirovicza, Josko Sindik, Pavao Rudan: Structures of morphological characteristics oof the population of the island of Rab the island Rab, separately for two sexes. After analysing 36 body variables, the results revealed the existence of statistically significant heterogeneities in 13 morphological variables in men and in 8 variables in women. Among the variables of head-related phenotypic characteristics, heterogeneity was statistically significant in 7 morphological variables in men and 5 morphological variables in women. Furthermore, the results of discrimination analyses showed that heterogeneity exists in some investigated continuous characteristics among certain populations (Pribačic Ambrožic et al. 2009). The scope of this study is to explore the population structure, by anthropometric analysis of complex phenotypic traits, of a representative sample of the population in the northern Adriatic island of Rab, separately for men and women. The homogeneity or heterogeneity in morphologic characteristics, as well as the degree of differentiation among subpopulations of Rab, which correspond to the population of certain settlements (Banjol, Barbat, Lopar, Rab, and Supetar Draga) will be determined. For this purpose, the research objectives are defined. The first objective of the study is to identify the latent dimensions of the body and head for anthropometric variables, separately for men and separately for women. The second objective of the study is to determine the factors of differences among the populations on the island of Rab in the latent dimensions of the anthropometric variables and the body in the latent dimensions of the anthropometric variables of the head, separately for men and separately for women. Sample and methods The sample of participants included 601 persons, aged from 18 to 75 years (Državni zavod za statistiku 2003). All anthropometric complex characteristics were selected according to the guidelines of the International Biological Program (Weiner & Louri, 1969). All data collection was performed using the techniques and instruments from the Institute of Anthropology, according to the guidelines of the 'Practicum of Biological Anthropology' as follows: 'Anthropometry' (Buzina et al. 1975) and 'Morphological and functional anthropometry' (Buzina et al. 1984). The principle of the selection of the methods applied to participant were performed according to the protocol of the Institute of Anthropology in Zagreb and of the International Biological Program (IBP; 1982, according to Pribačic Ambrožic et al. 2009). The analysis included 30 anthropometric variables of the body, as well as 14 anthropometric variables of the head. Biostatistical analyses were carried out with methods of univariate and multi-variate statistics, separately for subsamples of men and women. The influence of the variable age on the variables of the body and head has been eliminated by linear regression. A large number of manifest variables of the body and the head is reduced by using the principal components analysis (PCA) with Varimax rotation (hereinafter, the expressions factor analysis and PCA, factors and principal components are used as the synonyms). Comparisons among factor structures for the variables of body and head in males and females were performed using the simple robust method for quasi-confirmatory factor analysis (QCFA) (Sindik 2013). Further analysis of the differentiation among subpopulations of Rab, which correspond to the populations of some villages, were carried out 43 Anthropological Notebooks, XXI/2, 2015 using appropriate regression factor scores for each latent anthropometric variable. Canonical discriminant analysis (DA) was used for determining the differences among the studied settlements, separately in the latent space of anthropometric variables of the body, and separately in the area of anthropometric variables of the head. ANOVA was used in testing the differences among the populations in the studied villages, for each latent an-thropometric variable of the body or head. Statistical analysis was performed using the software package IBM SPSS Statistics 19.0. Results Descriptive statistics was performed separately for the subsets of body and head anthropometric variables. In previous research (Pribacic Ambrozic et al. 2009), among the total of 36 body variables in male participants, the existence of statistically significant heterogeneity was found in 13 anthropometric characteristics. In female participants, among 36 body variables, the existence of statistically significant heterogeneities was found in 8 anthropometric characteristics. With anthropometric variables of the head, in male participants, the existence of statistically significant heterogeneities