INFORMATION TO USERS

This dissertation was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted.

The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction.

1. The sign or "target" for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing pagels) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity.

2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame.

3. When a map, drawing or chart, etc., was part of the material being photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again - beginning below the first row and continuing on until complete.

4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced.

University Microfilms

300 North Zeeb Road Ann Arbor. Michigan 48106 A Xerox Education Company 73-15,943 GO, Rodney Chun-Pung, 1945- THE GENETICS OF PIGEON POPULATIONS ON OAHU. ~liversity of Hawaii, Ph.D., 1972 Biology-Genetics

University Microfilms, A XEROX Company, Ann Arbor, Michigan

. © 1973

Hnc!tll'Y Chull-Punr Go

ALL RIGHTS RESERVED

THIS DISSERTATION HAS BEEN MICROFIlMED EXACTLY AS RECEIVED. l~E GENETICS OF PIGEON POPULATIONS ON OAHU

A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAII IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN GENETICS

DECEMBER 1972

By

Rodney C. P. Go

Dissertation Committee:

Geoffrey C. Ashton, Chairman Ming P. Hi Chin S. Chung Hampton L. Carson Andrew J. Berger PLEASE NOTE:

Some pages may have

indistinct print.

Filmed as received.

University Microfilms, A Xerox Education Company iii

TABLE OF CONTENTS

Page

LIST OF TABLES. .. . vi

LIST OF FIGURES • . xvi

ABSTRACT· ..• ..xviii

I. INTRODUCTION. 1

II. ELECTROPHORESIS, THEORY AND APPLICATION 5

III. THE PIGEON· •• 14

3.1. History • 14 3.2. Origin- . . . 14 3.3. Taxonomy- •..•...••• 16 3.4. Previous Genetic Analysis . 16

IV. SHORT DESCRIPTIONS OF LOCALITIES TRAPPED· 21

4.1. The Honolulu Zoo· ... 21 4.2. St. Andrew's Belfry ...... 22 4.3. Bethe1-Pauahi Building. 23 4.4. Waldron's Grainery. 23 4.5. Piers 27 and 32 . 25

V. TRAPPING PROCEDURES AND BLOOD COLLECTION AND STORAGE. 26

5.1. Blood Collection and Storage. 28

VI. STATISTICAL PROCEDURES .• 30

6.1. Introduction. ..• 30 6.2. Estimation of Gene Frequencies and a, Inbreeding Coefficient...•..... 31

6.2.1. Two Autosomal Alleles with No Dominance 31 6.2.2. 3 Autosomal Alleles with Dominance Phenogram, 3-3-6.•••••..•. 32

6.3. Two Allelic Sex-Linked Genes with Dominance in the Homogametic Sex; Data on Both Sexes .•. 33 6.4. Estimations of F, the Inbreeding Coefficient. . 34

6.4.1. Chi-Square Assuming Panmixia. ...•• 34

6.5. Test of Homogeneity of Gene Frequencies, Two Co-dominant Alleles ....•.•....•. 34 iv

Page

6.6. Tests of Independence (interaction, homogeneity) Between Samples. An RXC Contingency Chi-Square. 35 6.7. Heterogeneity X2 Test for Testing Homogeneity of an Underlying Genetic Hypothesis in Different Samp les...... 35 6.8. Simple Linear Regression Analysis. .••. 36 6.9. Haldane's Test for Sex-Linkage in Samples From Populations ..•.• 38 6.10. Woolf's Analysis ..•.. ..•. 39 6.11. Summary .•..• 40

VII. ANALYSIS OF THE TRANSFERRIN LOCUS .• 42

7.1. Introduction •..... 42 7.2. Materials and Methods •. 45 7.3. Results and Discussion 47 7.4. Summary . 53

VIII. ANALYSIS OF THE 6-PHOSPHOGLUCONIC ACID DEHYDROGENASE LOCUS· ..•..• 54

8.1. Introduction .•.• 54 8.2. Materials and Methods· . 56 8.3. Results and Discussion . 58 8.4. Summary ..•..• 64

IX. ANALYSES OF THE ALKALINE PHOSPHATASE LOCUS 66

9.1. Introduction •...... 66 9.2. Materials and Methods .• 68 9.3. Results and Discussion • 70 9.4. Summary ..•..... 76

X. ANALYSES OF MORPHOLOGICAL CHARACTERS 77

10.1. Feathered Shanks ....• .. .•.. 77 10.2. Analysis of Plumage Color and Pattern Phenotypes . 79 10.3. Analysis of Eye Color Phenotypes . 84 10.4. Sex and Age Distribution . 86 10.5. Summary. . .. •...•. 89

XI. POPULATION STRUCTURE AND INBREEDING. 91

11.1. Randomly Mating Populations and Hardy-Weinberg Equilibrium. ..••.....••..• 91 11.2. The Inbreeding Coefficient: Its Methods of Estimation and Powers of Detection 92 11.3. Population Structure. •.•.•.••. 97 v

Page

11.4. Analysis of Population Structure in Pigeon Populations ..••.•..• .108 11.5. Assortative Mating and Inbreeding. • · .114 11.6. Analysis of Assortative Mating in the Zoo Population ...... ·. .120 11.7. Homing Behavior and Its Consequences .•. • .122 11.8. Evolutionary Consequences of Population Structure and Assortative Mating · .124 11.9. Summary . • .128

XII. ELECTROPHORETIC VARIABILITY IN A SINGLE PIGEON POPULATION .•... · ..131

12.1. Introduction •. ·• .131 12.2. Materials and Methods. • .131 12.3. Results and Discussion • .142 12.4. Summary...... 146

XIII. BREEDING DATA .. · .147

13.1. Introduction . · .147 13.2. Materials and Methods. · .147 13.3. Results and Discussion .148 13.4. Summary. .•.. · .148

XIV. SUMMARY. . •• .150

APPENDIX 1 • .153

LITERATURE CITED •. .273 vi

LIST OF TABLES

Page

7.3.1. Heterogeneity chi-square test for homogeneity of samples over all areas - Transferrin locus.... 157

7.3.2. Contingency table of transferrin phenotypes within sexes ...... · · · ·· · · · · · · ·· · 158 7.3.3. Contingency table of transferrin phenotypes within ages...... ··· · · · · · · ·· · · · 158 7.3.4. Contingency table of transferrin phenotypes within shank phenotypes. · ·· · · · · · · ·· ·· · 159 7.3.5. Contingency table of transferrin phenotypes within the white and colored populations ..• 159

7.3.6. Contingency table of transferrin phenotypes within eye color phenotypes ...••.. 160

7.3.7. 2 X 2 contingency table of transierrin homozygote and heterozygote frequencies within the sexes. • 160

7.3.8. 2 X 2 contingency table of homozygote and heterozygote frequencies within ages - the transferrin locus....•..•••..•.•... 161

7.3.9. 2 X 2 contingency table of homozygote and heterozygote frequencies within shank phenotypes - transferrin locus ...••...•.• 161

7.3.10 2 X 2 contingency table of homozygote and heterozygote frequencies within the white and colored populations - the transferrin

locus...... 't •••••• 162

7.3.11. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between sexes - the transferrin locus...... 163

7.3.12. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between ages - the transferrin, 6-phosphogluconate dehydrogenase, and alkaline phosphatase loci •... 164 vii

Page

7.3.13. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between the white and colored populations - the transferrin, 6-phosphogluconate dehydrogenase, and alkaline phosphatase loci. .. •.•••...... •165

7.3.14. Woolf's analysis of 2 X 2 comparisons of the eye color phenotypes, to test for differences in homozygote and heterozygote frequencies the transferrin locus ..•. ..•• .166

7.3.15. Contingency table to test first order interactions between the transferrin and 6-PGD loci••••...•167

7.3.16. Contingency table to test first order interactions between the transferrin and alkaline phosphatase loci...• ...... 167 7.3.17. Contingency table of transferrin phenotypes over the years 1969-1972 •.... .168

7.3.18. Contingency chi-square values and their associated p values, to test for homogeneity of transferrin phenotype frequencies between sexes within localities169

7.3.19. Contingency chi-square values and their associated p values, used for testing the homogeneity of transferrin phenotypes between ages within localities170

7.3.20. Contingency chi-square values and their associated p values, used to test first order interactions between the transferrin locus and shank locus, within their respective localities...... 171

7.3.21. Contingency table of transferrin phenotypes within the orange eye color and within localities...•..172

7.3.22. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between the sexes, and within localities - the transferrin locus 173

7.3.23. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between ages and within localities - the transferrin locus..•...... 174

7.3.24. Heterogeneity chi-square to test for heterogeneity between the samples from the different localities - the transferrin locus ••.....•••.•.•..175 viii

Page

7.3.25. Test of homogeneity of gene frequencies for the different localities - transferrin locus••.••176

7.3.26. Contingency chi-square values and their associated p values, used in testing the homogeneity of transferrin phenotypes bet'l1een two localities.. ..•. .. .•. .177

7.3.27. Contingency table to test the homogeneity of transferrin phenotypes between localities ••. .178

7.3.28. Estimates of p, q, and a for the transferrin locus gotten from Yasuda's G-type program •..••.179

7.3.29. Woolf's analysis of homozygote and heterozygote frequencies between two localities - the transferrin locus ....•••.••.... .180

8.3.1. Heterogeneity chi-square test of samples taken from different areas - the 6PGD locus •. .181

8.3.2. Table of pair-wise comparisons of heterozygote and homozygote frequencies, between areas within the zoo - the 6PGD locus ..•..•.• 182

8.3.3. Contingency table of 6PGD phenotype frequencies within the sexes - the zoo. ....•....• 183

8.3.4. Contingency table of 6PGD phenotype frequencies within ages - zoo ....••.•..•.•.• 183

8.3.5. Contingency table to test interaction between the 6PGD and shank loci - zoo ...••. .. . •.184

8.3.6. Contingency table of 6PGD phenotype frequencies within the white and colored population - zoo •...184

8.3.7. Contingency table of 6PGD phenotype frequencies within eye color phenotypes - zoo . . .•185

8.3.8 2 X 2 contingency table of homozygote and heterozygote frequencies within the sexes - zoo .••185

8.3.9. 2 X 2 contingency table of homozygote and heterozygote 6PGD frequencies within ages - zoo ...186

8.3.10. 2 X 2 contingency table of homozygote and heterozygote 6PGD frequencies within shank phenotypes - zoo ..••..•...... •...186 ix

Page

8.3.11. 2 X 2 contingency table of homozygote and heterozygote 6PGD frequencies within the white and colored population - zoo •...•..••. 187

8.3.12. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between shank phenotypes - transferrin, 6PGD, and APH systems ...... ••... 188

8.3.13. Woolf's analysis of pair-wise comparisons of homozygote and heterozygote 6PGD frequencies between two eye color phenotypes - zoo .... .•• 189

8.3.14. Contingency table to test for first order interactions between the 6PGD and alkaline phosphatase loci...... •.. .• 190

8.3.15. Contingency table of 6PGD phenotype frequencies over the years 1969-1972. • ...•.... 190

8.3.16. Heterogeneity chi-square test of the underlying genetic hypothesis - 6PGD locus ...• .. .•. 191

8.3.17. Estimates of p, q, and a for the different localities, obtained from Yasuda's G-type program ...... 192

8.3.18. Test of homogeneity of 6PGD gene frequencieo between localities. . •...... • 193

8.3.19. Contingency X2 values and their associated p values of 6PGD phenotype frequencies between two localities•...•...... •..• 194

8.3.20. Woolf's analysis of pair-wise comparisons of homozygote and heterozygote 6PGD frequencies between two localities...... 195

8.3.21. Contingency chi-square values and their associated p values of 6PGD phenotype frequencies between the sexes, with the data subdivided by localities ....••. 196

8.3.22. Contingency chi-square values and their associated p values of 6PGD phenotype frequencies between ages, with the data subdivided by localities.••...... •...• 197 x

Page

8.3.23. Contingency chi-square values and their associated p values of 6PGD phenotype frequencies between shank phenotypes, with the data subdivided by localities ...•• .. 198

8.3.24. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between the sexes, with the data subdivided by localities. ••• ...... • 199

8.3.25. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between ages, with the data subdivided by localities ...... 200

9.3.1. Heterogeneity chi-square test of samples taken from different areas within the zoo - alkaline phosphatase locus •.•.. 201

9.3.2. Woolf's analysis of pair-wise comparisons of APH homozygote and heterozygote frequencies between areas within the zoo .....•..• 202

9.3.3. Contingency table of APH phenotype frequencies within the sexes - zoo •.•.....••.• 203

9.3.4. Contingency table of APH phenotype frequencies within ages - zoo...... ••• 203 i 9.3.5. Contingency table of APH phenotype frequencies within shank phenotypes - zoo . 204

9.3.6. 2 X 2 contingency table of homozygote and heterozygote APR frequencies within the sexes - zoo . 204

9.3.7 2 X 2 contingency table of homozygote and heterozygote APH frequencies within ages - zoo . 205

9.3.8. 2 X 2 contingency table of homozygote and heterozygote APH frequencies within shank phenotypes - zoo •. ..••... .. 205

9.3.9. Contingency table of APH phenotype frequencies withi~ the white and colored populations - zoo 206 xi

Page

9.3.10. Contingency table of APH phenotype frequencies within the different eye color phenotypes - zoo ...... 206

9.3.11. Woolf's analysis of pair-wise comparisons of homozygote and heterozygote APH frequencies between two eye color phenotypes - zoo •....••. 207

9.3.12. Heterogeneity X2 test to test the homogeneity of the underlying genetic hypothesis over all the localities - APH locus. .... 208

9.3.13. Estimates of p, q, and a of the APH locus for the different localities, obtained from Yasuda's G-type program ...•..•...... •. 209

9.3.14. Test of homogeneity of APH gene frequencies from the different localities •• ...•. 210

9.3.15. Contingency chi-square values and their associated p values of APH phenotype frequencies between a pair of localities...•.•• 211

9.3.16. Woolf's analysis of pair-wise comparisons of homozygote and heterozygote APH frequencies between two localities. ••...•...... 212

9.3.17. Woolf's analysis to test for differences in homozygote and heterozygote APH frequencies between the sexes, with the data subdivided by localities ...... ••...... •.. 213

9.3.18. Woolf's analysis to test for differences in homozygote and heterozygote APH frequencies between ages, with data from the Bethel location...... •...... 214

9.3.19. Contingency X2 values and their associated p values of APH phenotype frequencies within the sexes, with data subdivided by localities .... 215

9.3.20. Contingency X2 values and their associated p values of APH phenotype frequencies within ages, with data s,~bdivided by localities...... 216

9.3.21. Contingency X2 values and their associated p values of APH phenotype frequencies shank phenotypes, with data subdivided by localities.... 217 xii

Page

9.3.22. Contingency table to test fox homogeneity of APH phenotype frequencies over the years 1969-1972 - zoo •.•.•...... •.. 218

10.1.1. 2 X 2 contingency table to test for inter­ action between shank phenotype and sex - zoo 219

10.1.2. 2 X 2 contingency table to test for inter­ action between shank phenotype and age - zoo 219

10.1.3. Contingency table of shank phenotype frequencies over the years 1969-1972 - zoo ..... 220

10.1.4. Contingency table of shank phenotype frequencies within localities...• 221

10.1.5. Test of homogeneity of shank allele frequencies between localities ... 222

10.1.6. Contingency table of shank phenotype frequencies within localities with the zoo excluded • ••• .•.. 223

10.1.7. Estimates of p, q, and a of the shank locus with separate estimates for each locality. . 224

10.1.8. Contingency chi-square values and their associated p values of shank phenotype frequencies within the sexes, with data subdivided by localities ....•... 225

10.1.9. Contingency chi-square values and their associated p values of shank phenotype frequencies within ages, with data sub- divided by localities.. . 226

10.1.10. Woolf's analysis to test for differences in homozygote and heterozygote frequencies between the sexes, with the data subdivided by localities•...••...... 227

10.2.1. Maximum likelihood frequencies of the alleles of the red plumage locus for each sex with data subdivided by localities .. 228

10.2.2. Haldane's test for deviations in phenotype frequencies from the expected in a dialle1ic sex-linked locus with dominance•.•.... 229 xiii

Page

10.2.3. Contingency chi-square values and their associated p values of red phenotypes within the sexes, with data subdivided by localities.•• . ... 230 10.2.4. Maximum likelihood estimates of the pattern allele frequencies for each locality .• . .. . 231 10.2.5. Contingency chi-square values with their associated p values of pattern phenotypes within two localities. • •.• 232

10.2.6. Contingency table of 5 major plumage color phenotypes within the localities, zoo excluded ...... 233

10.3.1. Contingency table to test for interaction between eye color and plumage color phenotypes .•••• ..•.. 234

10.3.2. Contingency table of eye color phenotype frequencies over the years 1969-1972, the numbers in parentheses are phenotype frequencies . 234

10.3.3. Chi-square values and their associated p values of eye color phenotype frequencies within the sexes, with data subdivided by localities ...... 235

10.3.4. Contingency table to test for homogeneity of eye color phenotypes between localities 236

10.4.1. Contingency table to test for differences in male and female frequencies between localities ...... 237

10.4.2. Contingency table to test for differences in young and mature frequencies between localities ...•.•...... 237

11.4.1. Variances and f values for the transferrin locus...... · 238 11.4.2. Variances and f values for the 6PGD locus. · .... 238 11. 4.3. Variances and f values for the alkaline phosphatase locus...... · . . .. 238 xiv

