Component Analysis of Osteometric Traits in Randombred House Mice
| Autor: | Larry Leamy |
|---|---|
| Rok vydání: | 1975 |
| Předmět: | |
| Zdroj: | Systematic Biology. 24:176-190 |
| ISSN: | 1076-836X 1063-5157 |
| DOI: | 10.1093/sysbio/24.2.176 |
| Popis: | Leamy, L. (Department of Biology, California State University, Long Beach, California 90840) 1975. Component analysis of osteometric traits in randombred house mice. Syst. Zool. 24:176-190.-A multivariate component analysis with an oblique factor solution was performed on 18 (15 osteometric and 3 external metric) traits in male and female randombred house mice of 3 different ages (1-month, 3-months, and 5-months). Approximately 77% of the total variance was explained by 4 factors in the 1-month mice, and by 5 factors in the 3-month and 5-month individuals. Skull, limb, pelvic girdle, and width factors (listed in order of importance) were identified for the 1-month age group, the external measures mostly associating with the skull factor. In the 3-month and 5-month age groups, girdle, skull length, limb, skull width, and tail factors were evident. There were considerable differences between the sexes throughout the age groups in the delineation of specific variables (especially the extemal ones) into factors. Factor patterns were explained largely in terms of differential growth and general anatomical and functional considerations. The direct contributions of the factors to the specific variables most affected were taken to represent the effects of a combination of additive genetic and residual environmental influences. The totals of the residual direct contributions and the indirect contributions of the factors for eaeh of the variables were found to be highly correlated with previous estimates of non-genetic maternal effects. [Osteometry; mice.] In morphological investigations involving a sufficiently large battery of metric variables, correlations among all variables typically generate a matrix of bewildering complexity. Fortunately several multivariate techniques have been developed which permit the simultaneous analysis of such traits even when no a priori assumptions are made regarding their functional interrelationships. Perhaps the most notable among these techniques are those of principal component analysis and factor analysis (Harman, 1967), both of which currently find application in a broad range of biological investigations (Seal, 1964; Gould and Garwood, 1969). Regardless of the specific design employed, the essence of these techniques is the reduction of a large set of interrelated variables into a smaller set of causal factors or components. The basic principles and procedures of these analyses may be found in Thurstone (1947), Adcock (1954), Seal (1964), and Harman (1967). Mammalian dental and/or skeletal dimensions constitute systems which have been much subjected to factor (Wallace and Bader, 1967; Riddle, 1971) and component (Jolicoeur and Mosimann, 1960; Rees, 1969) analysis. However, comparisons of the results of these and other investigations are somewhat difficult since different variables and factor designs are often chosen by various investigators. Another perhaps more fundamental problem is our basic ignorance of the relationship between the "factors" of factor analysis, and the conventional genetic and environmental components of variance obtained in a quantitative genetical analysis. Only rarely have such characters been concomitantly subjected to both associative and genetical techniques (Bailey, 1956; Yap Potter et al., 1968). Recently I completed a rather extensive quantitative genetical analysis of 15 osteometric and 4 external metric characters in a randombred population of house mice (Leamy, 1974). Estimates of the magnitude of genetic (heritabilities) and environmental (maternal effects) components were |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|