was found in the following 7 (out of 14) anthropometric head variables in total. Among female participants, of 14 anthropometric head variables, statistically significant heterogeneities were found in 5 characteristics (Pribacic Ambrozic et al. 2009). In this research, factor analysis is conducted separately on all samples of men and women, separate sets of variables for the body and head. In the analyses are used only simple anthropologic variables (excluding complex, or derived morphological variables, such as BMI), in other words a lower number of body variables (30, instead of 35) than in the previous study (Pribacic Ambrozic et al. 2009). All the variables used in this and previous study were standardised by age, which is important for considering real differences in morphological characteristics (Buffa et al. 2011). Table 1 shows the matrix of the factor loadings after Varimax rotation, for standardised variables of the body by age, separately for men and women subsamples. With both men and women, the application of PCA revealed the five principal components of the body morphological characteristics. Based on the variables that largely saturated each principal component, the factors in men are as follows: the transversal dimensions of the body, circumferences and body weight (factor 1), the longitudinal dimensions of the body and hip circumference (factor 2), the transversal dimensions of the extremities (factor 3), skin folds (factor 4), and the circumference of the knee (factor 5). Based on the variables that largely saturated each principal component, the factors in women are as follows: circumferences of the body and extremities, skin folds and body weight (factor 1), the longitudinal dimensions of the body and hip circumference (factor 2), the transversal dimensions of the extremities (factor 3), the circumference of the knee (factor 4), and the transversal dimensions of the body (factor 5). When applying the simple robust method for quasi-confirmatory factor analysis (QCFA), it was shown that no statistical significant differences are found among comparable factor structures at the morphological variables of the body, in males and females (when testing the differences among factor structures using the McNemar test). 44 Vanda Pribacic Ambrozic, Lajos Szirovicza, Josko Sindik, Pavao Rudan: Structures of morphological characteristics oof the population of the island of Rab Table 1: Matrix of the factors' loadings after Varimax rotation of the standardised anthropometric variables of the body by age Males (N=259) Females (N=342) Variable F 1 F 2 F 3 F 4 F 5 F 1 F 2 F 3 F 4 F 5 body height 0.894 0.889 sitting height 0.531 0.535 leg length 0.916 0.883 upper leg length 0.800 0.740 lower leg length 0.785 0.776 arm length 0.865 0.862 forearm length 0.803 0.743 upper arm length 0.724 0.672 biacromial length 0.521 0.406 0.395 0.425 0.379 0.568 thorax width 0.616 0.487 0.428 0.679 thorax depth 0.760 0.664 0.309 pelvis width 0.524 0.346 0.466 0.371 bicondiar with upper arm left 0.364 0.697 0.320 0.744 bicondiar with upper arm right 0.405 0.700 0.325 0.754 left knee width 0.815 0.794 0.303 right knee width 0.803 0.798 bicondilar width femur left 0.378 0.401 0.307 0.627 0.792 bicondilar width femur right 0.397 0.403 0.309 0.633 0.815 left ankle width 0.351 0.699 0.397 0.636 right ankle width 0.370 0.716 0.438 0.682 thorax circumferen. 0.885 0.867 abdominal circumferen. 0.829 0.857 ledge circumferen. 0.794 0.321 0.725 0.445 arm circumferen. 0.820 0.844 forearm circumfere. 0.769 0.468 0.738 0.310 upper thigh circumferen. 0.776 0.363 0.687 0.515 lower thigh circumferen. 0.656 0.331 0.303 0.568 0.511 biceps skinfold 0.417 0.741 0.748 triceps skinfold 0.354 0.803 0.670 0.414 suprailiac skinfold 0.816 0.455 0.663 body mass 0.848 0.312 0.854 0.318 Eigenvalue 7.697 6.367 4.594 2.643 1.885 7.465 5.939 4.234 3.177 1.439 Variance explained (%) 24.8 20.5 14.8 8.5 6.1 24.1 19.2 13.7 10.2 4.6 Cumulative % variance 24.8 45.4 60.2 68.7 74.8 24.1 43.2 56.9 67.1 71.8 Legend: Males: Factor 1- transversal dimensions of the body, circumferences and body weight; Factor 2-lon-gitudinal dimensions of the body and pelvis width; Factor 3-transversal dimensions of the extremities; Factor 4-skinfolds; Factor 5- knee width Females: Factor 1- body circumferences and the extremities, skinfolds and body weight; Factor 2-lon-gitudinal dimensions of the body and pelvis width; Factor 3-transversal dimensions of the extremities; Factor 4- knee width; Factor 5- transversal dimensions of the body 45 Anthropological Notebooks, XXI/2, 2015 However, all correlations among comparable factor structures are positive and significant: the contingency coefficient among first factors in males and females equals 0.595 (p<01), between