Page

11.4.4. Correlation coefficients over the different distances measured by steps for the transferrin locus..•...•..•...• 239

11.4.5. Correlation coefficients over the different distances measured by steps for the 6PGD locus ...... 240 11. 4.6. Correlation coefficients over the different distances measured by steps for the alkaline phosphatase locus•..••...•.•. 241

11. 4.7. Regression and correlation coefficients, and t values, for the transferrin, 6PGD, and alkaline phosphatase loci. 242

11.4.8. Variances and f values for the shank locus . 243

11. 4.9. Correlation coefficients over the different distances measured by steps for the shank locus...... 244

11.4.10. Regression table giving the regression and correlation coefficients and their t values. 245

11. 4.n. Variances and covariances found for the alleles of the pattern locus .•.••• 246

11.4.12. Correlation coefficients over the different distances measured by steps for the alleles of the pattern locus ...•.••.....• 247

11.4.13. Observed and expected correlation coefficients between alleles of the pattern locus and their associated Z values to test for differences between observed and expected. .... 248

11.6.1. Plumage color phenotype frequencies in the zoo population ..•...... • 249

11.6.2. Observed and expected frequencies of the various matings used in calculating the X2 obtained. 250

13.3.1. The different mating types and numbers and phenotypes of offspring produced from each type - the transferrin locus ..•.•.. 251 xv

Page

13.3.2. The different mating types and numbers and phenotypes of offspring produced from each type - the 6PGD locus••.. 252

13.3.3. The different mating types and numbers and phenotypes of offspring produced from each type - the alkaline phosphatase locus...... 253 xvi

LIST OF FIGURES

Figure Page

2.1. A Tise1ius electrophoresis cell and electrode assembly. •..• •. . •••••.•• 254

2.2. Electrophoretic pattern (descending) of human plasma ..•••..•.••.• 255

2.3. The counteracting effects of electroosmosis and electrophoresis . .• •••...•.• 256

2.4. Horizontal starch gel electrophoresis apparatus ..•. ••• .• ....•.• 257

4.1.1. A map of the Honolulu Zoo depicting the trapping areas within the Zoo complex •...•••. 258

4.5.1. A map of a portion of the Honolulu Harbor complex ...... •.•. 259

7.1.1. Diagramatic representation of the transferrin phenotypes. .•.•.•. .. • ...••• 260

8.1.1. Embden-Meyerhoff and pentose-phosphate pathways . 261

10.2.1. The three plumage pattern types found . 262

11.3.1. One dimensional stepping stone model of Kimura and Weiss ..•.•.•.....•..•..• 263

11.4.1. Graph illustrating the correlation of Tfr allele frequencies with distance (steps) .....•. 264

11. 4.2. Graph illustrating the correlation of 6PGD allele frequencies with distance. •.. •.•• 265

11.4.3. Graph illustrating the correlation of APH allele frequencies with distance... 266

11.4.4. Graph illustrating the correlation of shank allele frequencies with distance . 267

11. 4.5. Graph illustrating the correlation of pattern allele, CT, frequency with distance ....•.•.• 268

11.4.6. Graph illustrating the correlation of pattern allele, C, frequency with distance...•..••.• 269 xvii

Figure Page

11.4.7. Graph illustrating the correlation of pattern allele, c, frequency with distance. ...•••••. 270

12.3.1. Diagramatic representation of the 9 esterase phenotypes found...... 271 12.3,2. Diagramatic representation of the 6 possible 6PGD phenotypes ...... 272 xviii

Abstract

The extent of inbreeding found in natural populations of organ­

isms besides man has been examined in relatively few organisms or is

found to be insignificant. This thesis is therefore concerned in

finding the extent of inbreeding in pigeon populations, and if found,

to explain its significance. The major portion of this thesis is

also devoted to analysis of the genetic structure of pigeon popula­

tions and its implications toward evolution of the species.

Six populations of Columba livia were studied. A total of 1058 birds were trapped, with 792 of these coming from the zoo. An anal­ ysis of interaction between six loci: three electrophoretic,

transferrin, 6-phosphogluconate dehydrogenase, and alkaline phos­ phatase; and three morphological loci: plumage color, shank, and plumage pattern, failed to elucidate any interactions among the loci nor were any interactions found with sex and age. However, consider­ able heterogeneity in gene frequencies were found among localities, along with high F estimates in one of the populations.

With the discovery of this heterogeneity between populations, an analyeis into this heterogeneity was made in the ensuing section of the present thesis. The results revealed that different indicators of random differentiation between populations, in fact, suggest random differentiation for most loci, with gene frequency clines found for the APH and shank loci. Proof for the presence of three behaviors

(assortative mating, imprinting, and homing behavior) in the repertoire of Columba livia was given along with their effects on population xix

structure. The enhancing effects of the above behaviors on population subdivision is considered to supply the initial step in Wright's polyallelic random drift model.

Finally, a study of electrophoretic variation in a population of pigeons revealed that the proportion of loci polymorphic was only 18%.

This low proportion is ascribed to subdivision of the population, which through random drift and periodic reduction in size could lead to this low estimate. The choice of a majority of specific enzymes involved in energy production may also have given this lower estimate.

The average heterozygosity per individual was found to be .06, which was close to the maximum heterozygosity of .08.

Random differentiation is therefore considered the major force in gene frequency differences, caused by population subdivision and very little migration between subdivisions due to the homing behavior of the pigeon. This differentiation is therefore enhanced by assortative mating and imprinting. Hence, a low average heterozygosity would be expected due to the effects of subdivision and random differentiation. 1

I. Introduction

Within recent years, the study of population structure in humans as well as other animal populations has been undertaken. For, with the theoretical developments of Wahlund (1928), Dahlberg (1929),

Wright (1921 and later), Kimura and Weiss (1964), Malecot (1948, 1950,

1959), and Yasuda (1966), the importance of population structure in human and animal evolution was pointed out. The failure of the Hardy­

Weinberg Equilibrium Law to apply to natural populations, necessitated the development of new models of population structure, models that included the forces behind population structure, finite size, migra­ tion, and non-random mating.

Hence, Wahlund (1928) was able to formulate a mathematical expla­ nation of why subdivision of a large population results in an increase in homozygosity. In 1929 Dahlberg introduced his concept of isolate size, and its relationship between family size and consanguinity.

Wright (1921) prior to this had developed his concept of correlation between uniting gametes, the inbreeding coefficient.

Wright (1943, 1946, 1951) then formulated his model of isolation by distance; included within this model was the concept of isolate size. Assuming normal migration in two dimensions, he defined the neighborhood as the effective number in a circle radius of twice the standard deviation of offspring in one direction, and from this de­ rived a complicated relationship between neighborhood size, systematic pressure, and the inbreeding coefficient F. Wright (1943, 1946, 1951) also used a hierarchical definition of structure and applied his F statistics, of which F describes the degree of differentiation. ST 2

These F statistics were a great boon to investigators of population structure, for these could be defined in terms of the variance of gene frequencies.

Malecot (1948) then used a model taking advantage of marital distance between mates, the distance between birthplaces of two parents. He found that this eliminates the complication of neighbor- hood size and is independent from it. Therefore, F could be defined as a function of marital distance. Both Malecot (1950) and Kimura and

Weiss (1964) have looked at the stepping stone model of population structure, with the latter authors using the variance of gene fre- quencies to describe properties of the stepping stone model.

Yasuda (1968) then developed a theoretical model, based on

Wahlund's principle of subdivided populations, to estimate the coeffi- cient of kinship from mating type and phenotype frequencies. Morton later incorporated Yasuda's formulations_with the marital distance 1, , model of Malecot (1950, 1959) to develop the bioassay of population structure under isolation by distance model (Morton et al., 1968).

Several investigators have attempted to apply these models to natural populations. Nei and Imaizumi (1966) applied the stepping- stone model to blood group data in Japan. They found that the corre- lation of gene frequencies decrease with distance, and the covariance of the ABO blood group alleles were all negative. Therefore, they concluded that there was a significant random differentiation between subpopulations, defined as prefectures.

Petras (1967) working with Mus musculus found a numerical defi- ciency of heterozygotes for the Es-2 and Hb loci. Ecological data 3 supported the hypothesis of subdivision of the population. Wright's isolation by distance model was used to estimate Ne, which was esti­ mated to be 6-80.

Selander (1970) worked also on Mus musculus, but instead of using

F as a measure of differentiation, he used the variances of gene fre­ quencies as a measure of random drift and correlation of gene frequen­ cies to determine evidences for the tribal structure found in Mus.

The limited dispersal behavior of Mus, therefore, is a prerequisite for the population structure found.

Several other investigators also found incidences where behavior influenced the structure of populations. Wright and Hasler (1967) found that the homing behavior of the white bass (Roccus chrysops) and geographic distance were effective isolating mechanisms. The degree of isolation was determined by differences in serum proteins, and taxonomic characters between localities. Milne and Robertson

(1965) found that there was a splitting in a population of eider ducks

(Somateria M. mollissima). The allele frequencies of an egg white polymorphism showed significant differences between two populations occupying 1,700 acres in the Sands of Forvie National Nature Reserve.

Here ecological data revealed that the two populations differed in that one population was migratory and returned to breed on the east side of the moor, while the other was sedentary and bred on the west side. The data also showed that 95% of each population were mated before the return of the migrant population. Hence, a single dif­ ference in a behavioral trait can effectively cause differentiation between populations. 4

Coach and Beardmore (1959) suggest that a low frequency of mixed matings between two color phases of Chen caerulescens, Blue Goose and the Lesser Snow Goose, may be due to assortative mating and, there­ fore, leads to differentiation between the two color phases. O'Donald

(1960) also found assortative mating in the pale, intermediate, and dark phases of the Arctic Skua, Stercorius parasiticus. Hence, another behavior was found in natural populations, which caused popu­ lation structuring.

Therefore, in this thesis an application of one of the models for population structure will be applied to the data to look for differen­ tiation between localities. If structuring is found, then an attempt will be made to find the mechanisms leading to the structure of the populations. 5

II. Electrophoresis, Theory and Application

Electrophoresis is defined as the movement of a charged surface

(i.e., dissolved or suspended) relative to a stationary liquid by an electric field. Electrophoresis has burst into the foreground as one of the most significant biochemical analytical tools. Within the last two decades it has established itself as a very important tool in separate mixtures of proteins and other biological material. The phenomenon of electrophoresis is just one of the four electrokinetic phenomena known. The others are electroosmosis, streaming potential, and sedimentation potential; however, electrophoresis has the most potential for use as an analytical tool, and, therefore, has been utilized in many separatory techniques, which have as a basis the dif­ ferences in charge between colloidal particles in an electric field.

As early as 1899 Hardy discovered that many bicol1oids had char­ acteristic electrophoretic mobilities. This greatly stimulated interest in electrophoretic work leading to the characterization of proteins and enzymes by their electrophoretic properties, especially their isoelectric points. Michaelis (1909) was the forerunner in this field. He used a U-tube to determine the migratory direction of enzymes at different pH's, thus enabling him to establish isoe1ectric points for these enzymes. Although the above properties were impor­ tant, the diversity of electrophoretic behavior in biologically important substances allowed separatory electrophoresis to come to fore. This property of electrophoresis enabled scientists to study mixtures of biological material rather than isolated and purified material. 6

Two important techniques will be described here. The first being

Moving Boundary Electrophoresis, which is used as a historical build­

up to the second, Zone Electrophoresis.

Tise1ius in 1937 perfected the technique of Moving Boundary

Electrophoresis. This technique is able to separate, identify, and

estimate particles having different electrophoretic mobilities. It

can yield accurate electrophoretic mobility data. The main features

of the Tise1ius moving boundary technique are:

1. Sharp boundaries are initially formed.

2. Correctional boundary disturbances are minimized by an

effective thermostatic control.

3. Large electrode vessels contain reversible electrodes.

4. An optical system follows the movement of the boundaries,

due to a change in retractive index.

Figure 2.1 shows the basic set-up in a Tise1ius Apparatus.

Its widest application has been in the analysis of naturally occurring mixtures of proteins. Tise1ius first applied his apparatus to the detection of blood serum components. This apparatus is able to separate human serum into six major components: albumin, a1pha-1­ globulin, a1pha-2-g10bu1in, beta-globulin, fibrinogen, and, gamma globulin (Figure 2.2). This technique, although its resolution is not as great as in chromatography or zone electrophoresis, was able to show that these components were not homogeneous.

It was not until zone gel electrophoresis was developed that non­ homogeneous components could be further separated. Zone Electrophore­ sis is similar to the above technique except the migration is supported 7 by a homogeneous solid or gel framework. This minimizes convectional disturbances. Most of the early work was done with filter paper electrophoresis, but this technique gave similar resolutions as the

Tiselius method. The use of other supporting media (e.g., cellulose acetate, starch gel, and poly-acrylamide) greatly increased separa­ tions and resolution of the 6 major components found earlier.

HmTever, the use of a supporting medium introduces some complicat- ing variables: (Shaw, 1969)

1. capillary flow

2. electro-osmosis

3. adsorption

4. molecular sieving

The occurrence of electro-osmosis, which is the movement of liquid relative to a stationary charge surface by an applied electric field, compliments electrophoresis (Figure 2.3) and, thus, makes accurate mobility and isoelectric measurements difficult, since mobility could vary depending on the position of a sample in the blpporting medium.

The phenomenon of molecular sieving enables gel electrophoretic techniques to separate the inhomogeneous components.

There are some very important advantages to zone electrophoresis.

First, it can achieve complete separation of electrophoretically dif­ ferent components. Second, it is a less costly and a more simple technique. Third, it can be applied to simple ions as well as complex macromolecules. Last, smaller quantities of samples are needed, which may be an important necessity in some research. The most significant advantage is in the greatly increased separation and resolution. 8

Gel electrophoresis came into t~c limelight when Oliver Smithies in 1955 came across a new supporting medium, hydrolyzed starch. He recognized that the zone electrophoresis media used at that time gave a poorer resolving power than Tiselius' free boundary method. The paper eleccrophoresis method of Kunkel and Tiselius (1951), due to the absorption effects of the paper, reduced the resolving power of this technique, since the starting zone was not narrow enough. Therefore,

Smithies wanted a technique which would combine the ease of staining in the filter paper technique of Kunkel and Tiselius and the low absorption effects of the starch grain method of Kunkel and Slater

(1952).

This technique is called Starch Gel Electrophoresis. Smithies first prepared the gels from reagent grade soluble starch supplied by

Merck and Company, Ltd., Montreal, Quebec, Canada. The starch was later hydrolysed in a solution of acetone and HCl, washed and dried.

Concentrations of 109 of starch/lOOml. buffer to l6g/l00ml. buffer were used. After suspending his hydrolysed starch in buffer, he heated it until the starch grains were ruptured and a viscous homo­ geneous solution was obtained. This solution was later attached to a vacuum line to remove all air bubbles and then poured into a plastic tray. Introduction of the sample was made by either filter paper or starch grains soaked in serum.

The horizontal set up in Figure 2.4 is taken from Smithies 1955 paper. A regulated variable power supply was used with a maximum out­ put of 70 rna and 300 V. 9

He used a 0.03 mole of H B0 and O.lZ mole NaOH buffer. A 3 3 potential gradient of 6V/cm was applied for 6 hours at room tempera­

ture. The electrolyte used was 0.3 mole of H B0 and 0.06 mole of 3 3 NaOH. In a gel tray of the following dimensions, 4 to 10 mm in depth,

15 to 80 rom in width, and ZOO to 450 rom in length, a current of 4 to

5 ma for each gel was observed.

His results showed that most of the new bands were from the a ­ Z globulin fraction of the filter paper technique. He had one band moving in the opposite direction to the rest of the proteins. This came from the y-globulin fraction. There were two components he referred to as pre-albumins. There were, also, faint components migrating just in back of the albumin zone, which he called post­ albumins. The other components present were the following:

1. Fast az-globulin-migrating behind the post-albumin

S-globulin, the next slowest moving

Z. Slow az-globulin-migrating between S-globulin and the

starting position

3. as-proteins-found in two other groups of individuals,

migrating between slow a and S-globulin Z All the above except for the S-globulin component are derived from the az-globulin complex of paper and moving boundary electrophoresis.

Smithies (1955) was able to divide the serum into three distinct groups, I, IIA, and, lIB; with group I having no as proteins, and group IIA having as proteins moving faster than the lIB group. When serum was complexed with hemoglobin he found that the proteins of the three groups behaved differently. This difference Smithies observed 10 was due to the difference in haptoglobin genotypes. Previous to this, only Bernfield, Donahue and Homburger (1953) described "considerable dissimilarities" in human plasma from different individuals by using the Tiselius method. They found heterogeneity in the a globulin fraction, but could not make any definite generalizations due to their inability to resolve the as proteins.

This increased resolving power, Smithies attributed to the molecular sieving properties of starch gel which has pore sizes similar to those of the molecules under study. This conclusion was reached, since he observed that components in free moving boundary electrophoresis may have similar mobilities, but, in starch gel, their mobilities may be quite different and may even be reversed when com­ pared one to another. He also attributed the success of this technique to the high potential gradients which could be used with low ionic strength buffers. Another factor is the sharpening of bands due to the greater mobilities of the proteins in the slot compared to those in the gel.

Smithies correctly attributed this increased separation to the phenomenon of molecular sieving. For, in gel media such as starch, the spaces in the gel network approach macromolecular dimensions.

