University of Kansas Publications MUSEUM OF NATURAL HISTORY The University of Kansas Publications, Museum of Natural Histoi-y, beginning with volume 1 in 1946, was discontinued with volume 20 in 1971. Shorter research papers formerly pub- lished in the above series are now published as Occasional Papers, Museum of Natural History. The Miscellaneous Pub- lications, Museum of Natural History, began with number 1 in 1946. Longer research papers are published in that series. Monographs of the Museum of Natural History were initiated in 1970. All manuscripts are subject to critical review by intra- and extramural specialists; final acceptance is at the discretion of the publications committee. Institutional libraries interested in exchanging publications may obtain the Occasional Papers and Miscellaneous Publica- tions by addressing the Exchange Librarian, University of Kan- sas Library, LawTence, Kansas 66045. Individuals may pur- chase separate numbers of all series. Prices may be obtained upon request addressed to Publications Secretary, Museum of Natural History, University of Kansas, Lawrence, Kansas 66045. Editor: Richard F. Johnston PRINTED BY UNIVERSITY OF KANSAS PRINTING SERVICE LAWRENCE, KANSAS S'/l/^Lf^' I — :■ i:;ii"C.' ,..,0 OCCASIONAL PAPERS ^...v. of the MUSEUM OF NATURAL HISTORY The University of Kansas Lawrence, Kansas NUMBER 53, PAGES 1-8 APRIL 15, 1976 ESTIMATING VARIATION IN BONY CHARACTERS AND A COMMENT ON THE KLUGE-KERFOOT EFFECT By Richard F. Johnston^ Abstract Variation witliin and between samples of skeletal size characters of 3,295 specimens of house sparrows, Passer domesticus, was examined using sample variance, s^, and a measure of interlocality variation derived from analysis of variance. The specimens were from botli European and Nortli American localities, so possible ancestor-descendent relationsliips could be examined. It is clear diat variation in morphology within locality samples is uniform for a given variable, in^espective of continental sample source. Populations in North America show significant restriction of interlocality variation relative to Euro- pean samples, from wliich the fonner are lineally descendent witliin the past 123 years. Interlocality variation is not correlated with intralocality variation if the former is estimated from analysis of variance; there is such a correlation when simpler measures are used, as suggested earlier by Kluge and Kerfoot. Variation in natural populations ultimately is genetically-based, although envii-onments are important in dictating the ways in which phenotypes reflect genotypes. Such variation is thus a product of genetic-environmental interaction, and for certain assessments of the evolutionary process this can be ideal — natural selection occurs on the phenotype. But, if there is concern for morphological varia- tion itself, then not knowing much about relative contributions from genetic and environmental sources is a real constraint, limiting the range of questions that can be asked and the sorts of answers that may be expected. This paper examines aspects of variation within and between populations as estimated by samples of morphologic variables. The data are sizes of 14 bony elements in 3,295 specimens of house spar- 1 Museum of Natural History and Department of Systematics and Ecology, The Univer- sity of Kansas, Lawrence, 66045. 2 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY rows, Passer domesticus, assembled as several dozen locality samples from both Europe and North America. The two sexes serve as replicates for each continent. However, the continental samples do not allow replicate studies, for the European birds are known to be lineally-ancestral to the American birds as of 1852 (Barrows, 1889). We can expect the North American birds to show certain descendency characteristics, which may be as useful as replication. This paper is thus an extension of an earlier one in which the character-set was from skin specimens (Johnston and Selander, 1973). The questions asked here stem from consideration of the supposed translation of intralocality variation, between individuals, into inter- locality variation, between populations. Is there in fact an evident translation, at the simplest possible level? The answer is clouded by occurrence of environmental differences between localities, but a series of questions can still be approached, such as the following: Does variation in morphology within locality samples occur in such a way that one can predict the format of \'ariation in given char- acters, irrespective of locality? Is interlocality variation a predict- able consequence of intralocality variation, or, are the two corre- lated? Do descendent populations show interlocality variation that is predictable from the nature of either intralocality or interlocality variation of the ancestral populations? Materials and Methods The specimens used in this study were 1,357 males