Craniometrical sex determination of wild cat Felis silvestris in Bulgaria

Transcription

1 Acta Theriologica 37 (4): , PL ISSN Craniometrical sex determination of wild cat Felis silvestris in Blgaria Ivan PETROV, Hristo NIKOLOV and Svetoslav GERASIMOV Petrov I., Nikolov H. and Gerasimov S Craniometrical sex determination of wild cat Felis silvestris in Blgaria. Acta theriol. 37: Analyses were made of 118 skll measrements of adlt wild cats Felis silvestris Schreber, 1777 from Blgaria, taken on 24 males, 20 females, and 10 animals whose sex was not known in advance. Grop (clster) analysis of cases, factor analysis, and stepvise discriminant analysis were adapted. The clster analysis of cases indicated a high level of sex mixtre (p to 40%), which sggested the importance of the otliers in the data. Six keys to sexal dimorphism, of very high statistical significance were prodced, throgh the stepvise discriminant analysis. They inclded from 9 down to 1 variables each, which provided from 100% down to 93.8% of correct sex classification of wild cat sklls. Forest Research Institte, 132 St. Kl. Ohridski str., 1756 Sofia, Blgaria (IP); Institte of Zoology, 1 Tsar Osvoboditel bol., 1000 Sofia, Blgaria (HN, SG) Key words: Felis silvestris, sexal dimorphism, discriminant keys, Blgaria Introdction Sexal dimorphism of the cranial featres of wild cat Felis silvestris Schreber, 1777 was noted by French et al. (1988). Arrighi and Salotti (1988) fond for wild cats from Corse a male bigger than a female. Petrov et al. (1990) noted significant differences between males and females for a poplation of wild cats from Blgaria for most of 121 skll measrements. This forms a basis to apply discriminant analysis for obtaining keys to the sexal dimorphism in wild cat. Similar approach was sed for Martes foina and M. martes by Gerasimov (1985). Principal component analysis (PCA) cold be applied in order to determine the main sorces of the morphometric differences between sexes. Both discriminant analysis (althogh withot providing keys to classify new individals) and PCA were applied to stdy sexal dimorphism in two poplations of domestic cats (Wetzel et al. 1982, Petrov et al. 1992). The aim of this stdy was to analyse the sexal dimorphism in sklls of wild cats in Blgaria, sing the above mentioned methods. [381]

2 382 I. Petrov et al. Material and methods For the stdy was sed 24 males and 20 females of adlt wild cats, Felis silvestris Schreber, 1777, from Blgaria. Additional 10 individals of nknown sex were sed for some verifications. The material examined is kept at the National Msem of Natral History, Sofia, at the Forest Research Institte, and in some private collections. All the material sed was collected as "wild cat", and additionally, was verified by means of the Schaenberg's index (Schaenberg 1969). The vales of the index were 2.46, SD = for males and 2.45, SD = for females. The age of wild cats was determined by analysing annal growth layers in dentine of lower canines. The teeth were decalcified in 6.5% HNO3 soltion, cross-sectioned to 10 fjm. For slides were obtained for each specimen: one of them was kept nstained while each one of the other three were then stained with either of Ehrlich's, or Mayer's, or Delafield's hematoxylin. Annal growth layers were observed and conted nder a microscope. The wild cats sed were from 1+ to 7 years old. Reslts from PCA by French et al. (1988) showed that increasing the nmber of variables sed increased to a greater extent the distance between sexes for old wild cats (from 3.2 to 7.1) than for recent and modern ones (from 2.6 to 5.7). Ths, a wide set of measres to be applied for stdies of sexal dimorphism in wild cat cold be recommended. The list of all 118 skll measrements and their determinations sed are given in the Appendix 1. Variables were denoted frther in the text as VI, V2, etc. The scheme of measrements was the same as in Petrov et al. (1990), except V10, V53, and V81. All measrements, except the nerocranim capacity (V56), were taken by a vernier caliper, with an approximation of 0.1 mm. The brain volme (V56) was measred sing lead balls of 1.5 mm diameter. All measrements were taken by the same person. Locations of measrement points on wild cat skll are shown in Appendix 1. The mltivariate methods applied cold be specified as follows: clster analysis of cases based on Eclidean distances (withot standardization of data prior to the analysis), principal component analysis (PCA) based on correlation matrices (with standardization of the variables sed), and stepwise discriminant analysis with a choice of F-to-enter vale of 4.0 (see Forsythe's remark in Dixon 1981, pp ). The statistical analysis was performed sing BMDP software (Dixon 1981): BMDPKM for clster analysis of cases, BMDP4M for PCA analysis, BMDP7M for stepwise discriminant analysis, and BMDPAM was applied for a preliminary testing of the assmptions of the above mentioned mltivariate methods. Reslts The arithmetical means, and standard deviations of 118 skll measrements of wild cat from Blgaria are listed in an Appendix 2. Mltivariate analyses were tilized in order to distingish between sexes on the basis of the whole complex of variables, and not on single variables, as it was in all nivariate analyses (Petrov et al. 1990). Clster analysis of cases was applied on the whole data matrix not giving a priori any information abot sex. Althogh the two grops (clsters) were formed predominantly by individals of the same sex (Table 1), it shold be noted that abot 40% of the males belonged to the "female" clster. Ths, despite of the large nmber of highly significant differences that were previosly fond between the means of 121 dimensions for the two sexes (Petrov et al. 1990), otliers in the data seem to be important in the morphometrical characteristics of the wild cat