This sieving effect retards molecules larger than the pore sizes of the gel network. This sieving effect was explored by Singe and

Tiselius (1950) recognized the possibilities of this phenomenon, but explored it with no success. The mechanism involved in this sieving effect involves a particle moving through a gel experiencing frictional resistance, which is a complex function of r, the particle's radius. 11

When 2r is smaller than the average pore size of a gel, the viscosity

of a gel is low. However, when 2r is larger than the pore size, then

the viscosity becomes infinite. If enough momentum can be transferred

to displace the lattice, then the particle can "squeeze" through the

pore with a certain frictional resistance (Ornstein, 1964).

Another factor involved in the increased resolution and separa-

tion in gel electrophoresis is the thinner starting zone formed in the

direction of the starting zone. For, the thinner the starting zone,

the higher will be the resol11tion, at least, until the point where

diffusion spreading of a zone's edges becomes large compared to the

starting dimensions. This thin starting zone is achieved by two means.

The first is by the lack of absorption of the gel and, the second, is

achieved ~y Kohlrausch's regulating function. He observed that if two

solutions of ions were layered one over the other, such that a solution with ions of high mobility electrophoretically were placed below a

solution of a slow ion of the same sign of charge, then the boundary between the two ionic species would be sharply maintained, as migra-

tion occurs in an applied electric field. In gels, sharp moving boundaries will be maintained irregardless of the densities of the starting solutions, providing the faster ions precede the slower.

This principle is mathematically formulated in the following equation

(Ornstein, 1964):

(A) = X Ca = maZy (my - mS) -y- (f) xacy mrZa (rna - mS) .

Where X is the fraction of dissociation, C is the concentration, m is 12 the mobility and a, y, S are the ions with S, of opposite charge to a and y, and common to both solutions.

Ornstein (1964) suggested that if a protein molecule with a mobility of -1.0 units is placed in a glycine solution (pH 8.3) one centimeter above the glycine-chloride boundary, by the time a glycine molecule has migrated one centimeter, the protein will have migrated two centimeters and, thus, would be located at the boundary and will continue to run at the boundary, for the chloride ion (high mobility) migrates faster than the protein. If a mixture of proteins is used, ranging in mobilities from -1.0 to -6.0 units and is placed in an electric field, this function will concent~2te all proteins into very thin zones one behind the other, in order O'~ decreasing mobility. As the proteins enter the gel, the resistaace retards the proteins and, therefore, the glycinate ion (low mobility) overruns all zones and places the zones in a uniform linear volLage.

Smithies (1955) used a continuous buffer system, one in which the gel buffer and electrolyte are identical. This continuous system gave a rather broad and diffused albumin band. Poulik (1957) described a discontinuous system which greatly increased the resolution and sharp­ ened the albumin band. Poulik did not know then what caused this in­ creased resolution. Ornstein (1964) attributed this phenomenon of the discontinuous system to the Koh1rausch regulating function.

Smithies' starch gel electrophoresis so greatly increased the separation of human sera when compared to paper electrophoresis that

20 bands of proteins can be resolved as compared to 5 in the paper system. This increased resolution is not only attributable to 13 molecular sieving but also to inhibition of diffusive zone broadening by the strong gel matrix (Shaw, 1969). Smithies' technique also en­ ables one to examine proteins with a minimum risk of denaturing them.

Hunter and Burstone (1958) combined this technique with their zymograms, enzyme-substrate, visual staining technique, which allows one to lo­ calize individual enzymes on a gel. This greatly increased the use of gel electrophoresis as an analytical tool.

However, starch gel had several inherent problems. The major problem was that of variability from lot to lot and difficulty in preparation. Raymond (1959) subsequently suggested acry1amide as a supporting medium. The advantages of acry1amide over starch are the following: (Shaw, 1969)

1. Electro-osmosis is negligible.

2. Molecular sieving can be controlled over a much

wider range to achieve optimum separation.

3. Gel structures can be reproduced.

4. Adsorption, even of positively charged molecules

is negligible.

5. Separation is achieved in a shorter time.

6. A wider range of buffer solutions can be used.

7. The gel is transparent.

8. The gel is easier to handle.

Ornstein (1964) and Davis (1964) used this medium on their disc elec­ trophoresis technique which increased the resolution found in gel slab techniques. However, this will not be described, since it was not used as an analytical tool in this research. 14

III. The Pigeon

3.1. History

The pigeon was one of the first animals to be domesticated and

held in reverence. As far back as 5000-2000 B.C., B. C. Langdon

(1931) tells of the revered position held by doves by the Accadian,

Canaanitish, Armean and Arabic races. Darwin (1859) states that the earliest known record of pigeons is in the Fifth Egyptian Dynasty

about 3000 B.C. Lyell (1887) suggests that they probably were present

in the previous dynasty. The homing ability of the pigeon was first

put to use when the pigeon was used as a military courier for nations

at war. King Solomon, the Judean king was thought to have used this ability in about 1000 B.C .. Chamberlain (1907) states that the early

Persians, Assyrians, Egyptians, Lydians, and Phoenicians must have used pigeons in their military and naval campaigns. King Cyrus the

Great, of Persia (529 B.C.) is said to have used pigeons as message

carriers throughout his kingdom (Levi, 1969). Past records indicate

that Eastern countries were the cradle of pigeon domestication and

Western countries imported these breeds from the East. Levi (1969)

states that there are no pigeons indigenous to the Western countries; most breeds in these countries were derived from importations from

Persia, India, and Asia Minor.

3.2. Origin

Darwin in 1859 stated his theory that all domestic pigeons are derived from just one species. He stated in 1897 that 15

"The rock pigeon, Columba livia, may be confidently viewed

as the common parent form."

However, Whitman in 1919 believed; along with Ghigi (1908), that the

ancestors of the domestic pigeons may be polyphyletic. Whitman (1919) performed a simple experiment in which he selected two populations,

the barred and the chequered. In the barred he tried to select for chequered pattern, but with no success. However, starting from the chequered and selecting for less cheques, he could get blue barred patterns and even barless. This led him to erroneously hypothesize

that the chequered pattern was ancestral, since chequered could not be gotten from barred. He failed to recognize the phenomena of dom­ inance and recessivity on these characteristics.

Cole, in 1939, a geneticist from the University of Wisconsin, supported by his knowledge of the genetics of different characters, and the immunogenetic studies of blood by himself and Irwin, concluded that the origin of the domestic pigeon was monophyletic, with muta­ tions being the primary raw materials for the different breeds ob­ served. Here they found that most red cell antigens in the domestic breeds were identical to those in Columba livia, therefore suggesting a monophyletic origin, and not polyphyletic as suggested by Whitman

(1919). 16

3.3. Taxonomy: (Levi, 1969)

Class Aves Birds

Sub-class Neornithes True Birds

Super-order Neognathae Typical Birds

Order Columbiformes Sandgrouse and Pigeons

Sub-order Columbae Pigeons, Dods, and Solitaries

Family Columbidae Pigeons, Doves

Genus Columba

Species Livia

Breed Domestica

Domestic Pigeon

There are two major characteristics of pigeons and doves which distinguish them from other birds. The first is the production of crop milk. The male pigeon is the only male vertebrate which produces milk for its young. The female pigeon is the only feathered female that produces this milk. This, along with the manner in which they drink, also distinguishes them from all other Aves. Pigeons and doves immerse their beaks into the water up to their nostrils and drink in one continuous draught. Other birds dip their bills into the water and have to elevate their heads to let the water run down their throats.

3.4. Previous Genetic Analysis

The procedure for analyzing morphological and physiological char­ acters has been to relate every character to the typical phenotype of 17 the wild rock pigeon, blue-barred, plain headed, and, clean legged.

Most of the characters studied were morphological, due to the many phenotypes observed in domestic pigeons. Loisel (1903-1905) was the first to do a study of inheritance in pigeon color and patterns. His works suggested that red color and chequer pattern were dominant and not sex-linked. The following is a list of characters looked at and their references:

Color Inheritance

Intense Their Dilutes Non-Pigmented

Blue Silver White

Black Dun

Brown Khaki and Drab

Red Yellow

(Table taken from Levi, 1969)

Silver - Recessive to blue - Bonhote and Smalley (1911)

Sex-linked - Staples - Browne (1908) and Cole (1912, 1914)

Black - Dominant to blue - Hollander (1938a)

Linked to opal and other pattern factors - Hollander (1938a,

1938b)

Dun - Dilute of black - Cole (1914)

Sex-linked - Cole (1914)

Ash Red - Dominant to blue - Cole (1914), Cole and Kelley (1919)

Christie and Wriedt (1923); Metzelaar (1924, 1926, 1928),

Steele (1931), Horlacher (1930), Hawkins (1931), and

Hollander (1937).

Sex-linked - Cole (1914), Cole and Kelley (1919). 18

Red (recessive) - Recessive to blue - Cole (1914).

Yellow - Dilute to Ash Red and Recessive Red - Bonhote (1911)

Sex-linked - Cole and Kelley (1919).

Brown - Recessive to blue and allelic to Ash Red - Metzelaar (1926,

1928), Steele (1928, 1931), and Hawkins (1931)

Sex-linked - Christie and Wriedt (1927).

Khaki and Drab - Dilute of Brown - Metze1aar (1926, 1928).

White - Dominant or recessive to Blue - Cole (1939).

Albinism - recessive to Blue - Levi (1969).

White epistatic to many colors - Levi (1969).

Grizzle - Dominant to Blue - Bonhote and Smalley (1911), Bol (1926),

and Hollander (1937). o Opal - Recessive to Blue - Hollander (1938a).

OpaloP - Dominant to Blue - Hollander (1938a).

Smoky - Recessive to Blue - Hollander (1938a).

Milky - Recessive to Blue - Hollander (1938a).

Almond - Dominant to Blue - Hollander (1944).

Faded - Dominant to Blue; Allelic to Almond - Hollander (1938a).

Sex-linked - Hollander and Cole (1940).

Sandy - Dominant to Blue - sex-linked - Levi (1969).

Reduced - Recessive to Blue - sex-linked - Graefe (1951). 19

Pattern Inheritance CT T-pattern - dominant to Bar - Hollander (1938a), Jones (1922)

Bol (1926), Bonhote and Smalley (1911) C Checker-pattern - dominant to Bar

Bar-pattern+ - wild type c Barless-pattern - recessive to Bar

Inheritance of Eye Color

Phenotypes: Bull, Pearl, Orange and Red, Yellow, Albino

Pearl - Recessive to Orange and Red - Hollander and Owen (1939),

Bond (1919)

Bull - Recessive (not simple) to Orange - Levi (1969),

Recessive to Pearl - Christie and Wriedt (1923).

Inheritance of Crest and Hood

Crest - Recessive to plain head - Staples-Browne (1905), Morgan

(1911), Christie and Wriedt (1923), Soderberg (1927),

Bessmertnaja (1928), Horlacher (1930), Vecchi (1935)

Hood - Recessive to wild-type - Christie and Wriedt (1927).

Inheritance of Silky (Lace), Scraggly, Porcupine

Silky - recessive to normal feathering - Cole and Hollander (1939)

Scraggly - recessive to normal feathering - Riddle and Hollander (1943)

Porcupine - recessive to normal feathering - Cole and Hawkins (1930). 20

Inheritance of Feathered Leg

Muffed - co-dominant to clean legged - Doncaster (1912)

Grouse - recessive to clean legged - Hollander (1937)

Slippered - co-dominant to clean legged - Hollander (1937).

Inheritance of Foot Abnormalities

Web-foot (Syndactylism) - recessive to normal - Staples-Browne (1905)

Polydactyly - recessive to normal - Hollander (1954).

Inheritance of Other Abnormalities

Rump1ess - recessive to normal - Levi and Hollander (1939)

Eyeless - recessive to normal - Hollander (1948)

Featherless - recessive to normal - Cole and Owen (1944)

Wingless - recessive to normal - Levi (1969).

Inheritance of Physiological Characters

Ataxia - recessive or partially dominant - Riddle and Hollander (1943)

Clumsy - recessive to normal - Hollander (l938c). 21

IV. Short Descriptions of Localities Trapped

4.1. The Honolulu Zoo

The major trapping area was the Honolulu Zoo. It is located on

the Kapiolani Park grounds on the corner of Kapahulu Avenue and

Kalakaua Avenue. The zoo is bordered on its mauka (North) side by

Paki Avenue, on its ewa (West) side by Kapahulu Avenue, on its makai

(South) side by Kalakaua Avenue, and, on its Diamond Head (East) side

by Monsarrat Avenue.

The zoo covers a total area of 40 acres. The first zoo was built

in 1914. It was later expanded in 1935-1940, at which time the zoo

started its collection of tropical birds. Newspaper records show that pigeons were flying free prior to 1934. According to Dr. K. I. Chun,

one of the oldest pigeon fanciers in Hawaii, the zoo population was

started originally from four pairs of White Kings. Where these pigeons were obtained is still a mystery.

The major breeding areas for these pigeons at present are the date and coconut trees throughout the zoo and adjacent areas, in addition to the eight barns in the hoofed-animal area. Prior to this

time, the zoo had a large breeding 10ft to which the pigeons had free access, but this was torn down in the 1950s. The boundaries of this population are the ocean on the south side, Diamond Head on the east side, Ala Wai golf course on the north side, and Ala Moana Boulevard on the west side.

Seven trapping areas within the zoo were chosen mainly because of accessibility (Figure 4.1.1.). Areas 2 and 3 were the major trapping 22

areas since these were the feeding areas for the pigeons and also the

largest open areas.

A total of 792 samples was obtained from this locality. The

total population size was estimated to be 3,368 ± 15.2 as calculated

from capture-recapture data.

4.2. St. Andrew's Belfry

St. Andrew's Cathedral is located on the corner of South Beretania

Street and Alakea Street. The belfry was built in 1912. It is an

ideal breeding niche for pigeons since there are windows (1 ft. x

3 ft.) every eight feet on each side of the tower. The birds have un­

hindered entry into the tower. Pigeon nests are built on the steps

adjacent to each window. There are three major levels to the belfry.

The first level contains the machinery for the pipe organ. The second

level is an empty storage room with four windows facing east. Nests were found on each window ledge here and also on the floor adjacent to

the windows. The third level contains the cathedral bell. This room

has a ceiling of approximately 75-100 feet. There are three levels of windows in this room with six windows on each level. Nests were found

on every window ledge. At the very top where the roof meets the high walls there is sufficient space for the birds to build nests and to

roost.

Approximately sixty nests were counted from the ground level to

the roof of the belfry. The total population size was estimated by direct counting to be about 150-175 birds. The belfry's steps and rooms are cleaned once a year. However, nests on the higher window levels and on the rafters are left untouched. It takes only a week 23 for the pigeons to return and rebuild their nests. A total of 73 birds was removed from this locality.

4.3. Bethel-Pauahi Building

This building is located on the corner of Bethel and Pauahi

Streets. It is a seven-story building, which was built in 1962.

Between each floor there is a ledge 3 feet wide, which is an ideal area for pigeons to build their nests. An average of about 8 nests was found on each ledge. The second floor ledge did not have any nests but every ledge above the second had nests. Usually the nests were found behind the metal decorative grillwork fronting the building on Bethel Street. This acts as a wind and rain protector.

The population here was estimated by direct count to be approx­ imately 200 birds, and 76 were taken from this area. Since permis­ sion to trap birds there was granted on the condition that all birds caught would be disposed of, no birds could be banded for later collection of data.

4.4. Waldron's Grainery

Waldron's Grainery is a commercial animal feed distributing company. It is located in the lwilei district half-a-mile from the

Aloha Tower. Its boundaries are Nimitz Highway on its north and south sides, Kaukahi Street on its east side and Pacific Street on its west side. There are no nesting sites here. It is one of the major feeding areas for the birds in the city of Honolulu. The birds usually feed on the roof where there is a great spillage of feed being transported from the conveyor belt to the feed dispensing tower. They also feed on the 24 tower and conveyor belt. The pigeons feed here after sun down or just before dusk. Approximately 300 to 400 birds have been seen here on certain occasions. Trapping here was extremely difficult due to the over-abundance of food and the prohibition of trapping on the roof of the building. 25

4.5. Piers 27 and 32

Piers 27 and 32 will be described together since they are very

similar to each other and are separated only by a quarter of a mile.

Both piers are within the Honolulu Harbor complex. They are left open

24 hours a day since there are no doors on the water side of the piers.

Each pier is 300 feet long and 150 feet wide. The ceiling height is about 50 feet. Nests in pier 27 are found between the steel girders

supporting the roof and the corrugated roofing tnaking up the ceiling.

The main nesting area for pier 27 was found at the east end of the

pier. In pier 32, the nests were predominately found on the south side of the pier facing the water. The nests were found in spaces between

steel girders supporting the front of the pier and the corrugated

sidings of the top frontispiece. Easy access to these spaces is obtained through large holes in the frontispiece.

The population size was estimated to be 100 for pier 27 and 125

for pier 32. Pier 27 was built in 1938 and pier 32 in 1947. See

Figure 4.5.1. for their locations in the harbor area. 26

v. Trapping Procedures and Blood Collection and Storage

Sampling was done during the years of 1969 through 1971 on a bi­

weekly schedule. Live four-cell Potter trip-traps and 2 large 5' x 5'

funnel-cone traps were used. Trapping from September, 1969 to

September, 1970 was done bi-weekly during the hours from 9 a.m. to

5 p.m., Monday through Saturday. This time table was later changed by

order of the Zoo Director from 5 p.m. to 7 p.m. and from 6:30 a.m. to

9 a.m•• This timetable was followed from September, 1970 through

September, 1971. In December, 1971 one large mass trapping was done

by use of the neurodepressant Avertin.