second factors 0.707 (p<.01), among third factors 0.475 (p<01), among fourth factors 0.675 (p<.01) and among fifth factors 0.465 (p<.01). Table 2: Matrix of the factors' loadings after Varimax rotation of the standardised anthropometric variables of the head by age Males (N=259)_Females (N=342) Variable F 1 F 2 F 3 F 4 F 5 F 1 F 2 F 3 F 4 F 5 head length 0.301 0.424 0.517 0.315 0.907 head width 0.786 0.709 forehead width 0.839 0.849 face width 0.828 0.853 lower jaw width 0.610 0.594 morphological cheek height 0.799 0.822 nose height 0.811 0.753 nose width 0.641 0.457 0.786 mouth width 0.749 0.803 lip thickness 0.696 0.656 ear length 0.659 0.699 ear width 0.780 0.757 interorbital width 0.627 0.333 head circumference 0.630 0.393 0.463 0.739 Eigenvalue 2.996 1.763 1.600 1.424 1.275 2.737 1.834 1.680 1.398 1.339 Variance explained (%) 21.4 12.6 11.4 10.2 9.1 19.6 13.1 12.0 10.0 9.6 Cumulative % variance 21.4 34.0 45.4 55.6 64.7 19.6 32.6 44.6 54.6 64.2 Legend: Males: Factor 1- transversal dimensions of the head and head circumference; Factor 2-longitudinal dimensions of the face; Factor 3-transversal dimensions of the face and head length; Factor 4-length and width of the ear; Factor 5- lip thickness and interorbital width Females: Factor 1- transversal dimensions of the head; Factor 2-l longitudinal dimensions of the face and lip thickness; Factor 3-head length, interorbital width and head circumference; Factor 4- length and width of the ear; Factor 5- transversal dimensions of the face Table 2 shows the matrix of the factor loadings after Varimax rotation, for standardised variables of the head by age, separately for male and female subsamples. Both with men and women, the application of PCA revealed the five principal components of the head morphological characteristics. Based on the variables that largely saturated each principal component, the factors in men are named as follows: the transversal dimensions of the head and head circumference (factor 1), the longitudinal dimensions of the face (factor 2), the transversal dimensions of the face and head length (factor 3), length and width of the ear (factor 4), and thick lips and inter-orbital width (factor 5). Based on the variables that largely saturated each principal component, factors in women are as follows: the transversal dimensions of the head (factor 1), the longitudinal dimension of the face and thick lips (factor 2), head length, inter-orbital width and head circumference (factor 3), length and width 46 Vanda Pribacic Ambrozic, Lajos Szirovicza, Josko Sindik, Pavao Rudan: Structures of morphological characteristics oof the population of the island of Rab of the ear (factor 4), and the transversal dimensions of the face (factor 5). When QCFA was applied, it was shown that no statistically significant differences are found among comparable factor structures at the variables of the head, in males and females (when testing differences among factor structures using the McNemar test). However, all correlations among comparable factor structures are positive and significant: the contingency coefficient among first factors in males and females equals 0.652 (p<01), among second factors 0.499 (p<.05), among third factor in males and fifth factor in females 0.542 (p<.05), among fourth factors 0.616 (p<.01) and among fifth factor in males and third factor in females 0.499 (p<.05). Box's M tests for all sets of variables met the assumption of homogeneity of variance-covariance matrix, which allowed further step discriminant analysis. Table 3: Discrimination analysis in a space of latent anthropometric variables of the body in males among certain settlements on the island Rab Discrimination Eigen- Variance Canonical Test Wilks's function value explained correlation dis.fun. lambda ^-test df p 1 0.243 71.7 0.442 1 - 4 0.733 78.552 20 <0.001 2 0.067 19.8 0.250 2 - 4 0.911 23.612 12 <0.050 Latent Places on SE of F (df1/df2= Corr. Corr. variables Rab N Mean SD Mean 4/254) with 1st discr. funct. with 2nd discr. funct. transversal Banjol 33 0.229 1.017 0.177 dimensions Barbat 73 -0.051 0.941 0.110 of the body, Lopar 71 0.026 0.931 0.111 1.615 0.757 0.210 circumferences Rab 40 0.158 1.183 0.187 and body Sup.Draga 42 -0.286 0.982 0.152 weight longitudinal Banjol 33 0.076 0.915 0.159 dimensions Barbat 73 -0.146 0.911 0.107 of the body Lopar 71 0.136 1.022 0.121 0.814 -0.129 0.786 and pelvis Rab 40 0.028 1.007 0.159 width Sup.Draga 42 -0.062 1.165 0.180 transversal Banjol 33 0.042 0.838 0.146 dimensions Barbat 73 -0.076 1.075 0.126 of the Lopar 71 0.104 0.958 0.114 3.182* 0.522 -0.149 extremities Rab 40 -0.413 0.876 0.139 Sup.Draga 42 0.318 1.058 0.163 skinfolds Banjol 33 0.006 1.034 0.180 Barbat 73 0.551 1.154 0.135 Lopar 71 -0.298 0.775 0.092 