3 Sexal dimorphism in skll of wild cat 383 Table 1. Percentages of individals of Felis silvestris forming two clsters based on cranial or mandiblar measrements. Sex Cranial measrements Mandiblar measrements Clster I Clsterll Clster I Clster II Males 62.5% 37.5% 59.0% 41.0% Females 5.0% 95.0% 19.0% 81.0% in Blgaria. Therefore, finding working discriminant keys to the sexal dimorphism in this species is a priority task. Principal component analysis with standardization and varimax rotations, based on correlation matrices was applied. The reslts for the first two principal components (called ftrher PCI and PC2) based on both absolte and relative cranial measres showed that PCI explains 42.7% of the total variance in the data space, while PC2 - only 9.9%. Both principal components cold strongly be related to sex (see Fig. la). The first PC was related mainly to 43 absolte and 13 relative variables, predominantly lengths, while PC2 was explained by the brain volme (V56) and it's indices, the sizes of P 4, and three cranial widths. It is worth mentioning that specimens deviating from their grops on Fig. la are practically ff f f f f f l I Fig. 1. Individal vales of specimens of Felis silvestris from Blgaria for the PCI and PC2, based both on absolte and relative cranial measrements (a), and mandiblar measrements (b). Factor scores is along the abscissa (factor 1) and the ordinate (factor 2). Males are denoted by'm', females by T, and specimens of nknown sexes by ''. Nmbers of highly deviating individals are also given.

4 384 I. Petrov et al. f f f fff ff f ffff ffff ff f + F H m m mm m m mm m m mmmmmmmmm m mm m m f f f f f mmm f fff ff f f ff ffm mm mm m mmmmmmmm m F f m f m m m ffff f ff f ffffff f m f mmmm m m mm mmmmm mmm F M d m f f f m m mmm m f ff ff f fffff fff ff f m m mmmm m mm mmmmm m mm + + F ^ fff ff <i f f ff f ff f ffff fff F m mf m ii m fjii mm m I. II,I..mmmm mmm mmm -+ M f m mf f mm f mm f f mm f f mm ffff ff mmmmmmm ffffffmf mmmmmmmmm -+--+F M Fig. 2. Histograms of the canonical variable X corresponding to the discriminant keys fond for sex determination of the sklls of wild cat Felis silvestris from Blgaria. X is along the abscissa. *M' and 'F' stand for the arithmetical means, while 'm', T and '' denote the vales for male, female or ndetermined individals, a - first key, based on 9 dimensions of the cranim, b - second key, based on 6 dimensions of the cranim, c - third key, based on 4 dimensions of the cranim, d - forth key, based on 3 absolte and 2 relative dimensions of the cranim, e - fifth key, based on 4 absolte and 2 relative dimensions of the mandible, f- sixth key, based on the length of Ci (Vlll).