The anesthetic was introduced into the feed in the following manner:

1. One gallon of pigeon feed was measured.

2. One cup of Avertin was mixed with two cups of 200-proof

grain alcohol. The alcohol acted as a dissolving agent.

3. This Avertin mixture was then poured over the grain and

a tight lid was placed over it to prevent evaporation.

The grain was soaked for 48 hours with periodic mixing.

4. The liquid was then poured off and saved for future use.

The grain was then allowed to dry thereby evaporating the

alcohol and leaving the anesthetic on the grain.

After the birds consumed the feed, it took from 10 to 15 minutes for the anesthetic to act causing the birds to fall asleep and remain so from 1 to 3 hours. 27

In April and June of 1971, 136 birds were trapped at the Honolulu

Zoo by use of a fish net. This net was 12 feet in diameter and ringed with heavy guage wire to give it a stable shape. The net was hung above the ground through use of a tree, rope and pulley. Feed was used to entice pigeons into the trapping area. When the area was sufficient­ ly filled with pigeons the net was dropped. The trapped birds were then removed from under the net and taken to the laboratory for bleed­ ing and banding. Following this they were returned to the zoo and released.

In the years 1969-1971 all pigeons were banded. The color of plumage, color of eyes, age, sex, and type of shanks were recorded.

The birds were then released back into the population. The only group not released but kept for breeding studies was one trapped by Avertin drugging.

The Potter trip-traps were pre-baited and the trip doors were set.

The birds were then lured to the trapping area. Once a bird entered and stepped on the platform, this would release the trip lever and the door would drop. The funnel traps were pre-baited and allowed to sit in major feeding areas. The birds followed a trail of feed into the trap and were unable to find the opening again.

The best results for trapping were obtained always after a large storm. This weather condition kept the public away from the zoo and, therefore, the birds went unfed for one or two days. These major storms usually occurred during the fall and winter months. This accounts for the disproportionate amount of birds trapped during these rainy months. 28

In June of 1972 the drug Avertin was used to capture all the

birds from Saint Andrew's bell tower and from Bethel Street. All

samples from Piers 27 and 32 were obtained by shooting the birds

from the rafters with a .177 caliber pellet air rifle with a scope.

The birds were then etherized and bled from the heart with a 3cc

syringe and a 12 guage needle. The birds were then autopsied and sex

and breeding condition were recorded. Birds from St. Andrew's and

Bethel Street were brought back to the laboratory and etherized. An

autopsy was then performed, and, as in the Pier samples, data on sex

and breeding condition were recorded and a blood sample was then taken.

5.1. Blood Collection and Storage

Venous blood was collected through the alar vein found under the

pigeon's wing. The feathers were plucked from the area adjacent to where the wing meets the side of the breast. Alcohol was then swabbed

over the exposed skin area making the vein clearly visible. A lee.

syringe with a 25 guage needle was used to draw the blood. After the

blood was drawn, pressure was placed on the vein until blood ceased to

exude from the needle hole. No ill side-effects were observed, except

for occasional hematomas formed by double piercing of the vein.

All blood was collected in a 0.2 mI. ACD solution containing:

2.4 grams Dextrose/IOO mI.

0.08 grams Citric Acid/IOO mI.

2.2 grams Trisodium Citrate/IOO mI.

Storage in ACD at 4°C usually allowed survival of red cells for

21 days without any drastic changes (Bartlett and Shafer, 1961). 29

Storage in ACD usually was for 24 hours; however, on occasions, they were stored for 48 hours. No difference in enzyme patterns were seen between the 24-hour and the 48-hour stored samples.

The blood was then centrifuged at 3,000 rpm for 15 minutes after which the plasma was pipetted off and placed in 1/2 dram vials. The cells were then washed three times in ACD solution at which time the cells were lysed in a .05 mg solution of TPN. Two parts of TPN solu­ tion was added to one part red blood cells. The lysed cells were then mixed with .2 mls. of carbon tetrachloride or toluene and spun down again at 10,000 rpm. The layer of carbon tetrachloride or toluene plus the buffy layer was pipetted off. The remaining lysed cell solu­ tion was then pipetted into 1/2 dram, screw-cap vials and stored at

-12°C in an upright freezer. This type of freezer goes through an automatic defrost cycle every 30 days. This led to difficulties since the samples would thaw and the temperature within the freezer reached room temperature before the freeze cycle could cool the freezer again.

Many samples could not be typed due to deterioration of the protein patterns by this continual thawing and freezing. This, plus the many occasions on which electrical power was lost, invalidated many samples kept over too long a period of time. 30

VI. Statistical Procedures

6.1. Introduction

This section will deal with the statistical methods used to analyze the data presented in the following sections. Since the esti- mation of allele frequencies was determined by maximum likelihood methods, a general description of this method, derived by Fisher

(1922) should be presented.

If X is a random variable whose distribution is defined solely by q, and the sample size, n, is drawn at random from a population of XIS, then the likelihood, L, of obtaining a given sample will be

n IT f (x..q) i=l J.J

(Neel and Schull, 1954)

The theory of maximum likelihood states that the best estimate of q will then be the real positive root, if one exists, of the solution of

dL -= 0 dq

However, since L must be positive or non-negative, the maximum like- lihood estimate of q is then obtained from the solution of

d (log L) = 0 dq

The variance of this estimate will then be the negative inverse of the second derivative of the likelihood function. 31

2 0 2 = -E ["""d_...... ;l-:-'O~g~L) -1 d q 2

The total amount of information with respect to q, Iqq, is then

defined as

2 log L] Iqq = -E [d dq2

6.2. Estimation of Gene Frequencies and a, Inbreeding Coefficient

6.2.1. Two autosomal alleles with no dominance (Yasuda, 1966)

Let a, band c be the observed numbers of genotypes AA, AB, and

BB respectively and let p equal the frequency of A; then, assuming the individuals are unrelated and from a panmictic population except for a

proportion a and that the genotypes were sampled before differential

selection has taken place, the genotype probabilities will be

P (AA) = p2 + p(l-p)a

2P (Aa) = 2p (l-p) (I-a) P (aa) = (1_p)2 + p(l-p)a.

The log of the likelihood equation will then be

L = (a+b)lnp + (b+c)ln (l-p) + aln(p+a-ap) + bln(l-a) + cln(l-p+ap).

The maximum likelihood scores resulting from taking the first deriva- tive of L with respect to the various parameters are 32

Up (a+b) [lip] + a [ ~-a ] 1-a ] = p -ap - (b+c) [1/1-p] - b [ 1-p+ap

The solutions to the above will then be p = (2a+b)/2(a+b+c)

a = (4ac-b2 )/(2a+b)(2c+b)

The variances of the scores will be

Kpp [11 + 1-a 2 + b [ 1-2p ] 2 + [1 + 1-a ] 2 a p p+a-apJ lP(l-p)] C[l_p 1-p+ap]

1-a ] [ 1-P ] -b 1-2p J-c--+·1 1-a ] Kpa = a [ lip + p+a-ap p+a-ap [p(l-p) (l-a) t1-p 1-p+a.p

This will give estimates of p similar to those obtained if a was not estimated simultaneously.

6.2.2. 3 autosomal alleles with dominance phenogram, 3-3-6

(Cotterman, 1954)

The solutions of the maximum likelihood equations are: 33

p* = 1 - Jb:C

q* = Jb:C -"fc/n

r* = J~/n

where a, b, and c are the observed numbers in the three phenotypic 2 3 l 2 2 2 l l classes; D3D3,D3D , or D d ;D D or D d ; and dld respectively.

N is equal to a + b + c. The information functions are

4n 4n(1-p) 4n I = ---"'''':''::':;""""7""" I = --::-::-~-:-"--.<.....:--:;- I P p(2-p) q q[2-q(l-p)] r = l-r2

No test of goodness of fit is possible, since there are no degrees of

freedom left; ~ - m = 0, ~ is the number of phenotypes and m is the

number of alleles.

6.3. Two Allelic Sex-linked Genes with Dominance in the Homogametic

Sex; Data on Both Sexes (Cotterman, 1954)

Let a and d be the observed numbers of the A_ and aa phenotypes

respectively, and e and f the observed numbers of Aw and aw phenotypes

in the heterogametic sex. Then the maximum likelihood estimates will

be

p* l-q*

-e + le2+4(2d+f) (2n+n') q* = 4n + 2n'

4n n' I = l-q2 + pq where n = the total number in the homogametic sex and n' is the total number in the heterogemet.ic sex. 34

6.4. Estimations of F, the Inbreeding Coefficient

6.4.1. Chi-square assuming panmixia (Li and Horvitz, 1953)

The derivation of f from the X2 formula is as follows:

.[2NFq .q . ] 2 1. 1. + I 2Nq.q. i

= NF 2 (k-2+1) = NF 2 (k-l)

Therefore, if qi = ni/N and the zygotic proportions are calculated under the assumption of panmixia the X2 calculated will give an esti- mate of F; viz.,

2 f2 = N(k-1),x where N = sample size and k = number of alleles which is equal to

2 f2 1f for 2 degrees of freedom.

All estimates of f were determined in this way. S

6.5. Test for Homogeneity of Gene Frequencies, Two Co-dominant

Alleles (Neel and Schull, 1954)

If n equals the sample size; X, the number of A genes; and p equals X/2N, then, a X2 value can be computed according to the following equation:

x2 = l:pX - pl:X l:X where p = l:2N and q = l-p pq 35

The degrees of freedom will be k-l, where k equals the number of localities.

6.6. Tests of Independence (Interaction, Homogeneity) Between Samples

An R x C Contingency Chi-square (Snedecor and Cochran, 1967)

If each sample can be classified by one characteristic into R classes, and into C classes by another characteristic, where R is row, and C is column, the data may be represented in a matrix con- taining R rows and C columns. The entries in each RC cell is the numbers of the sample classified as such. Therefore, the numbers expected for each cell may be computed by

F = (row total) (column total) n where n is the sum of the row totals and the row total and column total refer to the row and column the cell is in.

Therefore, a test for independence is a X2 value computed by x2 = E(f-F)2/F, where f is the observed number in that cell. The degrees of freedom will be (R-l)(C-l). Since all but one of the f. IS ~ can be independent, for one can put R-l expected frequencies in a column, but the remaining cell is fixed since the sum of the Fls must equal the column total and similarly for the explanation for (C-l), the degrees of freedom are (R-l) (C-l).

6.7. Heterogeneity X2 Test for Testing Homogeneity of an Underlying

Genetic Hypothesis in Different Samples

Given a sample which can be classified into m classes with expected frequencies, Pl,PZ, •.. ,Pm' respectively, a test to see if 36

multiple samples from a population does not differ in their p 's can m be derived in the following way. First, a computation of the

individua1x2's are made

m X2 = L i=l

Then the sum of these X2 's are taken with n-1 degrees of freedom,

where n equals the number of samples. The pooled X2 is then computed

X~ ~(I~(Ti = - NPil-1/2)2/NPi for 1 d.f., where Ti = totals for the individual classes. Then, subtracting X2_X~ will give us the

heterogeneity X2 with n-2 degrees of freedom which tests the null

hypothesis that the p 's are not significantly different from sample m to sample. However, if the p 's change in a linear manner and to the m same extent, they can depart from the ratio in the null hypothesis without increasing the heterogeneity X2; therefore, this test failed

to show any heterogeneity between localities due to the presence of

this linear trend. Hence, a more appropriate test of homogeneity would be an RXC contingency X2 test (Steele and Torrie, 1960) with

the latter test, testing the null hypothesis that the proportion of

individuals is the same in the various classes.

6.8. Simple Linear Regression Analysis (Snedecor and Cochran, 1967)

Simple linear regression has three assumptions about the relation

between Y and X: 37

1. For each X there is a normal distribution of Y, from which

the sample Y is drawn at random.

2. "The population values of Y have a mean value ]J that lies

on a straight line ]J = a+f3(X-X) = a+f3x."

3 . a+f3x is the population deviation about its mean.

The mathematical model is defined by

Y = a + f3x + e: where a is the mean of the population that corresponds to X=O, thus specifying the height of the line. f3 is the slope of the population rerr,.ssion line, or the change in Y per unit increase of X. e: is a random term drawn from a population with mean 0, and standard devia- tion a E is also independent of X. Therefore, the sample y·x· regression equation is

Y = Y + bx.

The null hypothesis to be tested is that b=O. This test is made by computing s ,which is the sample standard deviation from regres- y.x. sion, from the sum of squares of deviations

h(y_y)2/n - 2 where n = the number of classes.

From this statistic another statistic, the standard deviation of the regression coefficient, can be obtained,

s = s /1r.x 2 b y·x.

Therefore, a test of significance of b is given by 38

d.f. = n-Z.

6.9. Haldane's Test for Sex-linkage in Samples from Populations

(Haldane, 1963)

A test for sex-linkage in samples from populations was derived by Haldane. His formulations were followed exactly, except for the need to switch males for females and vice-versa, since males are homogametic. The hypothesis to be tested is that s(b) = mq and

S(d) = fq2, where m is the number of females, f is the number of males, and band d are the number of recessive phenotypes in females and males, respectively. The d value usually can be tested for agree- ment with the following equation:

b:~ ± :2 laf(rn+b) where a is the number of dominant

phenotypes in females or by

6 = b(b-1) f-m(m-1)d

(b-1)[ (bc-Zad-(a+1)] if d is not small. The s(6)=O and a significant positive value imply an excess of recessive females, and a high negative value, implying a deficiency of recessive females. To obtain the sampling variance of

6, Haldane first computed the maximum likelihood estimation of q which was

1 A = [4(Zf+m)(b+Zd) + a2]~-a q Z(Zf+m) 39

Then, using this estimate in the following equation will give you an estimat~ of the standard error of ~

~ a~ = mqLfm(l-q){m + (4m+f)q})2.

Hence, dividing ~ by the a~ will give a test of the hypothesis stated at the beginning of the section.

There is a restriction in using this test. That is, the members of the sample must be unrelated, for panmixia is assumed in this formulation. Deviations from expectations can be caused by inbreed- ing, genetic heterogeneity, or selective mortality.

6.10. Woolf's Analysis

This analysis was originally derived for use in determining significant associations between blood groups and diseases (Woolf,

1955). This analysis is easily applied to data which can be formed into 2x2 comparisons, such as comparing homozygote and heterozygote frequencies of two localities.

The first estimate to be made is that of X, which is the ratio of one incidence rate to another. Therefore an estimate of X is

X = hK/Hk where h is the homozygote frequency in one class and k the hetero- zygote frequency in the same class; and H is the homozygote frequency in the other class, and K its heterozygote frequency.

This X value is then transformed into its logarithm which avoids aSYIDIDetry, since the comparison of a with band b with a will have the same numerical value for X, with only the sign changed. In 40 addition, the variance of y, log X, will be a simple formulation e

v = l/h + l/k + l/H + l/K w = l/V where w is the weight of y. ~(y) = 0 is the null hypothesis to be tested. This test will then be y2/V or wy2 for one degree of freedom.

If more than one sample is taken or a test of the homogeneity of the y estimates is needed, the following statistics need to be esti- mated: Y, which is the mean estimate of y is estimated by Ewy/Ew, and its antilogarithm, X, is taken as a combined estimate of X. The significance of Y can be tested by X2 = (Ewy)2/Ew = y 2Ew with one degree of freedom. Therefore, the heterogeneity X2 is tested by x2 = Ewy 2 - (Ewy)2/Ew = Ewy 2 - y 2Ew with degrees of freedom equal to n-l, n being the number of sets of data.

Therefore, this analysis is a valid test to test for differentia- tion between the proportions of homozygotes and heterozygotes between different groups of data or classes. Another appealing characteristic of this analysis is that the estimates are weighted estimates.

6.11. Summary

Maximum likelihood scoring methods of Yasuda (1966) are applied to the data to estimate simultaneously the gene frequencies and a.

Maximum likelihood estimation procedures are also applied for cases with 3 alleles in a dominance hierarchy and two sex-linked alleles with dominance. Contingency X2 tests for homogeneity is applied to the data to test that the proportion of individuals are the same in 41 the respective classes. Failure of the heterogeneity X2 to detect heterogeneity of p was discussed. A simple linear regression model is applied to the data to test for linear changes of gene frequency with distance. Haldane's test is applied to test for deviation of the observed phenotype frequencies from expected phenotype frequencies determined by sex-linked alleles. Finally, Woolf's analysis is applied to the data to test for the homogeneity of homozygote and heterozygote frequencies between two classes. 42

VII. Analysis of the Transferrin Locus·

7.1. Introduction

Transferrin is a B-globulin whose function is to prevent the loss of iron from the body by binding to the iron released by hemo­ globin. Schade et al., (1949) found that two atoms of ferric iron are ionically bound to the transferrin molecule. It is stable at pH values ranging from 7.2 to 6.5, with a pH of 5 causing ~omplete dis­ sociation. However, this reaction is reversible by raising the pH to 7.2. Aasa et al. (1963) suggest that the chelating site of the

Tf molecule involves two imidazole groups plus three tyrosyl residues.