9.364** -0.014 -0.532 Rab 40 -0.278 0.747 0.118 Sup.Draga 42 -0.195 0.869 0.134 knee width Banjol 33 0.266 0.837 0.146 Barbat 73 0.284 0.971 0.114 Lopar 71 -0.382 0.975 0.116 4.969** -0.195 -0.121 Rab 40 0.003 0.934 0.148 Sup.Draga 42 -0.061 1.088 0.168 Legend: ** F statistically significant at p<.01; *F statistically significant at p<.05; bold - highest statistically significant differences 47 Anthropological Notebooks, XXI/2, 2015 In Table 3, which refers to the differences in the latent anthropometric variables of the body in men, it is evident that the two discriminant functions statistically significantly differ among male residents of five different settlements on the island of Rab. The first discriminant function is the best described by latent variables of the transversal dimensions of the extremities and transversal dimensions of the body, circumferences and body weight. The second discriminant function is the best described by the latent variables of the longitudinal dimensions of the body and pelvis width and the negative projection of skinfolds. Based on the results of the ANOVA with respect to the discriminant function, it was showed that in a total of five factors of anthropometric variables of the body, three statistically significant indices of heterogeneity among male subpopulations in Rab are found in the transversal dimensions of the extremities; skinfolds; the width of the knee. Based on the discrimination functions, 40.5% of the participants are correctly classified. The most stable group is the subpopulation of males from Lopar with 67.6% of the participants correctly classified, followed by the population of the village Barbat with 67.1% of the participants correctly classified. Table 4: Discrimination analysis in a space of latent anthropometric variables of the body in females among certain settlements on the island Rab Discrimination Eigen- Variance Canonical Test Wilks's function value explained correlation dis.fun. lambda ^?-test df p 1 0.332 SS.0 0.499 1 - 4 0.71S 111.30 20 <0.001 Latent Places on SE of F (df1/df2= Corr. with 1st variables Rab N Mean SD Mean 4/254) discr. funct. circumferences Banjol 63 0.045 1.023 0.129 of the body Barbat S4 -0.104 1.017 0.111 and extremities, Lopar 6S 0.306 0.956 0.116 2.425* 0.682 skinfolds and Rab 69 -0.06S 1.041 0.125 body weight Sup.Draga 5S -0.176 0.S97 0.11S longitudinal Banjol 63 0.030 0.947 0.119 dimensions Barbat S4 -0.144 0.973 0.106 of the body Lopar 6S 0.204 0.99S 0.121 1.209 0.533 and pelvis Rab 69 -0.053 0.950 0.114 width Sup.Draga 5S 0.001 1.139 0.150 transversal Banjol 63 -0.150 0.979 0.123 dimensions Barbat S4 0.140 0.94S 0.104 of the Lopar 6S 0.214 1.097 0.133 2.263 -0.091 extremities Rab 69 -0.191 0.937 0.113 Sup.Draga 5S -0.063 1.006 0.132 knee width Banjol 63 0.2S4 0.962 0.121 Barbat S4 0.382 0.S9S 0.09S Lopar 6S -0.636 0.S44 0.102 13.158** -0.219 Rab 69 0.025 0.9S7 0.119 Sup.Draga 5S -0.145 0.997 0.131 transversal Banjol 63 0.13S 0.935 0.11S dimensions Barbat S4 0.409 1.023 0.112 of the body Lopar 6S -0.457 0.9S2 0.119 8.326** -0.190 Rab 69 -0.065 0.S45 0.102 Sup.Draga 5S -0.12S 0.993 0.130 Legend: ** F statistically significant at p<.01; *F statistically significant at p<.05; bold - highest statistically significant differences 48 Vanda Pribacic Ambrozic, Lajos Szirovicza, Josko Sindik, Pavao Rudan: Structures of morphological characteristics oof the population of the island of Rab In Table 4, which refers to the differences in anthropometric variables latent body of women, it is evident that one discriminant function was significantly different among the residents of five different settlements on the island of Rab. The first discriminant function is the best described by latent variables: circumferences of the body and extremities, skinfolds and body weight and longitudinal dimensions of the body and pelvis width. Based on the results of the ANOVA with respect to the discriminant function, it was shown that in five factors of anthropometric variables of the body, statistically significant indices of heterogeneity among female subpopulations in Rab are found: circumferences of the body and extremities, skin-folds and body weight; knee width; transversal dimensions of the body. Based on the discriminant function, 37.4% of the participants are correctly classified. The most stable group is the subpopulation of females from Lopar, with 67.6% of the participants are correctly classified. Table 5: Discrimination analysis in a space of latent anthropometric variables of the head in males among certain settlements on the island Rab Discrimination Eigen- Variance Canonical Test Wilks's function value explained correlation dis.fun. lambda ^-test df p 1 0.318 76.4 0.491 1 - 4 0.690 94.00 20 2 0.072 17.2 