5 Sexal dimorphism in skll of wild cat 385 the same specimens which were wrongly classified throgh some discriminant keys, based on the crania (nmbers 21, 25, and 41), which confirms the explanation of both components by sex. The PCI obtained on the basis of both absolte and relative sizes of mandibles explains 49.3%, and the PC2-18.7% of the total variance in the data space. Both principal components cold too mch be related to sex (see Fig. lb). PCI was more related to common lengths and then to widths, heights, or their indices, while the PC2 - only to teeth sizes and their indices. Once again the same specimens which were wrongly classified throgh the discriminant keys based on mandible measres (nmbers 21, 41, and 112; Fig. lb), deviated from their grops, which confirms the explanation of both PC's by sex. The reslts of the PCA demonstrated the sexal dimorphism being the main sorce of variability in the craniometrical data for the wild cat from Blgaria. By means of stepwise discriminant analysis, several classification fnctions were fond. The first key provided a 100% correct classification and was based on 9 dimensions of each skll: the greatest diameter of the orbit (V14), the distance between processs zygomatics and crista sagittalis externa posterior (V15), the greatest mastoid breadth (V35), the frontal breadth (V41), the least distance between the meata acstica externa (V42), the least distance between the aditory blla (V43), height of the foramen magnm (V57), the distance between foramina incisiva (V61) and the least diameter of C 1 (V79). In terms of a standardized (by the pooled within-grop variances) canonical variable x, the key was as follows: * = x V x V x V x V x V x V x V x V x V with the means of and for males and females, respectively, (Fig. 2a). The second discriminant key presented, provided a 97.7% correct classification (only the male No 25. was determined wrongly as a female) and it was based on 6 dimensions of each skll: the cranial breadth behind cheek-bone arcks (V33), the distance between proc. zygomatics and os zygomaticm (V40), the prot. occipitalis height (V52), the distance between os incisim anterior and os frontale posterior (V66), the height of nasal opening (V69), and the greatest diameter of C 1 (V78). In terms of a standardized canonical variable x, the key was as follows: ^ = x V x V x V x V x V x V with means of for males and for females (Fig. 2b). A third discriminant key was prodced, basing only on 4 variables. It provided a 95.5% correct classification, since a male, No. 41 and a female, No. 21 were in the intersection zone of the two "normal" crves (Fig. 2c), being somewhat closer to the center of the opposite grop. Nevertheless, the key is sefl for its simplicity, since it involves only the distance between the hind edges of osa palatina (V32), the length of the aditory blla (V45), the height of C 1 (V80), and the breadth of P 4 (V88). The third key was as follows:

6 386 I. Petrov et al. x = x V x V x V x V with means of for males and for females (Fig. 2c). In attempt to improve the discrimination between sexes, 32 relative indices were constrcted by dividing some variables describing widths, heights or teeth sizes by other ones, mainly skll lengths. On the basis of both absolte and relative sizes a forth key was obtained, providing a 100% correct classification. It involves the condylobasal length (V2), the distance between processs zygomatics and crista sagittalis externa posterior (V15), the least breadth of the spraorbital processes (V31), the cranial breadth behind cheek-bone arcks (V33), the palatal breadth measred between alveoli C 1 (V38), the distance between foramina incisia (V61), and the least diameter of C 1 (V79). The forth discriminant key was as follows: x = x V x V x V x V61/V x V79/V with means of for males and for females (Fig. 2d). As sometimes in practice only mandibles and not entire sklls are available, discriminant keys based on 32 mandible measrements, and on 17 indices formed ot of them, were also prodced. Ths, the fifth key provided a 95.8% correct classification (the male No. 35 and the female No. 112 were closer to the center of the opposite grop). It is based on 8 mandible sizes: the mandiblar length (V90), the mandiblar length (V92), the total length of the mandiblar teeth row (V96), the height of corps mandiblae (V102), the mandiblar breadth after Ci (V105), the distance between processs coronoidei (V107), the greatest gage of corps mandiblae (V120), and the total length of Ci (V121). The fifth (mandiblar) key was as follows: x = x V x V x V x V x V90/V x V96/V with means of for males and for females (Fig. 2e). A very simple sixth key based only on one measrement, namely the greatest diameter of Ci (Vlll), was fond. It provided a 93.8% correct classification (a male, No. 41 and two females: No. 21, and No. 112 were attached to the opposite sex, Fig. 2f). The key is as follows: x = x Vlll with means of for males and for females (Fig. 20. The six keys above show a very high statistical significance (p < ). Discssion Univariate statistics for all the variables sed here were presented, together with the reslts of a -test between the corresponding vales for male and female wild cats (Petrov et al. 1990). Univariate analysis showed the means for males being significantly higher than the corresponding vales for females in most of