and 1,079 females from North America, and 479 males and 380 females from Europe (cf. tables 1, 2, 4, and 5). A subset of the European speci- mens from northwestern Europe was also employed and consisted of 184 males and 189 females (tables 3 and 6). The 14 bony char- acters included 5 from the skull, 4 from the central body core, and 5 from the limbs, and these are identified in each of the tables 1-6. Intralocality variance was studied by computing sample vari- ance, S-, for each character for each locality. For each character sample variance proved to be homogeneous by F,nax test (see also Selander and Johnston, 1967 ) , and thus the basic statistic used was mean sample variance for each sex for each continental set of localities. Interlocality variation was computed as that fraction of the total variation for a character that was owing only to variation between samples (Sokal and Rinkel, 1963; Sokal and Rohlf, 1969; cf. John- ston and Selander, 1973), in accord with the following relationship: I.V. = ''- (100), s.' + s^- in which I.V. is interlocality variation, s~ is mean intralocality ESTIMATING VARIATION IN BONY CHARACTERS 3 Table 1. — Ixtralocality and Ixterlocality Variation in Skeletal Characters of North American Male House Sparrows: 1962-1970/ Mean Intralocality Per Cent Variable Variance Interlocality Variation Skull premaxilla length 0.077 9.34 skull width 0.129 6.25 skull length 0.316 17.95 dentary length 0.068 6.13 mandible length 0.267 10.21 Mean interlocality variation for skull variables 9.98 Core coracoid length 0.253 13.26 sternum length 0.438 14.57 keel length 0.677 11.10 sternum depth 0.178 15.63 Mean interlocality variation for core variables 13.64 Limbs humerus length 0.225 23.38 ulna length __ 0.296 ■ 12.35 femur length 0.276 17.18 tibiotarsus length 0.759 11.42 tarsometatarsus length 0.458 11.35 Mean interlocality variation for limb variables 15.17 Overall per cent of interlocality variation 12.93 1 Number of groups (localities), 37; number of specimens, 1357. Table 2. — Intralocality and Interlocality Variation in Skeletal Characters of European Male House Sparrows: 1962-1970.^ Mean Intralocality Per Cent Variable Variance Interlocality Variation Skull premaxilla length 0.079 22.25 sk-ull wdth 0.127 26.08 - skull length 0.359 36.29 dentary length 0.071 18.39 mandible length 0.273 22.66 Mean interlocality variation for skull \'ariables 25.13 Core coracoid length 0.229 29.96 sternum length 0.278 39.83 keel length 0.778 25.97 sternum depth 0.171 31.05 Mean interlocality variation for core variables 31.70 Limbs humerus length 0.262 27.02 ulna length 0.323 29.01 femur length 0.283 23.51 tibiotarsus length 0.722 24.56 tarsometatarsus length 0.453 24.37 Mean interlocality \ariation for limb variables 25.69 Overall per cent of interlocality variation 27.51 ^ Number of localities (groups), 25; number of specimens, 479. 4 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY variance, and .?a" is obtained as the ANOVA mean square between groups less the mean square within groups divided by the mean locaHt)^ sample size. The resultant values of I.V. are per cent frac- tions of the variation for the character over the set of localities represented. I.V. was computed for both sexes in the North Ameri- can, European, and northwestern European sample sets, and these values appear in column 2 in tables 1-6. Results Variation icithin locality samples. — Mean character variance is scaled in an uncomplicated way on size of character. The means hold no surprises, from sex to sex or from continental set to conti- nental set. For females, the North American samples have variances slightly smaller than for Europe as a whole, and slightly larger than for eight samples from England, France, and Germany. In males the picture is similar, although the means for European and North American samples are much more nearly alike. There are no sig- nificant differences in variance between the sexes, although for both continental sets females have slightly larger variances for most variables. European females would be significantly more variable than European males for sternum length if only European locality sets were considered, but this disappears if the North American as well as the European specimens are used. Adjusting for size of character by computing the coefficient of variation shows the premaxilla, dentary, keel, and sternum depth to be around 4%, and these are the most variable characters among the 14 here examined. This is as true for the North American sam- ples as it is for the European, a point of some note. Vanation between locality samples. — For females, about 10% of the character variation in North American populations is between localities, which contrasts sharply with values of about 30% for European samples and 27% for the northwestern European samples. The same pattern is seen in males, although it is somewhat less pronounced: about 14% of the variation