0.258 2 - 4 0.909 24.19 12 Latent Places on SE of F (df1/df2= Corr. Corr. variables Rab N Mean SD Mean 4/254) with 1st with 2nd discr. discr. funct. funct. transversal Banjol 33 -0.066 1.067 dimensions Barbat 73 0.088 1.006 of the head Lopar 71 0.263 0.869 3.125* 0.918 0.343 and head Rab 40 -0.244 1.039 circumference Sup.Draga 42 -0.313 1.008 longitudinal Banjol 33 0.167 0.974 dimensions Barbat 73 -0.060 1.163 of the face Lopar 71 0.002 0.990 1.095 -0.194 0.671 Rab 40 0.186 0.715 Sup.Draga 42 -0.208 0.953 transversal Banjol 33 -0.008 1.137 dimensions Barbat 73 -0.088 0.967 of the face Lopar 71 0.172 1.015 0.763 0.016 0.151 and head Rab 40 -0.080 1.070 length Sup.Draga 42 -0.055 0.849 length Banjol 33 0.427 0.943 and width Barbat 73 0.497 0.909 of the ear Lopar 71 -0.660 0.853 17.530** -0.182 0.633 Rab 40 0.052 0.895 Sup.Draga 42 -0.134 0.884 lip thickness Banjol 33 -0.226 1.029 2.794* -0.177 0.082 and interorbital Barbat 73 0.070 0.973 width Lopar 71 0.214 1.053 Rab 40 0.056 0.888 Sup.Draga 42 -0.360 0.946 Legend: ** F statistically significant at p<.01; *F statistically significant at p<.05; bold - highest statistically significant differences 49 Anthropological Notebooks, XXI/2, 2015 In Table 5, which refers to the differences in the latent anthropometric variables of the head in men, it is evident that the two discriminant functions significantly differ among the residents of five different settlements on the island of Rab. The first discriminant function is the best described by the latent variable transversal dimensions of the head and head circumference. The second discriminant function is the best described by the latent variables: longitudinal dimensions of the face; length and width of the ear. Based on the results of the ANOVA, with respect to the discriminant functions, it is showed that in a total of five factors of the anthropometric variables of the head, three statistically significant indices of heterogeneity among male subpopulations in Rab are found: in the transversal dimensions of the head and head circumference; length and width of the ear; lip thickness and interoibital width. Based on the discrimination functions, 42.1% of participants are correctly classified. The most stable group is the subpopulation of males from Lopar, with 70.4% of participants correctly classified. Table 6: Discrimination analysis in a space of latent anthropometric variables of the head in females among certain settlements on the island Rab Discrimination Eigen- Variance Canonical Test Wilks's function value explained correlation dis.fun. lambda ^?-test df p 1 0.141 65.1 0.352 1 - 4 0.813 69.409 20 <0.001 2 0.052 23.7 0.221 2 - 4 0.928 24.962 12 0.015 Latent Places on SE of F (df1/df2= Corr. Corr. variables Rab N Mean SD Mean 4/254) with 1st with 2nd discr. discr. funct. funct. transversal Banjol 63 -0.031 1.168 0.147 dimensions Barbat 84 0.090 1.062 0.116 of the head Lopar 68 0.122 0.878 0.107 1.030 0.828 0.213 Rab 69 -0.184 0.803 0.097 Sup.Draga 58 -0.020 1.053 0.138 longitudinal Banjol 63 0.272 1.057 0.133 dimensions Barbat 84 0.063 0.961 0.105 of the face Lopar 68 -0.156 1.006 0.122 2.077 -0.128 0.858 and lip Rab 69 -0.037 1.011 0.122 thickness Sup.Draga 58 -0.159 0.933 0.123 transversal Banjol 63 -0.049 0.983 0.124 dimensions Barbat 84 0.102 1.019 0.111 of the face Lopar 68 0.313 1.041 0.126 3.523** 0.228 -0.025 Rab 69 -0.159 0.954 0.115 Sup.Draga 58 -0.271 0.906 0.119 length Banjol 63 -0.049 0.992 0.125 and width Barbat 84 0.482 0.972 0.106 of the ear Lopar 68 -0.395 0.929 0.113 8.537** 0.003 0.437 Rab 69 -0.120 0.973 0.117 Sup.Draga 58 -0.039 0.917 0.120 head length, Banjol 63 -0.049 1.070 0.135 interorbital Barbat 84 -0.207 1.010 0.110 width and Lopar 68 0.287 1.024 0.124 2.469* -0.440 0.119 head Rab 69 -0.043 0.900 0.108 circumference Sup.Draga 58 0.067 0.939 0.123 Legend: ** F statistically significant at p<.01; *F statistically significant at p<.05; bold - highest statistically significant differences 50 Vanda Pribacic Ambrozic, Lajos Szirovicza, Josko Sindik, Pavao Rudan: Structures of morphological characteristics oof the population of the island of Rab In Table 6, which refers to the differences in the latent anthropometric variables of the head in women, it is evident that the two discriminant functions significantly differ among residents of five different settlements on the island of Rab. The first discriminant function is the best described by the latent variables: transversal dimensions of the head; negative projection on head length, interorbital width and head circumference. The second discriminant function is the best described by the latent variables: longitudinal dimensions of the face and lip thickness; length and width of the ear. Based on the results of the ANOVA