7 Sexal dimorphism in skll of wild cat 387 the measrements taken. However, the reslts from clster analysis of cases in this stdy showed the importance of the otliers in the data, also demonstrated for some other Carnivora species (Bchalczyk and Rprecht 1977, Reig 1989). Both clster analysis and principal component analysis showed sexal differences appearing as the main sorces of variability. Sexal dimorphism was related both with dimensions (i.e. absolte sizes) and shapes (relative sizes) of sklls. Among the first ones, more important were lengths, followed by widths, heights, and teeth sizes while among the latter ones - the indices formed by dividing widths to lengths, or brain volmes to lengths differentiate sexes better than those that heights participate in. Similar reslts with respect of the importance of lengths, widths, heights, as well as their indices, were also obtained for the mandibles. Stdies of other species of Carnivora have shown size to be more important in terms of sexal dimorphism for stone marten Martes foina (De Marinis et al. 1990), and shape - for otter Ltra ltra and Eropean badger Meles meles (Wiig 1986). An important practical advantage of the stepwise discriminant analysis is that it provides keys involving only few dimensions (from 1 to 9 in or case). However, neither the factors responsible for the variability in the data-set cold be determined, nor the significance for the sexal dimorphism of all the dimensions sed in the analysis, cold be estimated. For these prposes a principal component analysis based both on absolte and relative sizes, was performed. The reslts of the stepwise discriminant analysis show highly significant differentiation between sexes, and also permit specimens of nknown sex to be classified. References Arrighi J. and Salotti M Le chat savage (Felis silvestris Schreber, 1777) en Corse. Confirmation de la presence et approche taxonomiqe. Mammalia 52: Bchalczyk T. and Rprecht A. L Skll variability of Mstela ploris Linnaes, Acta theriol. 22: De Marinis A. M., Nikolov'H. and Gerasimov S On the sexal dimorphism of stone marten Martes foina Erxl. (Carnivora: Mstelidae) in Italy. Hystrix 2: Dixon W. J. (ed.) BMDP Statistical Software. University of California Press, Berkeley, Los Angeles: French D. D., Corbet L. K. and Easterbee V Morphological discriminants of Scottish wildcats (Felis silvestris), domestic cats (F. cats) and their hybrids. J. Zool., Lond. 214: Gerasimov S Species and sex determination of Martes martes and Martes foina by se of systems of craniometrical indices developed by stepwise discriminant analysis. Mammalia 49: Petrov I., Gerasimov S. and Nikolov H Metric characteristics and sexal dimorphism of cranial featres of wild cat (Felis silvestris Schreber, 1777: Mammalia, Felidae) in Blgaria. Acta zool. blgarica 40: Petrov I., Nikolov H. and Gerasimov S Craniometrical characteristics and sexal dimorphism of domestic cat (Felis cats L. 1758) in Blgaria. Acta zool. blgarica 45: Reig S Errors and misjdgements of otliers in morphometric stdies. Fifth Intern. Ther. Congress, Rome, Abstracts of papers and posters, vol. 2:

8 388 I. Petrov et al. Schaenberg P., L'identification d Chat forestier d'erope Felis s. silvestris Schreber, 1777 par ne methode osteometriqe. Rev. sisse zool. 76: Wetzel M.-C., Solignac M. and Lefebvre J Application des analyses mltidimensionnelles a l'etde d dimorphisme sexel des chats harets des iles Kergelen, en comparison avec ne poplation de chats domestiqes d canton de Berne. Mammalia 46: Wiig O Sexal dimorphism in the skll of minks Mstela vison, badgers Meles meles and otters Ltra ltra. Zool. J. Linn. Soc. 87: Received 15 December 1989, revised 25 September 1992, accepted 30 October 1992.

9 Sexal dimorphism in skll of wild cat 389 APPENDIX 1. List of cranial and mandiblar measrements of wild cat Felis silvestris from Blgaria, taken according the scheme illstrated in Fig. 1 (after Petrov et al. 1990). Var. Fig. Measrement No. Measrements No. points Cranim 1. Total length L Condylobasal length Basal length Nerocranim length Basal nerocranim length Distance between os incisivm anterior and os palatinm posterior Soil palatal length Profile length Facial length Lateral length of snot oral border of the orbit Frontal length (between anterior edge alveols I 1 and proc. zygomatics) Distance between foramen infraorbital and crista sagittalis externa posterior 14. Greatest diameter of the orbit 15. Distance between proc. zygomatics and crista sagittalis externa posterior 16. Basifacial axis Check-bone length Crista occipitalis length Crista sagittalis length Greatest length of the nasal bone Palatal length: the medial point of inter-section of the line joining the deepest indentations of the Choanae - Prosthion 22. Length of the lamina horizontalis ossis palatini Length of the os maxillare Distance between os frontale posterior and crista sagittalis externa anterior 25. Os nasale length Rostrm breadth: measred across oter borders at the canine alveoli Facial breadth (between the infraorbital foramina - least distance) Distance between eye-orbits (external) Zygomatic breadth (greatest) Least breadth between the orbits Least breadth aboral of the spraorbital processes Distance between the hind edges of osa palatina Cranial breadth behind cheek-bone arcks Greatest breadth of braincase Greatest mastoid breadth Distance between pper-jaw joints (external) from to

10 390 I. Petrov et al. Appendix 1. Figs 1-6. Scheme of the wild cats' Felis silvestris skll showing the location of measring points (after Petrov et al. 1990). 1 - dorsal view, 2 - ventral view, 3 - lateral view, 4 - frontal view, 5 - posterior view of the skll, 6 - lateral lingal view of the mandible.

11 Sexal dimorphism in skll of wild cat 391 APPENDIX 1 - contined Distance between pper-jaw joints (internal) Palatal breadth: measred between alveoli C 1 (internal) not shown 39. Greatest palatal breadth: measred across the borders of the alveoli M 1 (internal) 40. Distance between processs zygomatics and os zygomaticm Frontal breadth Distance between the meata acústica externa (least) Least distance between the aditory bllae Breadth measred across the oter borders at the aditory bllae Length of the aditory blla: from the most aborolateral points to the not shown most oromedial point 46. Least diameter of the aditory blla: from the middle of the opening of the external acostic meats p to the most medial protrsion of the blla on the opposite side of the blla 47. Breadth of processs pterygoides not shown 48. Greatest breadth of the occipital condyles Facial height: measred throgh the hind end of os nasale 4 25 h 50. Greatest height of the cranim 4 45 h 51. Cranial height: measred throgh the aditory blla Height of the protberantia occipitalis Height of os incisivm at alveoli incisiva Rear height Nerocranim capacity not shown 57. Height of the foramen magnm Breadth of the foramen magnm Greatest diameter of the meats acstics externs not shown 60. Least diameter of the meats acstics externs not shown 61. Distance between foramina incisiva Distance between foramen ovale and foramen rotndm Distance between foramen ovale and fissa orbitalis Distance between foramen ovale and foramen opticm Distance between foramina ovalia Distance between os incisivm anterior and os frontale posterior Least diameter of the orbit not shown 68. Breadth of the nasal opening: measred between osa incisiva not shown 69. Height of the nasal opening: measred between os incisivm and os nasale Distance between lacrimal dcts (canalis lacrimalis) 71. Distance between os incisivm anterior and crista sagittalis externa anterior 72. Distance between os nasale posterior and crista sagittalis externa posterior 73. Distance between foramen sphenopalatinm and canalis lacrimalis Distance from the arch of os palatinm to canalis palatins