Komatsu and Feeny (1967) put forth evidence for the essential role of the tyrosyl residues. They observed that there was a loss of iron­ binding capability following treatment of human transferrin with

N-acetylimidazole and this could be reversed following deacylation with hydroxylamine. Yoshioka et al. (1966) were able to demonstrate only one Fe-Tf complex in acrylamide gel, that of the Fe Tf complex. 2 Many other metallic ions (Jones and Perkins, 1965) plus certain salts such as phosphate and citrate (Schade et al., 1954) can bind to the Tf molecule in vitro. However, Saltman and Charley (1960) and

Davis et al. (1962) reported extremely high stability constants of the Tf binding sites for iron, thus making it unlikely that it is used as a physiological transport of other metallic ions and competition would unlikely occur in vivo.

In the fowl, unlike the human Tf molecule described above, there are two ferric iron atoms bound by tyrosine hydroxyl groups and histidine nitrogen atoms (Windel et al., 1963, Aasa and Aisen, 1968). 43

Katz and Jandl (1964) found that the iron is rapidly transferred to the erythroid cells in the bone marrow and to the yolk (Williams,

1962). Katz and Jandl (1964) also showed that the half-life of apotransferrin in fowl is 16 hours compared to 1.32 hours in man.

They suggested that the Tf molecule adheres to the surface of the de­ veloping red cell for about 1 minute during which time the iron is removed and incorporated into the cell. The apotransferrin is then unlodged and is able to return to the plasma to repeat the cycle

(Katz and Jandl, 1964).

Another function hypothesized by Schade and Caroline (1946) and

Martin and Jandl (1959) is that of a bacteriostatic function, by rendering iron unavailable for the multiplication of bacteria and viruses, thus forming a secondary defense against infection. This function was observed in ova transferrin in egg white.

The synthesis of transferrin occurs both in the liver and yolk sac in the chick embryo after 16-18 days incubation (Gitlin and

Kitzes, 1967). Mc Donald and Riddle (1945) found that estrogen administered at high levels increases the proteins of blood and liver and causes hypertrophy of the latter in the chicken and pigeon. These effects were found to be counteracted by thyroxine administration, thus identifying the organ for synthesis of most plasma proteins in fowl and pigeons.

Transferrin has been found to be one of the most polymorphic of the plasma proteins. It exists in a polymorphic state in the follow­ ing organisms: (Lush, 1967) 44

Mus musculus - mouse - two variants (Shreffler, 1960)

Equus caba11us horse - six variants (Schmidt, 1962), (Braend,

1964b), and (Braend and Stormont, 1964)

Bos taurus - cattle - six variants (Hickman and Smithies, 1957;

Bos indicus Ashton, 1957; Ashton; 1959; Ashton, 1965;

Jamieson, 1965)

Ovis aries - sheep - six variants (Ashton, 1958; Khattab et a1.,

1964)

Capra hircus - goat - two variants (Ashton and McDougall, 1958)

Sus scrofa - pig - two alleles (Kristjansson, 1960a and 1960b;

Ashton, 1960; Kristjansson and Cipera, 1963)

Cervus e1aphus - red deer - two variants (Lowe and McDougall, 1961)

Rangifer tarandus - reindeer - eight genetic variants (Gahne and

Rende1, 1961; Braend, 1964a)

Macropus rufus - red kangaroo - two variants (Cooper and Sharman,

1964)

Macaca iris - cynomolgus monkey ­ eleven variants, seven of which

Macaca mu1atta - rhesus monkey - are in common with one another

(Goodman and Wolf, 1963;

Goodman et a1., 1965)

Pan troglodytes - chimpanzee - four variants (Boyer and Young, 1960)

Gallus gallus - domestic fowl - two variants (Ogden et a1., 1962)

Columba 1ivia - pigeon - two variants (Mueller, Smithies, and Irwin

1962)

Mueller et a1. in 1962 using vertical starch gel electrophoresis and the buffer system of Smithies and Hiller (1959) described three 45 transferrin phenotypes; TfLl, TfLl-L2, and TfL2 in pigeons. Their breeding data suggested two co-dominant autosomal alleles. Each allele consisted of three bands, two major and one minor, with the heterozygote having four bands (Figure 7.1.1.). In 1968 Jeffrey

Frelinger from the California Institute of Technology did breeding studies of pigeons to examine the maintenance of this transferrin polymorphism. His studies showed clearly that heterozygous hens hatched a substantially larger proportion of eggs than did homozygous hens. He then tested the bacteriostatic hypothesis of Schade and

Caroline referred to earlier and found that egg white from hetero­ zygous hens inhibited the growth of yeast more than those egg whites from homozygous hens. When excess iron was placed into the egg white no significant differences were seen. Since the young squabs exhibit only their mother's genotype, prior to two weeks of age, this would give a selective advantage to eggs laid by heterozygous hens.

7.2. Materials and Methods

The methods for collection of blood samples were described earlier in section 5.2; therefore, they will not be described here.

At the initiation of this study three buffer systems were employed to examine which would give the best resolution and most consistant patterns. The three employed were:

Poulik Discontinuous Buffer System (Poulik, 1957)

Gel Buffer: pH 8.65

0.076 M Tris-hydroxymethyl - amino methane

0.005 M Citric Acid 46

Electrolyte pH 8.65

0.3 M Boric Acid

0.05 Sodium Hydroxide

Kristjansson Pre-Albumin System (Kristjansson, 1963)

Gel Buffer: pH 7.5

0.004 M Citric Acid

0.14 M Tris-hydroxymethyl - aminomethane

Electrolyte pH 8.7

0.3 M Boric Acid

0.1 M Sodium Hydroxide

Ashton Buffer System (Ashton and Braden, 1961)

Gel Buffer - pH 8.1

.008 M- Tris-hydroxymethyl - aminomethane

.040 M- Citric Acid

450 mls. Gel Buffer + 50 mls. Electrolyte Buffer

Electrolyte Buffer - pH 7.9

.031 M Lithium hydroxide

.194 M Boric Acid

The size of the gel trays were 25.5 em. in length, 12.9 em. wide, and a gel depth of 3 mm. This tray was divided in half, giving two gels 6.4 em. in width. The inserts used were Whatman E 17 cut,S rom. by 3 mm. A constant voltage of 10-12 V per centimeter was applied across the gel for 4 hours. The initial milliamperage was usually

50 mAo per gel plate for the Ashton buffer system, which was the one chosen for all future transferrin typing, due to its better resolution and repeatability of patterns. 3.6 em. separated the wick from the 47 insert line and the distance from insert line to the positive pole wick was 16.8 em. The insert slot was cut horizontally across the gel with a razor blade. Fourteen to sixteen sample inserts could be placed into each half of the gel at one time. Sixty to sixty-five grams per 500 m1. buffer of starch was used for all systems depending on the batch of starch.

The gels were stained with a migrosine solution of the following chemical composition: (10-15 minutes)

15 gms. Nigrosine

500 m1s. Ethanol (95%)

500 mls. H 0 distilled 2 100 m1s. Glacial Acetic Acid (Reagent)

The gels were then washed three times every 10 minutes with a wash solution of 5:5:1 of Ethanol, H 0, and Glacial Acetic Acid. After 2 the final wash the gels could be read after 30 minutes had elapsed.

7.3. Results and Discussion

The analysis of the transferrin locus was done separately by locality, except in the case of Waldron's where the data were not sub­ divided due to the small numbers sampled. The results of this analysis will, therefore, be discussed by trapping localities and then sum­ marized in the end.

The first locality is the Honolulu Zoo. The heterogeneity X2 calculated to test for heterogeneity of samples taken from the seven trapping areas within the zoo was not significant (Table 7.3.1.).

Therefore, in future analyses the data were pooled over all areas. 48

With the pooled data, the X2 for goodness of fit (Table 7.3.1.) showed

a significant departure from the expected at the .001 level. A clear

deficiency of heterozygotes was observed. The discussion of this

result will be deferred until population structure and inbreeding are discussed.

The contingency X2 tests to test for independence of transferrin phenotype and other phenotypes will be discussed here. There were no significant evidences for interaction between this system and sex.

The contingency X2 with 2 degrees of freedom was 0.14 with a p value greater than .900 (Table 7.3.2.). The contingency X2 to test for interactions between the transferrin system and age again showed no significance (Table 7.3.3.). The X2 value was 3.95 with 2 degrees of freedom. Therefore, there seems to be no phenotype frequency dif­ ferences either in sex or in age. Sexing was done by phenotypic examination. By young it is meant that the bird had not completed its first moult. Therefore, the bird was younger than 6 months of age.

To test for independence of transferrin phenotypes and shank phenotype a 2x3 contingency X2 was performed. The results failed to find any significant interactions (Table 7.3.4.). The X2 value with

2 degrees of freedom was 0.91 with a p value greater than .500. Com­ parison of white plumage color with colored plumage color (Table 7.3.5.) again failed to show any interaction (X~ = 2.62, p > .250). Comparing the transferrin phenotype frequencies between five eye color phenotypes

(Table 7.3.6.) did not show any significant heterogeneity (X~ = 4.24, p > 0.750). 49

The data was then divided into two classes, homozygotes and

heterozygotes and a 2x2 contingency X2 with Yate's correction was

used to test for any interactions. As in using simple phenotype

frequencies, no significant interactions were found. The X2 for sex

interactions was 0.000 (Table 7.3.7); for age it was 1.667, p > .100

(Table 7.3.8.); for shanks it was 0.170, p > 0.750 (Table 7.3.9.);

and for white plumage versus colored plumage it was 0.962, p > 0.250

(Table 7.3.10.). This subdivided data was then subjected to Woolf's

analysis to test for differences in homozygote and heterozygote

frequencies between two phenotypic classes. As was expected from the

preliminary analysis using contingency X2 ,s, there were no significant

differences found in any two by two comparisons. The X2 with one

degree of freedom for sex was .44 with a p value greater than 0.500

(Table 7.3.11); for young versus mature the X2 value was 2.57 with p

greater than .100 (Table 7.3.12); and for white versus colored the X2

value was .02 with a p value of 0.900 (Table 7.3.13.). Two by two

comparisons of transferrin homozygote and heterozygote frequencies in

two eye color phenotypes was then subjected to Woolf's analysis

(Table 7.3.14.) but no significant differences were found.

2 . A three by three contingency X was then employed to test for

first order interactions between the two other electrophoretic loci

examined. The results shown on Tables 7.3.15 and 7.3.16 gave X2

values of 2.10 and 2.33, when transferrin phenotypes were analyzed with 6-Phosphogluconic Acid Dehydrogenase and Alkaline Phosphatase

phenotypes respectively. The p values were greater than 0.500 50 and greater than 0.250 respectively, thus indicating no interactions which are significant.

The last test was to check for the homogeneity of transferrin phenotype frequencies over the years 1969, 1970, 1971, and 1972. The

X2 with 6 degrees of freedom was 6.05 with a p value greater than

0.250 (Table 7.3.17.). Thus, there appears to be no significant dif­ ferences between years.

Similar tests were performed on the other localities. The con­ tingency X2 to test for sex interactions was not significant in St.

Andrew's, Bethel, Pier 27 or Pier 32 (Table 7.3.18.). None of the localities show any dependence of transferrin phenotypes and age either (Table 7.3.19.). Only the sample from Pier 32 had a X2 value which approached significance, X~ = 5.48, .10 > p > .05. However, the pooled data from Piers 27 and 32 failed to show any significance, which suggests that this occurrence was probably spurious since there seems to be no difference in ecology which could change the selection intensity between the two localities.

The contingency X2 test for independence of transferrin phenotype and shank phenotype in the different localities other than the zoo was not significant except for the X2 for Pier 32 (Table 7.3.20.). The X2 value of 12.07 with 4 degrees of freedom is significant at the .05 level. Examining the data more closely one sees an excess of

TfrA/TfrB phenotypes in those birds with no feathering on the shanks.

This seems to suggest that the heterozygote transferrin bird with no feathering on its shank is more fit than the homozygote-non-feathered bird. Another hypothesis might be that there may be a disproportionate 51

number of AA X BB non-feathered shank matings thus leading to the dis­

proportionate number of AB individuals with this type of shanks. The

last and probably most feasible hypothesis is that this high X2 value

be one that would be expected by chance 5 out of 100 times. This is

the most feasible explanation since no other locality shows this

interaction between shank phenotype and transferrin phenotype.

Birds with orange eye color from St. Andrew's, Bethel, Pier 27

and Pier 32 were classified as to locality and transferrin type and a

test for independence between transferrin phenotype and locality

within the orange eye color phenotype was performed. As expected,

due to the heterogeneity of gene frequencies between these localities,

the X2 value was significant at the .005 level (Table 7.3.21.), thus

indicating that there is no disturbing effect of orange eye color on

gene frequencies; for it would have been surprising if there was no

significance at all between localities.

There were no significant differences between frequencies of

homozygotes and heterozygotes in males and females tested by Woolf's

analysis (Table 7.3.22.). This agrees with the tests for independence

performed above. The same analysis was performed on the young and mature populations (Table 7.3.23.). Again, these results also agree with the previous contingency X2 tests for independence.

A heterogeneity X2 was performed to test whether the underlying

genetic hypothesis is valid over all localities. The heterogeneity

X2 with 5 degrees of freedom was 6.73 with a p value greater than .100

(Table 7.3.24.); therefore, demonstrating that the genetic hypothesis is valid over all areas. Therefore, a test of homogeneity of gene 52

frequencies was performed over all localities by the use of the method

of Neel and Schull (1954). The X2 with 5 degrees of freedom was cal­

culated to be 29.37 with p < .005 (Table 7.3.25.) which supports the

hypothesis that the transferrin gene frequencies are heterogeneous

from locality to locality. A pair-wise comparison was made between

localities to test for independence of phenotype frequencies and local­ ity (Table 7.3.26.). These analyses revealed that the X2 's for the

Zoo X Bethel, Zoo X Pier 32, St. Andrew's X Pier 32, Bethel X Pier 32, and Waldron X Pier 32 comparisons are significant (Table 7.3.26.).

Therefore, Pier 32 has the most deviant gene frequency. Only Pier 27 did not differ significantly from Pier 32. This is expected since

Pier 27 is the nearest neighbor to Pier 32. The test for independence between locality and transferrin phenotype gave a X2 value of 25.94 for 10 degrees of freedom. This value is highly significant at the

.005 level (Table 7.3.27.), thus confirming the results of the pre­ vious test. Further discussion of these results will be continued in the section on population structure and inbreeding. Table 7.3.28 gives the maximum likelihood estimates of p and q with the simulta­ neous estimation of a derived from Yasuda's G-type program. The dis­ cussion of these estimates will be in the section on population structure and inbreeding. The results of pair-wise comparisons of homozygotes and heterozygotes between two localities is illustrated in Table 7.3.29. These results show significant differences in the homozygote and heterozygote frequencies for Zoo X Bethel, Zoo X Pier

32, and St. Andrew's X Bethel. It is noted here that the Zoo has the largest positive a estimate and St. Andrew's the second largest 53 positive estimate. These being 0.208 and 0.202 respectively. While on the opposite end, stands Bethel with the highest negative estimate of a -0.172, and Pier 32 with the smallest positive a estimate.

Therefore, there seems to be a clear correlation between the a esti­ mates and the results of Woolf's analysis of homozygotes and hetero­ zygotes, which is to be expected due to the relationship between the inbreeding coefficient and heterozygosity, F = l-H/2pq.

7.4. Summary

These analyses suggest that there are no significant interactions between the transferrin system and any of the other electrophoretic loci looked at; whether we compare phenotypic frequencies or homo­ zygote and heterozygote frequencies. The contingency X2 tests and

Woolf's analysis give comparable results. There also were found no significant interactions between the transferrin system and shank pheno­ type, sex, and age. The phenotype frequencies were stable over the three years of study at the zoo. There was a significant excess of heterozygotes found in the non-feath:red shank phenotype but this was ascribed to chance occurrence due to its failure to show up in other areas and due to the excess of heterozygotes found in Pier 32 initially. There was also found a significant deficiency of heterozy­ gotes in the zoo population but not in the others (Table 7.3.24.).

Finally, the gene frequencies estimated for the different localities were found to be very heterogeneous (X 2 = 29.37, p < .005), the significance of which will be discussed in the section on population structure and inbreeding. 54

VIII. Analysis of the 6-Phosphogluconic Acid Dehydrogenase Locus

8.1. Introduction

The enzyme 6-Phosphogluconic Acid Dehydrogenase, hereafter re-

ferred to as 6PGD, is one of the vital enzymes in the Embden-Meyerboff

carbohydrate metabolism pathway, in that it catalyzes the reaction below: 6PGD . ------~ TPN + 6-PG TPNH + r1bose-5-phosphate +C02

For its position in the pathway see Figure 8.1.1. This step follows

the glucose-6-phosphate dehydrogenase step in the pentose phosphate pathway. Both of these enzymes perform a major function of this path- way. That is, the reduction of TPN to TPNH which then provides the cell with reducing potential to reduce oxidized glutathione. Another major function of 6PGD lies in the production of ribose which is necessary for the synthesis of both DNA and RNA in nucleated cells.