is between localities for North American, 27% for European, and 21% for northwestern Euro- pean sample sets. At the level of individual characters, inspection of the tables suggests there is no relationship between intralocality variance and interlocality variation; in fact, rank correlation analysis shows there to be no relationship between interlocality variation and intralo- cality coefficient of variation for either sex in any of the sample sets. Discussion One noteworthy aspect of this examination is that North Ameri- can samples shown significantly less interlocality variation than the European samples. Causes for lesser variation in the descendent ESTIMATING VARIATION IN BONY CHARACTERS Table 3. — Intralocality and Interlocality Variation in Skeletal Characters of Northwestern EuropeaxN Male House Sparrows: 1962-1970." Mean Intralocality Per Cent Variable Variance Interlocality Variation Skull premaxilla lengtli 0.093 14.76 skull width 0.121 26.51 skull length 0.336 24.73 dentary length 0.0S3 9.80 mandible length 0.261 23.43 Mean interlocality variation for skull variables 19.84 Core coracoid length 0.202 27.73 sternum length 0.429 22.87 keel lengtli 0.753 15.58 sternum depth 0.154 9.63 Mean interlocality variation for core variables 18.95 Limbs humeiTis length 0.159 27.51 ulna length 0.251 23.02 femur length 0.218 22.05 tibiotarsus length 0.530 15.49 tarsometatarsus length 0.451 13.55 Mean interlocality variation for limb variables 20.32 Overall per cent of interlocality variation 19.70 1 Oxford, London, England; Paris, France; Mainz, Oberstenfeld, Ludwigsburg, Stuttgart, and Radolfzell, Germany; number of specimens, 184. Table 4. — Intralocality and Interlocality Variation in Skeletal Characters of North American Female House Sparrows: 1962-1970." Mean Intralocality Per Cent Variable Variance Interlocality Variation Skull premaxilla length 0.078 3.70 skull width 0.097 12.27 sk-ull length 0.324 10.33 dentarv length 0.072 6.31 mandike length 0.290 4.11 Mean interlocality variation for skull variables 7.34 Core coracoid length 0.280 12.06 sternum length 0.469 10.17 keel length 0.652 8.97 sternum depth 0.175 11.31 Mean interlocality variation for core variables 10.62 Limbs humenis length 0.253 11.85 ulna length 0.362 9.36 femur length 0.310 13.68 tibiotarsus length _____ 0.844 7.78 tarsometatarsus length 0.530 9.52 Mean interlocal! t>' variation for limb variables 10.44 Overall per cent of interlocality variation 9.43 1 Number of groups (localities), 37; niunber of specimens, 1,079, 6 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY populations include restriction of variation in the innoculating sam- ples brought to North America, and something on the order of a population bottleneck has been suggested as having occurred in 1852-60. Yet, persistent small populations almost certainly did not occur, to judge from Barrows' account ( 1889). Also, it is simple to demon- strate normal within-population variation for all morphometric variables examined, something that could not happen had a major genetic bottleneck occurred. To account for restricted inter locality variation in North American sparrows, I suggest that we are wit- nessing the manufacture of continent-wide interlocality variation from antecedent intralocality variation — that \\'hich was originally present in the original samples of the predominantly English house sparrows. The English birds almost certainly contained only a frac- tion of the allele population then characteristic of continental Euro- pean birds. It is, of course, that considerably greater allelic popu- lation that supports the robust interlocality variation we currently see in the European samples. Incidentally, electrophoretic varia- tion supports this view — Klitz (1972) found five polymorphic enzyme loci in continental European samples, and only three of these were polymorphic in England, and just two in North America. We have a clear instance of founder effect sampling error without any of the genetic traumata associated with small effective popula- tion size. The analyses of the present data sets are not inconsistent with this interpretation (tables 1-6). The values for per cent I.V. for the northwestern European samples tend to be intermediate be- tween the total European and the North American specimens, especially the males. Another intriguing aspect of the examination is that there is no evident relationship between intralocality sample variance and vari- ation between samples. The Darwinian thesis assumes that evolu- tion occurs through selective reassortment of variation between indi- viduals in a population to produce variation between populations. Modern genetics supports the Dar\\'inian \iew, and allows us to think of second-order variation as coming from recombination, epistasis, nongenetic environmental induction, and selection. There should therefore be nothing mysterious about interlocality variation — intrapopulation