with respect to the discriminant functions, it showed that in a total of five factors of anthropometric variables of the head, three statistically significant indices of heterogeneity among female subpopulations in Rab are found in: transversal dimensions of the face; length and width of the ear; head length, interorbital width and head circumference. On the basis of the discrimination functions, 31.3% of the participants are correctly classified. The most stable group is the subpopulation of females from BaTbat (58.3% correctly classified participants), followed by Lopar with 45.6% correctly classified participants. Discussion and conclusions The factors revealed in this research followed the logic of morphological factors (cumulative environmental-genetic factor, longitudinal factor and transversal factors) obtained in the research about morphometric variables, measured on six metacarpal bones (Simic et al. 1992). These factors can be discussed within the context of their biological meaning affecting the phenotypic formation of the body and head (as well as of the metacarpal skeleton) in a given population (Smolej et al. 1987; Simic et al. 1992; Smolej-Narancic et al. 1994). Based on the analysis of complex phenotypic traits in subpopulations of Rab, it is possible to present the following general findings. First, by using the regression procedures, when we analysed the impact of age on anthropometric characteristics, it is found that age has a significant proportion of variability in these variables and in both sexes. Specifically, with aging, there is an increase in transversal variables and in circumferences, with a reduction of longitudinal variables. Among the variables of the head, there is an increase in the size of the face and ear. The impact of the factors of age, height and weight, on the investigated properties (variables) is eliminated by their standardisation, enabling an analysis of phenotypic variations in the subpopulations with a large age range. The estimation of the heterogeneity among the subpopulations of the island, using latent variables, showed that in both sexes, population settlements differ in three of the five factors, among anthropometric variables of the body. Heterogeneity among subpopulations exist for eco-labile properties (circumferences of the body and extremities, diameters of the extremities and skinfolds). This fact suggests that the observed phenotypic variations of the body are primarily the results of the interaction among phenotypic plasticity and adaptation. Results of the assessment of the heterogeneity among the subpopulations of Rab using the latent dimensions of the head in both men and women showed significant differences in three factors. Heterogeneity in eco-stabile properties of the head indicates that due to differing population settlement in different periods and due to reproductive isolation, the population from the examined settlements retain their 51 Anthropological Notebooks, XXI/2, 2015 original properties. Discriminant analysis, conducted in the space of anthropometric variables of the body, separate men from Lopar and Barbat according to their skinfolds. Men from Rab and Supetar Draga can be discriminated by their transverse dimensions of the extremities. Women from Lopar and Barbat can be differed by the transverse dimensions of the body and the width of the knee. Discriminant analyses, conducted in the space of the variables of the head, showed that in both sexes, people from Lopar have smaller ears and a larger head circumference. Therefore, the investigations in the space of morphological dimensions performed on the populations of inhabitants of the Island of Rab provided data useful for explaining micro-evolutional processes that shaped the present population of the island (Pribacic Ambrozic et al. 2009). Analysing the latent morphological variables enabled clearer insight that the examined population is heterogeneous in the space of the investigated body and head variables for both sexes. The heterogeneity of body variables in men from five locations on the Island of Rab is more emphasised in the torso and extremities circumferences, as well as in the diameters of extremities (eco-labile characteristics). In other words, their phenotype is more prone to the impact of external factors during the periods of growth and development (Rudan et al. 1994). The heterogeneity in women is more emphasised in eco-labile, but also in eco-stable morphological characteristics, which can be explained in terms of the female phenotype, in contrast to the male, less susceptible to the influence of ecological factors, as well as in terms of univariate analyses (Pribacic Ambrozic et al. 2009). The performed discrimination analyses