The Auk 109(3):530-542, 1992 EVOLUTION IN THE ROCK DOVE: SKELETAL MORPHOLOGY RICHARD F. JOHNSTON Museumof NaturalHistory and Department of Systematicsand Ecology, 602 DycheHall, The Universityof Kansas,Lawrence, Kansas 66045, USA ABSTRACT.--Domesticpigeons were derived from Rock Doves (Columbalivia) by artificial selection perhaps 5,000 ybp. Fetal pigeon populations developed after domesticsescaped captivity; this began in Europe soon after initial domesticationsoccurred and has continued intermittently in other regions. Ferals developed from domesticstocks in North America no earlier than 400 ybp and are genealogicallycloser to domesticsthan to European ferals or wild RockDoves. Nevertheless, North American ferals are significantlycloser in skeletalsize and shapeto Europeanferals and Rock Doves than to domestics.Natural selectionevidently has been reconstitutingreasonable facsimiles of wild size and shape phenotypesin fetal pigeonsof Europeand North America.Received 17 April 1991,accepted 13 January1992. Man, therefore, may be said to have been and southwestern Asia; this is known to be true trying an experiment on a gigantic scale; in more recent time (Darwin 1868; N. E. Bal- and it is an experimentwhich nature dur- daccini, pers. comm.). Later, pigeons escaping ing the long lapse of time has incessantly captivity either rejoined wild colonies or be- tried [Darwin 1868]. came feral, and are now found in most of the world (Long 1981). European,North African, Of the many kinds of animals examined for and Asiatic ferals may have historiesof several the study of variation under domesticationby thousandyears. CharlesDarwin, only for pigeons(Columba livia) North American ferals have a significantly did he describefully the chief domesticstrains, shorter history, stemming from British dove- along with "their history, the amount and na- cotepigeons (the earliestof which were brought ture of their differences,and the probablesteps to Britain by the Romans;Levi 1974)introduced by which they havebeen formed" (Darwin 1868: by Scottishand English immigrantsto Ameri- 1 [vol. I]). He did artificial selection and studied can Atlantic seaboard localities in 1600-1610 inheritance of plumage colors, color patterns, (Schorger 1952). North American ferals, there- and body size and shape in domesticpigeons; fore, are not directly lineally related to ferals the results of these studieswere important to of the Old World (Fig. I). Additionally, founder his work on natural selection (Darwin 1859, gene frequenciesseemingly departed signifi- 1868).Darwin's findings supported the ideathat cantly from thoseof Europeandomestics, judg- the range of colors,patterns, sizes, and shapes ing from allozymesof North Americanand Eu- shown by domesticstrains had antecedentsin the variation of wild Rock Doves. He also ropean ferals (Johnstonet al. 1989). Thus, the evolutionary derivation of ferals is thought that fetal pigeonswere an understand- more complex than it might have been. The able consequenceof domesticbirds escaping complexity is most useful--it is, for example, captivity. In the late 1850s,however, Darwin importantthat fetal pigeonswere derived from was heavily involved in writing his "big book" domestics more than once, because the devel- (Stauffer 1975), so origins of ferals from do- opmentof feralsin North Americacan be viewed mesticswere barely mentioned. Details of such as an independent replicate of the natural ex- origins, involving character variation molded presumablyby natural selection,are neverthe- periment in ferality tried in Europe and Asia. less of interest to thinking concerningpopu- Without the replicate, this study would almost lation differentiation; some details inferred from certainly not have been undertaken, nor would skeletal morphology are reported here. it in any event have a satisfactoryconclusion. Rock Doveswere domesticatedin the period Getting to that conclusionemploys assessment I0,000 to 5,000 ybp, earlier than has been pre- of skeletal similarities and differencesamong viously suggested(e.g. Sossinka1982). Domes- wild, domestic, and the two fetal lines of Rock tications evidently occurred many times Doves, and approximatinghow the similarities throughout the Mediterranean Basin,Near East, and differences could have occurred. 53O July1992] MorphologicEvolution in RockDoves 531 AMERICAN EUROPEAN WILD TAI•LE1. Sample sizes of wild, feral, and domestic DOMESTICS FERAL FERAL ROCK DOVE pigeon skeletonsused in different analyses.• Wild RockDoves.--Total sample (22 M, 23 F); Capo Caccia, Sardinia (19 M, 19 F). NorthAmerican ferals.--Total sample (111 M, 86 F); Lawrence, Kansas, USA (26 M, 22 F). Europeanferals.--Total sample (49 M, 61 F); Durham, Co., Durham, England (16 M, 27 F). Domesticpigeons.