The molecular weight of 6PGD was found by Kazazian (1966) to be

80,000 and since the enzyme is a dimer, the enzyme has two sub units each of 40,000 molecular weight. He found that the molecular weights were the same for man and Drosophila. Therefore, the molecular weight for the pigeon enzyme must not be too much different from this. In

Drosophila it is an X-linked locus (Kazazian et al., 1965); however, in man, pigeons, rabbits, ants, mice, rats, and monkeys it is auto- somally linked. Below are some species which show polymorphisms for this enzyme:

1. Man - Homo sapiens (Fildes and Parr, 1963)

2. Cat - Felis catus (Thuline et al., 1967) 55

3. Rat - Rattus norvegicus (Parr, 1966)

4. Deer Mouse - Peromyscus manicu1atus (Shaw, 1965)

5. Common Pigeon - Columba 1ivia (Parr, 1966)

6. Quail - Coturnix coturnix (Baker and Manwell, 1967)

7. Carp - Cyprinus carpio (Bender and Ohno, 1968)

8. Goldfish - Cyprinus curutus (Bender and Ohno, 1968)

9. Fly - Drosophila me1anogaster (Kazazian et a1., 1965)

10. Snails of the genus Cepea (Manwell and Baker, 1968)

11. Sen anemone - Actinia equina (Manwell and Baker, 1970)

12. Dove - Streptope1ia risoria (Cooper et a1., 1969)

Parr in 1966 found that the human type II, PGDA/PGDC, enzyme was about 80-100 per cent of type I; PGDA/PGDA, and type III was 70-90 per cent of type I. Carter et a1. (1966) found that treatment with urea reduced activity of types I, II and III, 37%, 70%, and 95% respectively.

Therefore, in man there seems to be differential levels of activity for each phenotype, with one being more efficient than the others. However, similar work has not been reported for the pigeon enzymes.

Personal communication with Dr. Childs in 1969 permitted me to obtain the data collected by him and his co-workers. A sampling of

103 wild birds were trapped in Baltimore and the results obtained were

54 6PGDA/6PGDA, 44 6PGDA/6PGDB, and 5 6PGDB/6PGDB. These phenotype frequencies were not significantly different from Hardy-Weinberg ex­ pectations. The gene frequencies obtained from this data were .26 for the 6PGDB and .74 for 6PGDA. 56

8.2. Materials and Methods

Initially three buffer systems were employed to test which system gave the most reliable and repeatable results. The three systems were as follows:

Fildes and Parr System (1963)

Gel Buffer - pH 7.0

.01M Phosphate

Electrolyte Buffer - pH 7.0

.1M Phosphate

Cooper et al. System (1969)

Gel Buffer - pH 8.5

24 mls. of stock buffer diluted to 600 mls.

Electrolyte Buffer - pH 8.5

.0998 M Tris-hydroxymethyl-aminomethane

.055 M Boric Acid

.0022 M Ethylenediamine-tetracetic-acid, tetrasodium salt

Stock Buffer - pH 8.5

O.09M Tris-hydroxyrnethyl-amino methane

0.5M Boric Acid

0.02M Ethylenediamine-tetracetic-acid, tetrasodium salt

Bowman et al. System (1966) modified

Stock Buffer - pH 8.6

0.9M Tris-hydroxyrnethyl-aminomethane

0.5M Boric Acid

0.02M Ethylenediamine-tetracetic-acid, tetrasodium salt 57

~~l Buffer - pH 8.6

25 mls. of stock buffer diluted to 500 mls.

Electrolyte Buffers

Cathode - 200 mls. stock diluted to 1 L.

Anode - 143 mls. stock diluted to 1 L.

For all systems 10 mg of TPN/ 100 mls. of buffer were added to

the cathodic tray. Horizontal electrophoresis was carried out at a

voltage drop of 6V/cm. or 25 m.A. current per run. The length of the

run was 16 hours. A 13% starch gel was used for all systems. The

samples were absorbed on a Whatman No.3 filter paper cut 3mm by 5mm.

They were inserted into a gel slot made by a horizontal cut across

the gel. The dimensions of the gel tray were 22.5 cm. long, 10.2 cm.

wide, and 3 mm. deep. The insert line was 3.8 cm. from the cathodic

wick and 15.2 cm. from the anodic wick. The total length of the gel

from wick to wick was 19 cm.

The bottom side of the gel was stained with the following stain

solution:

25 mg. 6-Phosphogluconic Acid

10 mg. TPN

2 mg. Phenazine Methosulphate

10 mg. Nitro Blue Tetrzolium

100 mls. Tris-maleic Buffer - pH 6.5

The gel was then incubated at 37°C in this mixture from 1-2 hours;

then, it was placed into wash of 5:5:1 ethanol, water and acetic acid, respectively. The wash made the background white and left the zymogram undisturbed. 58

8.3. Results and Discussion

Previous population work on 6PGD in Coturnix coturnix was done by

Manwell and Baker in 1969. Here they reported the 6PGD phenotypes and

found no significant deviation from Hardy-Weinberg expectations.

Their results were based on 73 samples.

Our study found that the 6PGD phenotypes in the zoo pigeons did

not deviate significantly from Hardy-Weinberg expectations. This was

surprising since a deficiency in heterozygotes was found in the trans­

ferrin locus. Therefore, analyses were performed to investigate whether

there could be any interactions which could explain this obvious di­

lemma. Hopefully, some evidence of 6PGDA/6PGDB, heterozygote advantage

could be found, which would not give rise to heterozygote deficiency.

A test to see if we could pull the data from all trapping areas was performed (Table 8.3.1.). The X2 value of 11.24 with 6 degrees of

freedom approached significance at the .05 level. A later pair-wise

analysis of areas comparing homozygotes and heterozygotes by Woolf's

analysis (Table 8.3.2.) showed that there were two significant differ­

ences between the homozygote and heterozygote frequencies between

localities. For area II compared with area III, the X2 value is 4.07 with one degree of freedom. This gives a p value of 0.050 > p > 0.025.

Examining the X value reveals that its value is greater than 1.0;

therefore, suggesting that the number of homo zygotes for area II is much higher than the homozygotes for area III. Or, looking at it from

the standpoint of heterozygotes, Area III has a larger freyuency of heterozygotes than does area II. Area III had a significant departure

from Hardy-Weinburg expectations. The X2 value was 8.46 with one 59

degree of freedom. This gives a p value of less than .005. Therefore,

a significant excess of heterozygotes was observed for area III.

The other significant difference was found for the Area III and

Area VI comparison. Again the X value was examined and was found to

be less than one; therefore, again suggesting that the frequency of

heterozygotes in Area III is much greater than in area VI. From these

two comparisons and Table 8.3.1. we see that area III will give a negative F value, while Areas II and VI will give positive values.

The question now arises concerning the differences between one area

and the others.

Area III is the common feeding ground of the whole population be­ cause this area is nearest to the snack bar where food is bought to

feed the pigeons (Figure 8.3.1.). Area VI is located behind the camel enclosure and giraffe enclosure. These are two breeding areas for the pigeons, for they are able to build adequate nests in the rafters of

these enclosures. Area II is a feeding area also, however, past trap­ ping data points out that 90% of the birds caught in this area were caught between the hours of 6 a.m. and 9 a.m. Our records show that

there are very few birds in the zoo at this time. The birds do not start to arrive at the zoo until 9:30-10:00 a.m. The reason is obvious for they are conditioned to the public arrival times. There­ fore, the conclusion reached is that the difference between area III and the other two areas is that the other two areas are samples of individuals breeding within the zoo or are individuals residing in the zoo proper; while those in area III are samples from both the zoo and adjacent areas. 60

Wahlund's principle of subdivided populations would suggest the opposite condition to be true. However, as stated by Selander (1970), heterotic effects may mask any effects of subdivision on phenotype frequencies. His results show very marked subdivisions of mice popu­ lations within barns; however, there was not always a deficiency of heterozygotes. He did find a correlation between D = (Ho-He)/He and population size, with small populations usually having positive D values and large, negative D values. These results can be used to explain our results, if we can assume that those birds from areas II and VI are from a large population within the zoo, while those from area III are predominately from small populations in outlying areas.

This appears to be very feasible since there are more actual breeding sites within the zoo than outside of it. In addition, we also know that at the retirement of Paul Breese, the former Zoo director, in

1965 all the pigeons were breeding within the zoo. Therefore, the populations breeding outside of the zoo are of recent origin (from personal communication with Paul Breese, 1972.)

Contingency X2 's were computed to test for any interactions between sex and 6PGD for the zoo population (Table 8.3.3.). The p value is greater than 0.500. No interactions were observed with age either (Table 8.3.4.). The test for independence of shank phenotypes with 6PGD phenotypes showed no significant interactions with X2 of

1.91 (Table 8.3.5.), and two degrees of freedom. There was also no difference in phenotype frequencies between the white and colored population (Table 8.3.6.). The p value was greater than 0.750. The analysis of independence of eye color and 6PGD showed no significant 61 interactions (Table 8.3.7.).

As in the analysis of the transferrin locus the data was then grouped into homo zygotes and heterozygotes and two by two contingency

X2 's were calculated with Yates' correction. There were no signifi­ cant interactions found for sex (Table 8.3.8.), age (Table 8.3.9.), shank (8.3.10.), and, plumage color (8.3.11.). Woolf's analysis was applied to the same grouped data but no differences in homozygote and heterozygote frequencies were found when compared by sex (Table 7.3.11.) by age (Table 7.3.12.), by shanks (Table 8.3.12.) nor by plumage color

(Table 7.3.13.). Two by two comparisons of eye colors resulted in no significant differences of homozygote frequencies and heterozygote frequencies between eye colors (Table 8.3.13.).

Contingency X2 's were calculated to test for first order inter­ actions between the other electrophoretic loci. No interactions were found between 6PGD and transferrin and APH. The X2 value for the latter comparison was 0.29 with 4 degrees of freedom. The p value was greater than 0.990 (Table 8.3.15.). Comparison of 6PGD phenotypes over the years 1969-1972, resulted in a contingency X2 of 12.92 with

6 degrees of freedom (Table 8.3.6.). This was significant at the

0.050 level. This difference in phenotype frequencies is to be expected in an inbred population.

Wright, in 1931, expressed the change in gene frequency for an inbred population with overdominance acting at a locus by the fo1low­ ign formula:

6q = q(l-q) [(s-Ft) - (s+t)(l-F)q]/w 62

Therefore, it is easy to see that with the absence of balance of se­

lection pressure one allele will be fixed over the other. Slight

selective pressure would have no appreciable effects on the relative

decrease of heterozygosity. However, very large selective pressure

would be quite important with the extreme case being the balanced

lethal system which results in 100% heterozygotes with any inbreed­

ing system.

The results of Robertson (1962) and Ewen and Thomson allowed them

to conclude that the amount of heterozygotes in a population depended

on the quasi-equilibrium point and also the initial gene frequencies

in a finite population where retardation of fixation due to hetero­

zygote advantage would be large for equilibrium values near central

frequencies. Large population size coupled with central equilibrium

frequencies will increase mean fixation time since there is a tendency

for the allele to approach its equilibrium value, and random effects

are, therefore, lessened. They found that small population size and

equilibrium points outside the range .20 < m < .80 tended to decrease

mean fixation time. However, when a (a = 2N[sl + s2]) was greater

than 16 this led again to an increase in fixation time. Hence,

changes in phenotype frequencies would be expected due to the large

population size involved. Reeve (1955) has compared the effects of heterosis on different systems of inbreeding. He computed a value n, which is the number of generations required to halve the amount of heterozygosis; given by n 10glO 0.5/log A; where A is the limiting = l0 ratio of heterozygosis gotten by taking the largest ~oot of the characteristic equation of the matrix of recurrence equations for the 63 different systems of close inbreeding. He concluded that weak selec­ tion causes relatively little delay in fixation by self-fertilization.

However, the delay increases greatly with a decrease in closeness of inbreeding. For example, with sl and s2 equal to .2, the half-periods, n, for heterozygosis under selfing, sib, half-sib, and double-first­ cousin are 3.80, 38.94, 149.77 and 1158 respectively. Thus, this illustrates the importance of the closeness of inbreeding upon the results of heterotic selection.

Looking at the other localities we see that Table 8.3.15. fails to show any heterogeneity of the underlying genetic hypothesis. Table

8.3.17. gives the maximum likelihood estimates of p, q and a from the different localities. The method of Neel and Schull (1954) was applied to test for homogeneity of gene frequencies between localities (Table

8.3.18.). The X2 with 5 degrees of freedom was 59.84 with a p value of less than .005. The discussion of this will be deferred until the section on population structure and inbreeding.

Pair-wise contingency X2 's were computed for locality-by-locality comparisons (Table 8.3.19.). The comparison of the zoo with all ~he other localities consistently had extremely high X2 values with all p values, except one, less than .005. No other locality-by-locality comparison was significant. To examine this difference, Woolf's analysis was applied to these localities with the data classified as to homozygotes or heterozygotes. Verifying the results of the previous test, all the X2 's were significant (Table 8.3.20.) for the zoo com­ parisons but for none of the others. All of the X values are less than one, therefore, the frequencies of homozygotes are significantly less 64

in the zoo sample than in the other localities. Examining the a

estimates in Table 8.3.17, reveals that all the a estimates are

positive, except for the zoo which has a negative estimate, -.050.

Therefore, there is evidence from the zoo data suggesting heterotic

effects at this locus, which, as pointed out by Selander (1970),

could mask any effects of inbreeding and subdivision of populations.

Therefore, it is not too surprising that a deficiency of hetero­

zygotes was not seen in the 6PGD locus.

Tests for independence between the 6PGD locus and the other

phenotypes were performed on data from these localities; as in the zoo

data. All localities failed to show any effects of sex on 6PGD pheno­

types (Table 8.3.21.). No interactions were found between age and

this locus at any of the localities (Table 8.3.22.). The X2 values

for comparisons of shank phenotypes and 6PGD phenotypes were all non­

significant (Table 8.3.23.). Woolf's analysis of two-by-two compari­

sons between 6PGD and sex, and age within the different localities

(Tables 8.3.24, 8.3.25.) failed to show any significant differences

between homozygote and heterozygote frequencies, supporting the

results obtained from contingency X2 analyses.

8.4. Summary

There were no interactions found between 6PGD and any of the other phenotypes, such as sex, age, eye color, and shank. A signifi­ cant excess of heterozygotes was found in area III of the zoo. This significance was discussed in the light of counteracting effects of inbreeding and selection and the correlation of population size with 65 deficiency and excess of heterozygotes. Woolf's analysis revealed that the zoo data differed from all other localities due to its deficiency of heterozygotes. Alpha estimates revealed that only the zoo estimate was negative. Significant differences in phenotypic frequencies over the period of study was discussed considering the joint effects of heterosis and inbreeding within a population.

Finally, a significant heterogeneity of gene frequencies between localities was found, discussion of which will be deferred to a later section. 66

IX. Analyses of the Alkaline Phosphatase Locus

9.1. Introduction

Alkaline phosphatase is an enzyme which hydrolyses esters of orthophosphoric acid in a pR optimum of about 10. The physiological

role played by this enzyme is in the hydrolysis of inorganic pyroshos­

phate. Inorganic pyrophosphate is a by-product of many anabolic path­ ways and is a metabolic poison to other processes; therefore, it is essential that it be removed from the system. This important role of alkaline phosphatase, hereafter referred to as APR, is supported by the occurrence of high turnover levels of cells, snythesis of new cells; or, a high anabolic rate, synthesis of intracellular materials (Bell,

1971). In both processes inorganic pyrophosphate is produced (Bell,

1971).

Tanabe and Wilcox, in 1960, found that APR plasma levels are higher in young chicks up to 4 weeks of age and usually decrease and rear.h a low point at 14 to 18 weeks of age. Common in 1936 also report­ ed high alkaline phosphatase levels in White Wyandotte chicks. This also gives evidence for the theoretic physiological role of APH.

Heritable differences in APH levels of fowl have been reported by

Stutts et a1., 1957; Tanabe and Wilcox, 1960; McDaniel and Chute, 1961;

Law and Munro, 1965; Wilcox, 1966. Therefore, there seems to be genetical differences in different breeds of fowl.

It is generally agreed that mature birds have lower levels of APH in plasma; however, the relationship seems to be less clear. Bell (1960) reported that APH activity of laying hens was greatly raised because of the increased osteoblastic activity and turnover, and an attempt to 67

replace bone lost in shell formation. In the domestic fowl, the hen,

having been bred to ovulate daily, must draw 2g. of calcium daily

from its food source to produce commercially important eggs; however,

her maximum capacity to absorb calcium through her intestines is

only 1.5g. per 24 hours. Therefore, she is forced to obtain this

extra .5g of calcium from her skeletal structure. If the laying

period is extended over too long a period she undergoes osteoporosis,

a condition where the entire bone structure becomes diminished; the

ratio of organic to minerals remains the same but the skeletal bulk

becomes lessened without overall change in the total bone biochemistry

(Bell, 1964). Bell, (1964), describes an inherited condition called cagelayers fatigue (osteoporosis). This condition is correlated in its acute stages with high APH levels; therefore, there seems to be a very strong association of APH levels and egg laying when this is ac­ companied with osteoclastic activity. However, Common (1936) and

Stutts et al. (1957) failed to find any differences in APH levels in laying and non-laying hens.

Electrophoretic variants of APH in fowl have been described by Law and Munro in 1965. They found two forms of the enzyme, APH 2 and APH4, with the latter being the slow component. Their breeding data confirmed 2 4 the dominance of the APH allele over the APH allele. In a subsequent paper, (Law, 1967), Law found a direct association between the APH isozyme pattern and that of the leucine aminopeptidase pattern. By 2 treatment of the fast APH and LAPF isozyme with neuraminidase, which is a carbohydrase that splits off terminal sialic acid components, Law found that the fast component could be changed to the slow component 68 which was not affected by the neuraminidase treatment; therefore, suggesting the hypothesis that the gene for the fast components is attaching sialic acid residues to the basic APH and leucine animopepti- dase protein structure.