showed the existence of differences in latent morphological variables of the body and head in both men and women. These differences are more clearly seen in more eco-labile variables; the greatest differences can be seen among the settlements of BaTbat and Lopar, both in head and body latent variables. In terms of historical course of settling on the Island of Rab, it has to be mentioned that the emergence of population groups on the island was followed by long periods of complete biological and socio-cultural isolation. This partial isolation, together with the interactive evolution processes, can have a result in phenotypic and biological (morphological) differentiation of a group (Rudan et al. 1988; Rudan et al. 1999). These results clearly reflect the grouping of current population communities on the Island of Rab, in terms of their morphological characteristics, which can be correlated with the historical facts on the settling of the island, together with the dynamics of the migrations of the inhabitants on the island. In the future studies, whether this biological variation in morphology variables is related to geographic, linguistic and migration differences could be examined. Acknowledgements This study has been carried out as part of the scientific project The Population Structure of Croatia - Anthropogenetic Approach, Ministry of Science, Education and Sports, Republic of Croatia (Project number: 196-1962766-2751, Project coordinator: Academician Pavao Rudan). We would also like to thank Professor Nina Smolej-Narancic for her valuable help and suggestions. 52 Vanda Pribacic Ambrozic, Lajos Szirovicza, Josko Sindik, Pavao Rudan: Structures of morphological characteristics oof the population of the island of Rab References Buffa Roberto, Giovanni U. Floris, Paolo F. Putzu & Elisabetta Marini. 2011. Body composition variations in ageing. Collegium Antropologicum 35(1): 259-65. Buzina, Ratko, Z. Grgič, M. Kovačevič, Hubert Maver, Konstantin Momirovič, Pavao Rudan, Ljerka Schmutzer & Bosiljka Štampar-Plasaj. 1975. Praktikum biološke antropologije - Antropometrija. Zagreb: ZLH, RSIZZ. Buzina, Ratko, Z. Grgič, M. Kovačevič, Hubert Maver, Konstantin Momirovič, Pavao Rudan, Ljerka Schmutzer & Bosiljka Štampar-Plasaj. 1984. Praktikum biološke antropologije - Morfološka i funkcionalna antropometrija. Zagreb: RSIZZ, ZLH. Državni zavod za statistiku Republike Hrvatske. 2003. Popis stanovništva 2001. http://www.dzs.hr/. Accessed on 2 June 2015. Pavičič Dino. 2004. Populacijska struktura antropogenetičkog izolata (otok Vis) procijenjena analizom kompleksnih svojstava. Unpublished PhD Thesis. Zagreb: Sveučilište u Zagrebu. Pribačič Ambrožič, Vanda, Milko Padovan, Lajos Szirovicza & Božo Vojnikovič. 2009. Anthropometrical Characteristics of the Population of the Island of Rab. Collegium antropologicum 33(4): 1329-34. Rudan, Igor, Pavao Rudan, Andre Chaventre, Branka Janičijevič, Jasna Miličič, Nina Smolej-Narančič & Anita Sujoldžič. 1998. Model-Bound and Model-Free Approach in Holistic Analysis of Population Structure: Example from the Island of Pag, Croatia. Homo — Journal of Comparative Human Biology 49: 201-24. Rudan, Igor, Pavao Rudan, Branka Janičijevič, Jasna Miličič, Nina Smolej Narančič, Andre Chaventre & Anita Sujoldžič. 1999. Sets of complentary phenotypic relationships across five isolate populations: Traces of expression of different gene complexes?. International Journal of Antropology 14: 227-42. Rudan, Pavao, Anita Sujoldžič, Diana Šimič, Linda A. Bennett & Derek F. Roberts. 1992. Population structure in the eastern Adriatic: the influence of historical processes, migration patterns, isolation and ecological pressures, and their interaction. In: Derek F. Roberts, Norio Fujiki & Kanji Torizuka (eds.), Isolation, Migration and Health. Cambridge Univ. Press, pp. 204-18. Rudan, Pavao, Božidar Finka, Branka Janičijevič, Veljko Jovanovič, Vesna Kušec, Jasna Miličič, Marjeta Mišigoj-Durakovič, Derek F. Roberts, Ljerka Schmutzer, Nina Smolej-Narančič, Anita Sujoldžič, Lajos Szirovitza, Diana Šimič, Petar Šimunovič & Sanja Marina Špoljar-Vržina. 1990a. Antropološka istraživanja istočnog Jadrana, Knjiga druga: Biološka i kulturna mikrodiferencijacija seoskih populacija otoka Hvara. Zagreb: Hrvatsko antropološko društvo. Rudan, Pavao, Branka Janičijevič, Diana Šimič & Linda A. Bennett. 1994. Migrations in the Eastern Adriatic and their consequences on population structure - Analyses of census data. In: Alexandar Lopašič (ed.), Mediterranean Societes: Transition and Change. Zagreb: HAD, pp. 113-22. Rudan, Pavao, Derek F. Roberts, Anita Sujoldžič, Bert Macarol, Eugenija Žuškin & Andrija Kaštelan. 