12 392 I. Petrov et al. APPENDIX 2 - conclded Length of maxilar teeth row from the fore end of alveols I 1 not shown to the hind end of alveols M Length of incisive teeth row not shown 77. Distance between alveols I 3 and alveols C 1 not shown 78. Greatest diameter of C 1 : measred near to the alveols not shown 79. Least diameter of C 1 : measred near the alveols not shown 80. Height of the C 1 : measred above the alveols not shown 82. Distance between posterior edge of alveols C 1 and anterior edge not shown of alveols P Length of the cheektooth row: measred on the bccal side 2 P 2 M Teeth row length: measred between posterior edge of alveols C 1 2 C 1 M 1 and posterior edge of alveola M Length of the premolar row: measred along the alveoli on the 2 p2 p4 bccal side 86. Length of premolar row on P 3 and P 4 2 P3 P Length of P 4 : measred on the bccal side not shown 88. Breadth of P 4 not shown 89. Distance between the tips of C 1 not shown Mandible 90. Length of the mandible: measred between anterior edge of alveols Ii and processs coronoides posterior 91. Length of the mandible: measred between anterior edge of alveols Ii and the middle of processs condyloides 92. Length of the mandible: measred between anterior edge of alveols Ii and processs anglaris posterior 93. Length of processs condyloides not shown 94. Distance between processs condyloidei not shown 95. Height of the rams mandiblae Total length of the mandible teeth row 6 Ii Mi 97. Distance between anterior edge of alveols Ii and incisra mandiblare Distance from anterior edge of alveols Ii to the angle between processs condyloides and processs anglaris 99. Distance between posterior egde of alveols Ci to processs coronoides 6 Ci Distance between posterior egde of alveols Ci and the angle between 6 Ci 57 processs condyloides and processs anglaris 101. Distance between posterior egde of alveols Ci and processs anglaris 6 Ci Height of corps mandiblae measred in the point between P3 and P4 not shown 103. Height of corps mandiblae measred behind Mi not shown 104. Total height of corps mandiblae Mandiblar breadth after Ci not shown 106. Distance between the middles of alveols Ci (internal) not shown 107. Distance between processs coronoidei not shown 108. Distance between processs anglaries not shown

13 Sexal dimorphism in skll of wild cat 393 APPENDIX 1 - contined Distance between anterior edge of alveols Ii to foramen mandiblare Distance between foramen mandiblare and processs anglaris Greatest diameter of Ci measred near to the alveols not shown 112. Distance from posterior edge of Ci to posterior edge of Mi 6 Ci Mi 113. Length of cheektooth row: measred between anterior edge of P3 6 Pa Mi posterior edge of Mi 114. Length of premolar row 6 P 3 P Length of Mi not shown 116. Greatest breadth of Mi not shown 117. Distance between the tips of Ci not shown 118. Breadth of Ci measred near alveols not shown 119. Height of Ci measred above the alveols not shown 120. Greatest gage of corps mandiblae not shown 121. Total length of the Ci not shown

14 394 I. Petrov et al. APPENDIX 2. Arithmetic means and standard deviations of 118 craniometrical measrements of wild cat Felis silvestris from Blgaria (nmeration and description of the measrements after Petrov et al. 1990, except V10, V53, V81). Males Females I V X SD min max X SD min max Cranim

15 Sexal dimorphism in skll of wild cat 395 APPENDIX 1 - contined

16 396 I. Petrov et al. APPENDIX 2 - conclded Mandible