Electrophoretic variants of alkaline phosphatase isozymes were found in Columba livia by Brown and Manley in 1970. They found 7 phenotypes which they concluded were the result of the codominance of A B 4 APH alleles with the following frequencies: APH , 0.284; APH , C D 0.602; APH , 0.045; and APH , 0.068.

Other organisms where variants of alkaline phosphatase have been observed follow:

Man - Homo sapiens, (Hodson et al., 1962)

Sheep - avis aries, (Gahne, 1963)

Cattle - Bos taurus, (Rendel and Stormont, 1964)

Drosophila Melanogaster, (Beckman and Johnson, 1964)

9.2. Materials and Methods

Two buffer systems were tested in the horizontal starch gel system.

The first buffer system adapted from Law and Munro (1965) failed to AAAB resolve the APH /APH and APH /APH phenotypes.

Law and Munro System

Gel Buffer - pH 8.6

.04M - Tris-hydroxymethyl-aminomethane

.003M - Boric Acid

Electrolyte - same as above 69

E-C System - (E-C Bulletin, Vol. 5, No.1, 1971)

Gel Buffer - pH 9.0

137 gms. - Tris-hydroxymethyl-aminomethane

9 grns. - Maleic acid

Water to 4 liters

The first system was used in a horizontal starch gel electro­

phoresis apparatus. 70 grams of Connaught starch was used per 500 mls. of buffer. The size of the gel trays were 22.5 cm. long, 10.2 cm. wide and 3 mrn. deep. The insert line was 3.8 cm. from the cathodic wick and 15.2 cm. from the anodic wick, the total length of the gel being 19 cm. from wick to wick. Whatman E 17 filter paper was used as inserts. The voltage was set at 250V and lOrn.A. The total run

took 24 hours for a migration of 2 cm.

A cyanogum horizontal electrophoresis system was used for the

E-C system. This system was preferred due to the better resolution and shorter running time. The size of the gel tray was 23.9 cm. long,

11.5 cm. wide, and 2 mrn. deep. The distance from wick to wick was 18.9 cm. The distance from the insert line to the cathodic wick was 2.6 cm., and to the anodic wick was 16.3 cm. The voltage was set at 250V for

3 hours; the milliamperage came to 20 m.A. per gel. A 7% cyanogum gel made up of the following chemicals was used as a gel matrix.

7 gms. Cyanogum 41

.10 gm. Ammonium Persulphate

.1 cc. Tetramethylethylene diamine

E-C Buffer to 100 mls. 70

Later disc electrophoresis was run on three representative

samples of the three phenotypes observed. The procedure is described

by Brown and Manley (1970):

7% Canalo gels that stack at 8.9 and run at pH 9.5

Electrolyte Tris 6.0 grms. per liter H 0 2 Glycine 28.8 gms. per liter H 0 2 pH 8.3

2.5 m.A. were applied per gel for about 90 minutes. The resolu-

tion was the same or worse than the E-C horizontal gel electrophoresis

system. Therefore, the disc electrophoresis system was not utilized further.

The staining procedure was the same for each system (Beckman and

Johnson, 1964):

100 mg. a-napthyl-acid phosphate

.5 g. polyvinyl pyrolidone

2 g. Sodium chloride

100 mg. Fast Blue RF. Salt

.25 mls. 10% Magnesium chloride

.25 mls. 10% Manganese chloride

The gels were incubated in this mixture for 1-2 hours at 37.5° C.

Samples of very young birds were diluted 1:2 with gel buffer since in an undiluted sample the activity was too strong to differentiate the AAAB APH / APH and APH / APH phenotypes.

9.3. Results and Discussion

A heterogeneity X2 test was performed on the zoo data to test 71 whether we could pull the samples from the different areas (Table

9.3.1.). The heterogeneity X2 was 11.03 with 6 degrees of freedom.

The p value approached significance at the .05 level. Looking at the individual X2 's for each area revealed that 3 out of the 7 areas had significant X2 's at the .005 level; areas II, III, and VI. All of these areas revealed a significant deficiency of heterozygotes. It should be noted that these same three areas showed a deficiency of heterozygotes in the transferrin system. However, one cannot conclude from this that only areas II, III, and VI have inbreeding, for the number of samples from the other areas were quite small, usually less than 20. The pooled X2 was again significant at the .005 level, which signifies a departure from Hardy-Weinberg Equilibrium Law, that being a significant deficiency in heterozygotes. When Woolf's analysis was applied to this same data with APHA/APHA and APHB/APHB pooled as homozygotes, we find two significant X2 's, area II compared with III and area II compared with IV (Table 9.3.2.). The X value reveals that the frequency of homozygotes is greater in area II than in the other two areas. This was expected since area II showed the largest value

2 for the goodness of fit X • The difference between area II and IV bears no information, for the total sample size for area IV was 11 birds and hence, no discussion will be conferred on this significance except, that according to Cannings and Edwards (1969), small sample sizes tend to show heterotic effects whether these effects are there or not.

However, comparing II with III, we can use the same argument put forth in section 8.3 for the differences found in 6PGD homozygote and 72 heterozygote frequencies for the same two areas. That argument is that the degree of inbreeding is not the same for the two areas because area three is a sampling of not only the zoo pigeons, but also is a sampling of birds from the outlying areas. From past history, it is known that this dispersal of breeding sites into the outlying areas were recent in origin (Paul Breese, Personal Communi­ cation, 1972) and, therefore, levels of inbreeding have not built up to the same extent as those in the zoo.

Contingency Chi-square tests for independence between APR pheno­ types and other phenotypes will be discussed here. As in the other electrophoretic loci no interactions were found with sex (Table 9.3.3.); age (Table 9.3.4.); and, shank phenotype (Table 9.3.5.). Woolf's ana­ lysis on the same data, except grouping all homozygotes, confirmed the previous results obtained from the previous tests (Tables 7.3.11; 7.3.12;

8.3.12.). Two by two contingency Chi-square tests were also performed on the same grouped data which confirmed the results of Woolf's analysis

(Tables 9.3.6; 9.3.7; 9.3.8). Therefore, no significant differences in homozygote and heterozygote frequencies were found.

Plumage color also had no effects on APR phenotype frequencies

(Table 9.3.9.). The X2 value with 2 degrees of freedom was .07 with a p value greater than .950. A 5x3 contingency Chi-square test was performed to test for interactions between eye color and APR phenotypes.

The X2 value obtained was 3.97 with 8 degrees of freedom and a p value greater than .750 (Table 9.3.10.). Thus, it can be concluded that no interactions were observed in this data. Applying Woolf's analysis on these same data again confirmed the results of the tests for 73 independence (Table 7.3.13. and Table 9.3.11, respectively). A pair- wise comparison of APH phenotypes in two different eye colors revealed no significant differences in homozygote-heterozygote frequencies

(Table 9.3.11.).

A heterogeneity X2 was then employed to test the validity of the underlying genetic hypothesis over all the localities trapped. Table

9.3.12. revealed that there was no heterogeneity observed (X~ = 2.35,

P > .750). Examining the individual X2 's in the table suggested that the zoo and Pier 32 had significant deviations from Hardy-Weinberg

Equilibrium Law. However, looking at the observed and expected numbers

2 for each individual X , it can be seen that the major contribution to the X2 value for Pier 32 carne from dividing .28 into (1.72)2, 2 (Observed-Expected) /Expected. Li in 1955 revealed that the calcula- tion of expected values by the formula Gq2 was not valid for small sample sizes and low frequencies of q. Levene's formula (1949) will not help since it would only decrease the number of homozygotes expected and therefore increase the X2 value. Snedecor and Cochran (1967) suggests that whenever the expectation is less than one, then classes must be pooled to satisfy this condition. By pooling the homozygous classes the X2 value becomes non-significant, therefore, suggesting no significant deficiency in heterozygotes. Table 9.3.13. shows an un- usually high estimate for F; however, this is a spurious result due to small sample size. The significance of the deficiency of heterozygotes in the zoo sample will be discussed in the section on population structure and inbreeding. 74

A test of homogeneity of gene frequencies was computed (Neel and

Schull, 1954). The X2 with 5 degrees of freedom was 18.68, which has

a p value of less than .005; therefore, it can be concluded that a

significant deviation in gene frequencies between localities was

observed. (Table 9.3.14). Discussion of the randomness of these

deviations will be discussed in a later section.

A contingency X2 comparing phenotypes of the six localities was

computed. The X2 value obtained was 42.85 with 10 degrees of freedom,

with a p value of less than .005. Therefore, confirming that the gene

frequencies are heterogeneous over localities.

Pair-wise comparisons of localities with respect to APH phenotype

frequencies disclosed that Zoo X St. Andrew's (X~ = 23.85, p < .005);

Zoo X Pier 27 (X~ = 9.75, .025 > p > .010); and Zoo X Pier 32

(X~ = 8.92, .025 > p > .010) for the zoo comparisons, gave significant

differences. St. Andrew's belfry compared with all other localities

gave significant differences, which is to be expected since the

frequency of the APHA allele was zero (Table 9.3.15). Woolf's ana­

lysis revealed that the frequency of homozygotes was significantly

lower in the Zoo than at Pier 27 and Pier 32 (Table 9.3.16). This is

to be expected due to the low frequency of APHA in each of these

localities.

Woolf's analysis of comparisons made by sex failed to disclose

any significant differences of homozygote and heterozygote frequencies within a locality (Table 9.3.17). No differences were found in young

compared with old birds for Bethel (Table 9.3.18). If the locality is not on the table, the reason is that the number of heterozygotes was 75

zero in one of the classes thus making X = 0 or undefined due to

division by O. Similarly, no differences in phenotype frequencies

between sex, age and shank phenotypes were found for the APR locus

(Tables 9.3.19; 9.3.20; 9.3.21). A test for independence of APR

phenotypes and years revealed a very significant heterogeneity as to

APR phenotypes over the years 1969-1972 (Table 9.3.22). The gene

frequencies calculated for these years are .145, .137, .193, and .180,

respectively. Therefore, there is a gene frequency change from year

to year with 1971 having the highest frequency. As stated by Selander

(1970), one of the difficulties of this kind of study is the problem of gene frequency variation over time. Calculating X = hK/Rk, a

Woolf's analysis variable, for comparisons of 1969 with the other years, seems to show a decrease in homozygote frequencies; therefore, suggesting heterotic effects. However, the change in gene frequencies would have caused this decrease. Wright (1969) expressed the change in gene frequency when heterozygote advantage and inbreeding are op­ posing each other in the following formula:

6q = q(l-q) [(s-Ft) - (s+t)(l-F)]q/w

Therefore, if the selection pressure is not balanced, one allele will tend to be fixed. However, if selection pressure is slightly larger, the relative decrease in heterozygosis is not affected, but strong selection would be very important. If selection is strong the change in gene frequency can be explained by this formula. However, the most probable explanation may lie in increased migration from other popula­ tions which would cause a change in gene frequency where 6q = -m(q-q), 76 m = proportion migrated (Li, 1955). Banding records support this hypothesis of increased migration. Both forces, however, may be act­ ing to give these changes. The effect of finite size on over­ dominance and inbreeding within a population was discussed in section

8.3. where a similar situation occurred for the 6PGD locus, and there­ fore, you are asked to return to this section.

9.4. Summary

A deficiency of heterozygotes was found to be significant in the zoo population but not in the other localities studied. A significant difference in homozygotic and heterozygotic frequencies was found between areas II and III. The reason for which was explained by the difference in the amount of inbreeding due to the different popula­ tions sampled in these two areas. A significant interaction of shank phenotype and APH phenotype was found; however, this was decided to be spurious due to an expectation of less than one. No other interactions were found with any of the other loci examined. A significant hetero­ geneity of gene frequencies between localities was found; the signifi­ cance of which will be discussed in the section on population structure and inbreeding. Pair-wise comparisons revealed that the Zoo differed greatly from St. Andrew's, Pier 27, and Pier 32 in phenotype frequencies, the discussion of which is again deferred to the section on population structure and inbreeding. Significant differences in Zoo APH phenotype frequencies over the years 1969-1972 were also found. These differences were ascribed to gene frequency changes, which may have come about by migration, interaction of heterozygote advantage and inbreeding, or, by both of these forces acting on the population. 77

X. Analyses of Morphological Characters

10.1. Feathered Shanks

A description of the different shank phenotypes (grouse, muffed, and slippered) is given in the Appendix. None of the first two pheno­ types were observed in the wild populations studied, which included

1058 birds. This is expected due to the fact that these first two phenotypes are commonly found in fancy breeds of pigeons, which would not fare too well in the wild. However, the slippered phenotype is the type of shank feathering found in many breeds including the breed from which the zoo population was derived (Levi, 1969). Birds with heavy feathering on the shanks and middle toe were classified as homozygote + +. Those birds with moderate or few feathers were classi­ fied as heterozygotes. Naked shanks were classified as homozygote - -.

Breeding obsernations revealed that naked X naked matings never produced young with feathers on its shanks; whereas heterozygote X naked produced both naked and moderately feathered shank young; therefore, supporting the hypothesis of a co-dominant two allele hypothesis.

Analysis of this locus in the zoo population, however, was restricted to feathered and non-feathered phenotypes since earlier recordings did not distinguish between homozygotes, + +, and hetero­ zygotes, + - No interactions were found for sex and age from results of a 2x2 contingency X2 (Tables 10.1.1,10.1.2.). However, there was a significant heterogeneity of feathered and non-feathered phenotypes over the years of the study (Table 10.1.3.). The phenotype frequency in 1969 was 0.12 and dropped to 0.02 in 1972. This difference can be 78 explained most clearly by the effects of finite population size, which leads to random fixation of one or the other of the alleles with no selection pressure. It has an even greater effect if genes are at low frequencies (Wright, 1931). The gene frequency of the offspring popu- lation with N individuals will vary by chance according to the follow- ing binomial expansion: (Li, 1955)

2N [p(A) + q(a)]

The probability of a certain value q. j/2N in the next generation is J (Wright, 1931)

2N 2Npj 2Nqj = (2Nqj) p q

Thus, it can be seen that values of q near 0 or 1, will have smaller magnitudes of chance variations than those of intermediate frequencies.

These observations, however, assumes no selection pressure. Robertson

(1962) considers the problem of heterozygote advantage in an inbred finite population and concludes that if the equilibrium frequency lies outside the range .20 ~ q ~ .80 there is an actual increase in fixa- tion time for small population sizes; therefore, also predicting a fixation of one of the alleles.

The X2 for goodness of fit to Hardy-Weinberg equilibrium was 9.79 with one degree of freedom. The p value obtained was less than .005.

Therefore, as in the transferrin and APH loci a significant deficiency in heterozygotes was observed. This significance will be discussed in the next section on population structure and inbreeding. 79

Comparison of the phenotype frequencies of all localities re­ vealed a significant X2 value of 60.86 with 10 degrees of freedom; p less than 0.005 (Table 10.1.4); therefore, suggesting significant differences in gene frequencies between localities. The test of homogeneity of gene frequencies (Nee1 and Schull, 1954) confirmed the above results. The X2 obtained was significant at the .005 level

(Table 10.1.5). Table 10.1.6 shows that this significant difference was still present even when the zoo sample was removed; however, the level of significance was lower, suggesting, as was expected from looking at the maximum likelihood gene frequency estimates shown in

Table 10.1.7, that the frequency of the feathered gene in the zoo population was very different from the other estimates. Discussion of whether these differences are random or not will be discussed with population structure.

Tables 10.1.8 and 10.1.9 failed to disclose any interactions with sex and age, respectively, by a contingency X2 test. Woolf's analysis did not reveal any difference in homozygote and heterozygote frequencies between males and females, and young and mature birds (Tables 10.1.10 and 10.1.11) in any of the localities. Interactions with the electro­ phoretic loci was found insignificant for all localities (Table 10.1.10) in which contingency X2 's could be calculated. The zoo contingency

X2 's are found in the sections on the individual electrophoretic loci.

10.2. Analyses of Plumage Color and Pattern Phenotypes

There are five major plumage color phenotypes in pigeons: black, blue, red, brown, and white. With each of these colors are four 80

T pattern alleles (C ; C; +; c), which causes heavy checking (very few

clumped granules); moderate to light checking (many clumped granules);

barred pattern (mostly clumped granules, except for barred area on

wings and tail); barless (all clumped granules, except for band on

tail). Therefore, the differences between the patterns are the

amount of spreading factor present. They form a dominance hierarchy T of C > C > + > c. In this study, c will stand for the barred pheno-

type since no barless patterns were observed. Therefore, c was used

for the lowest allele of the hierarchy found in the population.

Lloyd-Jones (1915) studied microscopically the nature of the red

and black color phenotypes. He found that the pigments were granular

under the microscope. Their sizes were either spherical or rod-like.

The pigment for black was black and for red was reddish brown. Hawkins

(1931) described brown pigeons as having brown pigment and spherical

granules. All of these pigments are chemically similar for they are

melanins and confined to the epidermal portion of the feather. Blue

is similar to black, for the pigment color is the same but the granules

are not evenly spread throughout the feather; they are clumped. White

has no pigment at all and can be either dominant, recessive, co-

dominant, or epistatic to the other color genes. The genetics of white is thus very complicated and fails to be a suitable color to

study in our populations as far as gene frequencies are concerned.