variability surely comes first, and there should be some traces of that variation in any instance of interlocality varia- tion. So, since there is no predictable trace, is there something peculiar about house sparrows, or should there be other means of assessing the kinds of variation with which we are concerned? House sparrows probably are not peculiar in the sense intended above. It is therefore almost necessary to conclude that the samples at hand cannot provide the link between intrapopulational and in- terpopulational variation. This could be because the measures of ESTIMATING VARIATION IN BONY CHARACTERS 7 Table 5. — Ixtralocality and Interlocality Variatiox ix Skeletal Characters of European Female House Sparrows: 1962-1970/ Mean Intralocality Per Cent Variable Variance Interlocality Variation Skull premaxiUa lengtli 0.100 20.00 skull ^yidih 0.136 21.84 skull length 0.343 31.94 dentary length 0.085 18.82 mandible length 0.346 28.14 Mean interlocality variation for skull variables 24.15 Core coracoid length 0.261 35.38 sternum length 0.611 31.03 keel length 0.799 33.03 sternum depth 0.183 40.98 Mean interlocality variation for core variables 35.11 Limbs humerus length 0.304 • 31.38 ulna length 0.390 30.48 femur length 0.325 25.48 tibiotarsus length 0.890 24.83 tarsometatarsus length 0.553 23.93 Mean interlocality variation for limb variables 27.22 Overall per cent of interlocality variation 28.82 1 Number of groups (localities), 23; number of specimens, 380. Table 6. — Ixtralocalit\' and Interlocality Variation in Skeletal Characters of Northwestern European Female House Sparrows: 1962-1970.' Mean Intralocality Per Cent Variable Variance Interlocality Variation Skull premaxilla length 0.104 18.59 skull uidth 0.105 19.31 skull length 0.301 31.43 dentary length 0.075 17.31 mandible lengtli 0.264 37.70 Mean interlocality variation for skull \'ariables 24.87 Core coracoid length 0.225 35.64 sternum length 0.547 31.87 keel length 0.761 23.28 sternum depth 0.188 20.05 Mean interlocality variation for core \'ariables 27.71 Limbs humems length 0.225 31.49 ulna length 0.321 32.04 femur length 0.291 31.82 tibiotarsus length 0.690 22.56 tarsometatarsus length 0.418 26.98 Mean interlocalit>' variation for limb variables 28.98 Overall per cent of interlocality variation 27.19 1 Oxford and London, England; Paris, France; Mainz, Oberstenfeld, Ludwigsburg, Stutt- gart, and RadolfzeU, Germany; niunber of specimens, 189. 8 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY intrapopulation variation always assess a relatively small, coadapted set of character-states, those within a locality sample. But, over the entire specimen set, representing all localities, there is no neces- sary coadaptation. In any event, the lack of a consistent effect of intralocality variance on interlocality variation is unexpected, and it does not support the classical Darwinian thesis as predicted above. Neither does it support Kluge and Kerfoot (1973), who examined the same set of phenomena with slightly different variables and concluded that interlocality variation reflected intralocality variation. The results of Kluge and Kerfoot have been obtained again by Sokal (in litt.) using a slightly different technique. His technique again avoids any measure of interlocality variation that includes great heterogeneity (defined here as owing to the statistical con- sideration of variables lacking a common selective background, as inevitably occurs when doing analysis of variance for a variable over a set of geographically widely-separated localities). Thus, I conclude here that any application of an ANOVA mean square between groups to give an estimate of interlocality variation for comparative use against measures of intralocality variation is likely to give irrelevant assessment of the question. Much simpler meas- ures of between-locality variability, such as range in character- state means as used by Kluge and Kerfoot, prove to be more satis- factory. Literature Cited Barrows, W. 1889. The English sparrow in North America. U.S. Dept. Agr., Div. Ornitliology and Mammology, Bull. 1:405 pp. Johnston, R. F., Selander, R. K. 1973. Evolution in the house sparrow. III. Variation in size and sexual dimorphism in Europe and North and South America. Amer. Nahir., 107:373-390. Klitz, W. 1973. Genetic consequences of colonization with subsequent ex- pansion: the house sparrow in North America. Ph.D. dissertation, Univ. Kansas. Kluge, A. C, Kerfoot, W. C. 1973. The predictability and regularity of character divergence. Amer. Natur., 107:426-442. Selander, R. K., Johnston, R. F. 1967. Evolution in the house sparrow, I. Intrapopulation variation in North America. Condor, 69:217-258. Sokal, R. R., Rinkel, R. C. 1963. Geographic variation of the alate Pemphi- gus populi-iransversiis in eastern North America. Univ. Kansas Sci. Bull., 44:467-507. Sokal, R. R., Rohlf, F. J. 1969. Biometry. Freeman, San Francisco.