--Total sample (64 ?); large speci- mens (36 ?); small specimens(28 ?); racing homer (11 M, 7 F); tumbler (5 M, 4 F); runt (5 ?); pouter (4 ?); English carrier (5 ?); turbit (5 ?);jacobin (3 ?); fantail (2 ?). Fig. 1. Tree diagram depicting genealogicalrela- tionshipsof wild Rock Doves,domestic pigeons, and male; F = female; ? = sex unknown. two lineagesof feral pigeons. as racing homers,a post-Darwinian,artificially se- MATERIALS AND METHODS lected strain of domestic pigeon (Levi 1974), were removedfrom locality samplesand usedas one of the The geographicregion of this study is chiefly Eu- varietiesof domesticpigeon. Domesticsamples have rope and North America. Specimenscome from Can- a great range in size and shape,so they (depending ada, the United States, the British Isles, Switzerland, on the analysis)were sortedto varioussubsets: pooled Italy, Egypt, Israel,and Yemen. Wild C. liviaare found large domestics(keel length > 69 mm); small do- from the Faeroes, Shetland, Scandinavia and Russia mestics(keel length < 69 mm); and a number of do- south to Ghana, northern Chad, eastern Sudan, Yem- mestic strains maintained and identified by Charles en, Pakistan, and India. Ferals occur in most cities Darwin (runts plus pouters,jacobins plus fantails,En- worldwide, extensivelyin agricultural and maritime glish carriers,turbits, and tumblers).These strains are habitats,and on many isolatedoceanic islands (Long composedof substrainsand some, such as tumblers, 1981). are more variable morphologicallythan others (Dar- Darwin appreciatedthe value and usesof biological win 1868). Most of the domestics from the Darwin collections, so at the conclusion of his work with collectionwere preparedwithout sexbeing recorded; pigeonshis captive birds were preservedand stored there is no reliable way in which to estimatesex for at the British Museum of Natural History, along with thesespecimens, forcing the useof both sexesin many his other pigeon specimens.Fifty-three now exist as analyses. Although the wild and feral samples are whole skeletons, and are included below, one as a satisfactoryin sample sizes,associated label data, and wild specimenfrom Shetland and the remainder as methodsof preparation,the domesticspecimens are part of the domesticsamples. suboptimalin someof these respects.In sampleswith I securedrecent samplesand prepared the speci- sexespooled, however, their size and shape infor- mensat the University of Kansas.They include: wild mation seemsto be satisfactory. Rock Doves from Capo Caccia, 25 km W Alghero, The 16 skeletalvariables employed are listed in the Sardinia, November, 1989; ferals from Fertilia and Appendix for some representative samples;I took Sassari, Sardinia, November, 1989; Zurich and Basel, measurementsfor all specimens,which was consis- Switzerland, November, 1989;Washington, 17 km E tent with past practice (Johnston 1990). Except for Durham, England, January,1989; Lawrence, Douglas purely descriptive purposes(as in the Appendix) or County, Kansas,1983 to 1989;and BacaCounty, Col- in clusteranalysis (for which the data were standard- orado,June, 1987.The remaining specimenswere ex- ized), variables were transformed to natural loga- amined either at the British Museum (Natural His- rithms. Statisticswere processedon an IBM main- tory), Tring, England, or at Kansason loan from a framecomputer using BMDP (Dixon 1988)and NTSYS number of other museum collections (listed in the (Rohlf 1985), or on an MS-DOS 80286/287 personal acknowledgments).Sample sizes employed in the computer using BMDP PC90 (Dixon 1990). Missing study are listed in Table 1. data were computed for specimenslacking no more Point-locality specimensas noted abovewere some- than two variablesby meansof maximum-likelihood times usedas discreteanalytical units. For mostanal- estimation in which missing values were estimated yses,samples were pooled (Zink and Remsen 1986), by regressing the variable on all variables of speci- owing to a need for a relatively high level of gen- mensof the samesex and locality that had acceptable erality: 15 locality or regional samplesprovide the values in the specimenwith the missingvalue. "North American feral" sample,and seven were used Three ways to examinesimilarities and differences for the "Europeanferals." One large and severalfrac- in wild, domestic,and feral sampleswere used. For tional samples were used for "wild Rock Doves." each sample, univariate product-moment correlation Eighteenspecimens with leg bandsidentifying them coefficientswere computed for the 16 variables of the 532 RICHARDF. JOHNSTON [Auk, Vol. 109 TABLE 2. Abbreviations used in tables. RESULTS Abbreviation Full name Sample statistics.--Descriptivestatistics for males and females over the 16 variables for five MAXL Premaxilla length MAXW Premaxilla