1982. Strategy of anthropological research on the island of Hvar. Collegium Antropologicum 6: 39-46. Rudan, Pavao, Derek F. Roberts, Branka Janičijevič, Nina Smolej, Lajos Szirovicza & Andrija Kaštelan. 1986. Anthropometry and the biological structure of the Hvar population. American Journal of Physical Anthropology 70: 231-40. Rudan, Pavao, Diana Šimič & Linda A. Bennett. 1988. Isolation by distance on the island of Korčula: Correlation Analysis of Distance Measures. American Journal of Physical Anthropology 77: 97-103. Rudan, Pavao, Dijana Šimič, Nina Smolej-Narančič, Linda A. Bennett, Branka Janičijevič, Veljko Jovanovič, Margaret F. Lethbridge, Jasna Miličič, Derek F. Roberts, Anita Sujoldžič & Lajos Szirovicza. 1987a. Isolation by distance in middle Dalmatia - Yugoslavia. American Journal of Physical Anthropology 74(3): 417-26. Rudan, Pavao, J. Lawrence Angel, Linda A. Bennett, B. Finka, Branka Janičijevič, Veljko Jovanovič, Margaret F. Lethbridge, Jasna Miličič, Marjeta Mišigoj, Nina Smolej-Narančič, Anita Sujoldžič, Lajos Szirovicza, Diana Šimič & Petar Šimunovič. 1987c. Antropološka istraživanja istočnog Jadrana, Knjiga prva: Biološka i kulturna mikrodiferencijacija seoskih populacija Korčule i Pelješca. Zagreb: Hrvatsko antropološko društvo. Rudan, Pavao, J. Lawrence Angel, Linda A. Bennett, Branka Janičijevič, Margaret F. Lethbridge, Jasna Miličič, Nina Smolej-Narančič, Anita Sujoldžič & Diana Šimič. 1987b. Acta morphologica Neerlando-Scandinavi-ca 25: 69-82. Rudan, Pavao, Linda A. Bennett, B. Finka, Branka Janičijevič, Veljko Jovanovič, Vesna Kušec, Margaret F. 53 Anthropological Notebooks, XXI/2, 2015 Lethbridge-Čeiku, Jasna Miličič, Ljerka Schmutzer, Nina Smolej-Narančič, Anita Sujoldžič, Diana Šimič, Petar Šimunovič & Sanja Marina Špoljar-Vržina. 1990b Antropološka istraživanja istočnog Jadrana, Knjiga treca: Biološka i kulturna mikrodiferencijacija seoskih populacija otoka Brača. Zagreb: Hrvatsko antropološko društvo. Rudan, Pavao, Milica Gomzi, Eugenija Žuškin & Marko Šarič. 1979. Anthropometric and physiological properties in the determination of biological distances. Human Heredity 29: 77-81. Rudan, Pavao. 1978. Biological structure of the Istrian population - Study of "distances" by anthropometric traits, dermatoglyphic properties and blood group gene frequencies. Collegium Antropologicum 2: 53-8. Rudan, Pavao. 1982. Dermatoglifi u proučavanju bioloških udaljenosti - Prilog antropologijskim istraživanjima populacije otoka Hvara. Rad JAZU 402: 167-206. Šimič, Diana, Andre Chaventre, Chris C. Plato, Jordan D. Tobin & Pavao Rudan. 1992. Factor structure of mor- phometric variables measured on six metacarpal bones. Annals of Physiological Anthropology 11: 3-12. Sindik, Joško. Simple robust method for quasi-confirmatory factor analysis (Three examples). Collegium antropologicum 37(4): 1071-80. Smolej, Nina, J. Lawrence Angel, Linda A. Bennett, Derek F. Roberts & Pavao Rudan. 1987. Physiological variation and population structure of the island of Korčula, Yugoslavia. Human Biology 59: 667-85. Smolej-Narančič, Nina, Andre Chaventre, Pavao Rudan. 1994. Anthropometric variation and population structure of the island of Pag, Croatia. Human Biology 66: 275-98. Wright, Sewall 1931. Evolution in Mendelian populations. Genetics 16: 97-159. Wright, Sewall. 1943. Isolation by distance. Genetics 28: 114-56. Wright, Sewall. 1951. The genetical structure of the populations. Annals of Eugenics 15: 323-354. Povzetek Glavni predmet pričujočega članka je del holistične antropološke interdisciplinarne študije populacijske strukture prebivalstva otoka Raba, ki temelji na multivariatnih analizah la-tentnih morfoloških spremenljivk, z namenom obrazložitve mikroevolucije prebivalstva. Cilji raziskovanja so usmjerjeni v proučevanje razlik med prebivalstvom v geografsko zamejenem področju Raba na osnovi latentnih morfoloških spremenljivk glave in telesa. Predstavljena je analiza reprezentativnega vzorca 601 odraslih oseb (v starosti od 18 do 75 let) iz naselij Banjol, Barbat, Lopar, Rab in Supetarska Draga. V analizo je vključenih 30 antropometrijskih spremenljivk telesa ter 14 antropometrijskih spremenljivk glave. Analiza glavnih komponent z varimax rotacijo je izolirala pet latentnih spremenljivk telesa in pet latentnih spremenljivk glave (pri obeh spolih), medtem ko je kvazi-konfirma-torna faktorska analiza pokazala, da ni statistično značilnih spolnih razlik v latentni strukturi morfoloških spremenljivk. Analizira razlik v latentnih morfoloških spremenljivkah je omogočila jasen vpogled v heterogenost proučevane populacije v prostoru proučevanih spremenljivk telesa in glave pri obeh spolih. KLJUČNE BESEDE: antropometrija, struktura prebivalstva, Rab CORRESPONDENCE: JOSKO SINDIK, Institute for Anthropological Research, Ljude-vita Gaja 32, Zagreb, Croatia. E-mail: jsindik@inantro.hr. 54