There are two types of red phenotypes, Ash Red, a dominant sex-

linked gene and Recessive Red, which differs from Ash Red in that it is not sex-linked and is recessive to blue. Phenotypically they differ

since Ash red exhibits areas (head, tail, rump, and wing) of ashy grey 81 coloration in combination with the red phenotype in Ash Red; while

Recessive Red is solid red throughout. None of the birds captured exhibited the Red Recessive phenotype; all were Ash Reds. A Examining the maximum likelihood frequencies of the B allele in the male and female populations of the different localities, re- vealed a marked heterogeneity in frequencies (Table 10.2.1). The frequencies in the male populations varied from 0.37 in the zoo popu- lation to 0 in Pier 32; and in the female populations they varied from

0.31 in the zoo population to 0 in Piers 27 and 32. None of the localities, except St. Andrew's revealed any significant differences between male and female frequencies.

The male and female frequencies estimated for St. Andrew's

Belfry were 0.07 ± 0.02 and 0.01 ± 0.01 respectively. This hetero- geneity between the sexes obviously represents a case of disequilibrium A in the B allele frequencies between the sexes; for if the population were in equilibrium, both male and female frequencies would be equal, since the frequency in Lhe heterogametic sex would be equal to the frequency in the homogametic sex in the previous generation. Due to A the low frequency of the B in the population, this disequilibrium could easily have been the result of random genetic drift. Since the A frequency of the B allele will vary by chance in accordance with the following binomial expansion: (Li, 1955)

with the + and b alleles pulled to form q(a), and from the formulations of Wright (1931) on stochastic processes, one can easily see that the 82 gene frequencies near to 0 and 1 will have smaller variations in gene frequencies and thus are subjected more to the ultimate fixation of the allele in low frequency.

The comparison of male and female frequencies for the wild-type allele in the zoo population reveals that this allele is not in equilibrium in the two sexes and is actually decreasing in frequency in the males, since the female frequency is .61 ± .05 and, in males,

.46 ± .08, and since the qn = qn-l This decrease in q is accompanied, A as expected, since p+q+r = 1, with a concomitant increase in B and b frequencies. Therefore, in the zoo population the frequency of the A B and b alleles seem to be on the increase. The opposite effect is observed in the St. Andrew's and Bethel localities, which exhibit a rise in the + allele frequency which, again, is due to the affects of random genetic drift on low frequencies of alleles in these popula­ A tions. The increase in B and b frequencies can easily be explained from the standpoint of the masking effects of the white gene. Since more heterozygous birds would mate, more red and brown phenotypes would be unmasked, therefore leading to a higher estimation of these gene frequencies in subsequent generations.

The observation of the presence of the brown allele in females only in the St. Andrew's and Bethel localities is far from surprising, for, since the frequency in these populations is so low and the sample sizes are 66 and 59, respectively, the probability of capturing a brown male would be very slim indeed. Unless these females are recent migrants, and were produced from one heterozygote sire, brown males would be expected to be found in a large sampling. Piers 27 and 83

32 had 0 frequencies for the brown allele.

The analysis of the b allele by J.B.S. Haldane's (1963) method of studying sex-linked loci in samples from populations reveals that -1 ~ cr ,a measure of difference between male and female frequencies explained in Section 6.9. was equal to -2.15, which has a probability level of less than .005 (Table 10.2.2); therefore, suggesting a deficiency in female brown phenotypes. The data, however, is not adequate in numbers to allow one to conclude whether this observation is real or not, for J.B.S. Haldane's technique is based on large sampling theory. Inbreeding would also lead to excess males in the population. The numbers observed in males and females for the red, wild, and brown plumage phenotypes in males and females are shown in

Table (10.2.3). Contingency X2 's failed to show any heterogeneity between sexes.

The analysis of maximum likelihood pattern frequencies showed marked heterogeneity of frequencies between the zoo and all other locations. The zoo had a frequency of .17 ± .03 while all other localities had frequencies of 0.24 to 0.39 with standard errors of

± .05 (Table 10.2.4). The frequencies of the C allele were very stable over most areas with all standard errors overlapping each other except for that at Pier 32 with Pier 27, Waldron and Bethel.

There may be two explanations for this, either genetic drift or mis- classification of blue checks and black checks; how~ver, this would mean mistakingly classifying a black check as a blue check, which is very unlikely due to the marked differentiation between the two types

(Figure 10.2.1). The frequencies of the c, barred allele, as in the 84

T C allele varied greatly from locality to locality; with frequencies ranging from 0.58 ± .05 in the zoo population to a low of 0.31 ± .06 in the Pier 32 population. The differences found at this allele also is apparently due to genetic drift. Further discussion will be deferred to the section on population structure.

Pair-wise comparisons of localities for three phenotypic patterns,

Black Check, Blue Check and Blue Bar, gave three significant contingency

X2 's, the Zoo X St. Andrew's, Zoo X Pier 27, and Zoo X Pier 32 compari­ sons (Table 10.2.5), which were expected due to their large allele frequency differences. A contingency X2 for all localities except the zoo was computed to be 14.12 with 12 degrees of freedom, p greater than .250; therefore, displaying no significant phenotype frequency differences in these localities (Table 10.2.6); these phenotypes being

Red, Blue Bar, Blue Check, and Black Check.

10.3. Analysis of Eye color Phenotypes

In all populations studied there were only five major eye color phenotypes, Red, Brown, Pearl, Orange, and Yellow. The Brown pheno­ type is the Bull Eye described in Appendix 1. Pearl eye was shown by

Hollander and Owen (1939) to be a simple recessive to orange. Bull eye is suggested to be simple recessive of orange (Levi, 1969). Pearl is suggested by the matings of Christie and Wriedt (1923) to be dominant to bull. However, the relationship between yellow and the rest of the color phenotypes is unknown. Therefore, the analysis of eye color phenotypes will deal with only phenotypic frequencies. 85

In the Zoo population a significant interaction was observed between white and colored plumage and eye color. The contingency X2 observed was 89.68 with four degrees of freedom; the p value is less than .001 (Table 10.3.1). This interaction confirms the observations of Levi (1969) that there is a tendency for bull eye to be associated with white plumage color. The other eye colors found in white birds usually were accompanied by foul feathering, feathers of other colors besides white present on the bird. Levi (1969) states that birds with colored plumage usually have eyes ranging in color from red through orange to yellow, hence, the high frequency of brown-eyed birds in the white population and lower frequency in the colored population (Table 10.3.1).

The contingency X2 to·test for differences in eye color phenotype frequencies over the years 1970, 1971, 1972, revealed a significant difference (X~ = 23.83, p < .01) (Table 10.3.2). Examining the indi­ vidual chi-squares reveals that the biggest differences have been in the Red, Orange, and Yellow eye colors in the years 1970 and 1971.

The phenotype frequencies reveal that there is a decrease in the red phenotype frequencies, an increase in orange phenotype frequencies, and, a decrease in yellow phenotype frequencies. However, these results are difficult to interpret since they are confounded by the plumage color differences found over the same years. Therefore it can only be concluded that the difference in eye color freqllencies from year to year is associated with plumage color phenotype frequency changes. 86

No interactions between sex and eye color phenotypes were found

(Table 10.3.3). Age had a very obvious effect on eye color. The eye color of all young birds was the same; a greenish-brown eye was always observed. The true eye color of the young bird is therefore masked until usually the time of its first moult.

Some mature birds were observed with the color of one eye differ­ ent from the other; sometimes half or a portion of one eye would be of a different color. A similar observation was reported in Levi

(1969). However, the genetics of these abnormal phenotypes are un­ known.

In comparing the phenotype frequencies between localities in a contingency X2 analysis, the X2 value obtained was 66.60 with 12 degrees of freedom, and a p value of less than .005 (Table 10.3.4).

Therefore there is a significant heterogeneity of eye color phenotype frequencies in the different areas, with the zoo excluded. It is noted that no associated differences in plumage color phenotype frequencies in these same localities were found (Table 10.2.6). The discussion of this significance will be deferred to the section on population structure.

10.4. Sex and Age Distributions

Sex was determined in the zoo sample mainly by phenotypic obser­ vations. The males tend to have a larger crown, wattle, and body size.

The neck of a male is also thicker than the female. However, the probability of being wrong is 20-30%. Miller in 1955 reported an accurate method to determine sex in Columbiformes. The method was ~7 employed only on birds kept for the breeding colony. Due to the chance of injury to hens, who are ovulating at the time of the exami­ nation, no birds which were to be banded and released were subjected to this method of sexing. The method has two major requirements, good lighting and an infant nasal speculum to spread the cloaca. The feathers surrounding the cloaca are removed first; then the prongs of the nasal speculum are placed as deeply as it can go into the cloaca.

It is essential that the prongs go below the muscular lip of the cloaca; the cloaca is then spread open. Cocks will have two papillae at the 3 and 9 o'clock positions, while hens will have a whitish circular area at about 3 o'clock. This whitish area is the opening of the oviduct into the cloaca. The function of the papillae in the cocks is probably for grasping the vent of the hen during copulation.

All birds in the St. Andrew's, Bethel, Waldron, Pier 27, and

Pier 32 localities were sexed by autopsy. The feathers were removed from the breast and a horizontal cut was made on the posterior end of the sternum. Then two lateral cuts were made parallel to the mid­ ventral line, cutt~ng anteriorly. The digestive tract was then lifted out of the way on the left side of the bird, thereby revealing the sex organs.

By young, it is meant that these birds had not started or completed their first moult. The feathers of the neck were finer and softer in appearance in the young birds. The birds recorded as young, therefore, ranged in age from 3 weeks (fledgling) to about

4-6 months of age when moulting is completed. The young usually were smaller in size and the wattle and eye cere was less developed. 88

The eye color also was greenish brown. The eye color was used as the major criterion when classification was questionable. However, the age at which the eye color changes from the juvenile to the adult color is unknown. It is, however, prior to sexual maturity which is usually

6 months of age. Therefore, all birds classified as young were younger than 6 months of age.

The contingency X2 to test for differences in sex distribution among localities was 5.66 with 5 degrees of freedom. This corresponded to a p value of greater than 0.250 (Table 10.4.1). Therefore, this value suggests no difference in male and female frequencies between localities. It is also noted that there was no significant deviation from a 50% occurrence of the sexes when X2 's with Yates' correction were computed.

There was a very significant difference found in the age distri­ bution among populations. The contingency X2 value obtained was

25.33 for 5 degrees of freedom, corresponding to a p value of less than .005. However, this was expected since the zoo data included all birds caught over the 4 years. Shooting birds only at night in Pier 27 failed to result in an unbiased sample since the adults were sitting the nests at night and were, therefore, out of sight behind the steel girders. The drugged grain at St. Andrew's and Bethel were placed predominately near nests and therefore would result in more mature birds being caught.

It can be concluded that no sex distribution differences were found among or within localities. However, no conclusion could be reached about age distribution, because of unbiased sampling. 89

Therefore, differences found between localities cannot be ascribed to differences in sex distribution and probably cannot be ascribed to age distribution differences.

10.5. Summary

Analysis of the shank phenotype revealed no significant inter­ actions with sex and age; however, a significant heterogeneity of phenotype frequencies was found over the period of study. The signi­ ficance of which was discussed in terms of random genetic drift and heterosis in a finite population. A significant deficiency of heterozygotes was found for the shank locus, discussion of which was deferred to the next section. Heterogeneity in both gene frequencies and phenotype frequencies were found between localities.

Description of the five major color types and pattern phenotypes was given. Haldane's test for agreement with expected phenotype fre­ quencies in a randomly mating population revealed a significant deficiency of recessive females, or likewise, an excess of male recessives. This deficiency or excess was explained in the light of A inbreeding and chance occurrence. The increase in the B and b alleles are discussed. The marked differences in allele frequencies are dis­ cussed from the view of random genetic drift. Analysis of the pattern alleles also revealed considerabl~ heterogeneity, except for the C allele which tended to be stable over all localities. This signifi­ cance was discussed in conjunction with Mainardi's theory of assortative mating favoring dominant alleles. 90

Analysis of the eye color phenotypes revealed a significant interaction with plumage color phenotypes, which confirmed previous data by Levi (1969). No interactions were found with sex; however, the interaction with age is discussed. Again, a heterogeneity of phenotypes over localities was observed. This significance is discussed in the next section. Finally, it was concluded that no heterogeneity was found in the sex distribution between localities, and the significant age distribution heterogeneity was accounted for by bias sampling. 91

XI. Population Structure and Inbreeding

11.1. Randomly Mating Populations and Hardy-Weinberg Equilibrium

By randomly mating is meant that the probability of choosing a

particular type for a mate is equal to the relative frequency of

that type in the population. Hardy (1908) and Weinberg (1908) first

reported that the genotypes in the generation following random mating would be equal to the following proportion, p2 (AA), 2pq (AB), and q2 (BB). This has since then been known as the Hardy-Weinberg

Equilibrium Law. However, it should be pointed out that this law operates under rather rigorous and unnatural assumptions. That is, it holds if there are no mutation selection, migration, or random drift forces present. It should also be noted that the Hardy­

Weinberg principle holds true for any infinitely large population, for, otherwise, the effects of random drift will apply.

There are two major departures from random mating. The first is inbreeding, where the correlation between mated individuals is greater than if they were chosen at random. The second is assorta­ tive mating; that is, the greater phenotypic similarity between indi­ viduals. The effects of these non-random perturbations will be discussed at this time. However, it should be noted that compliance with Hardy-Weinberg expectations does not imply the absence of deterministic and stochastic forces. It can only imply that the sum effects of these forces equals O. 92

11.2. The Inbreeding Coefficient: Its Methods of Estimation and

Powers of Detection

The inbreeding coefficient, F, was first defined by Wright (1922).

This inbreeding coefficient of Wright's was derived from the corre1a- tion of uniting gametes. Therefore, it was defined as the correlation between uniting genetics. Ma1ecot (1948) later defined F in terms of probability theory, which says that F is the probability that two genes are identical by descent. An obvious difference between the two is that Wright's F can vary from -1 to +1, while the inbreeding coefficient of Ma1ecot can only vary from 0 to +1. Therefore, Wright's

F lends itself better to the study of population structure, for, cases of negative f, which is an estimate of F, are negative assortative mating, and heterosis. The equations given by Wright in 1965 are:

A f m = genotypic correlation

F = -st/[s+t-st] sand t = selection coefficients of the homozygotes

One of the effects of inbreeding on a population is that the variance within groups is reduced while the variance between groups is increased and, hence, the total variance is increased. The net effect of inbreed- ing is to decrease heterozygosity by the amount of 1/2N if the popu1a- tion is moderately large and the frequencies of the sexes are equal

(Wright, 1931). Wright in 1921 gave the genotypic frequencies expected if F is the average inbreeding coefficient: 93

Genotype Frequency

AA q2 (l-F) + qF

Aa 2q (l-q) (l-F) (1)

aa (1_q)2 (l-F) + (l-q)F where qF, (l-q)F are the inbred portions.

Li and Horvitz (1953) described several methods which are in use in estimating F, the inbreeding coefficient. The first method described was that in which the total proportion of heterozygotes observed was compared to the expected proportion derived from the

Hardy-Weinberg Law. The relevant equations are:

H f = 1 -­ H O where H is the Hardy-Weinberg frequency and H the frequency from O F equation 1, and where H is the observed frequency.

The second method involves arranging the zygotic proportions of

Wright's model into a gametic correlation matrix and taking its determinant, which is a simple function of F. This forms a matrix of observed numbers, and by taking its determinant and dividing by the th product marginal totals yield an estimate of the (k-1) power of F, where k equals the number of alleles. 94

The third method uses the proportion of alleles in homozygous condition, Z.. = (l-F)q~ + Fq., the proportion of A.A. individuals. l.l. l. l. l. l. Therefore, the proportion of A. in homozygous condition for all A. 's l. l. in the population is Z.. /q. and where the sum of these proportions l.l. l. is equal to

Zkk x --- = 1 + F(k-l) qk

Obviously, if F = 0 this equals 1 in a random mating population.

Therefore, the sample estimate of F is equal to

1 f = k-l

Another method employs the product moment correlation, for it can be shown that the product moment of the gametic correlation table for

Wright's model is F, no matter which numerical values are assigned to the alleles (Wright, 1948); however, in the case of small numbers the order of arrangement of the numbers may give different estimates.

Therefore, F may be estimated by the correlation coefficient

a (nl-n3)~ N(all- za13+ 33) - f = N(n +n )- (n -n )2 l 3 l 3 in the case of three alleles and 1, 0, and -1 assigned to A , A ' and l Z A respectively. 3 The maximum likelihood method in estimating F for two alleles without dominance is easy to calculate. The log likelihood equation is given by Yasuda (1968): 95

L = (a+b)ln p + (b+c)ln (l-p) + aln(p+a-ap) + bln(l-a)

+ cln (l-p + ap), a=f

The maximum likelihood score for a is

Therefore, the solution for a is

a = (4ac-b2 )/(2a+b)(2c+b)

General equations to expand this method to multiple alleles are given by Yasuda (1968). Li and Horvitz suggest reducing the number of alleles and solving the equation:

a .. (l-q.) l.l. l. (i~j) q. + e = Eaij = ni - aii l. where a.a. is the number of a. homozygotes, q. the frequency of the l. l. l. l. a. allele, a .. is half the number of heterozygotes and is equal to l. l.J F/(l-F). The estimate of F then is

2 Na .. - n~ ai'l. -N.9l. l.l. l. f n. N. = N. (l-N ,) = n. (N-n,) , l. = ql. ql. ql. l. l.

The last method to be described was the method preferred and used for all subsequent estimates of F, unless stated otherwise. This method was used because of its simplicity and built-in testing of the null hypothesis, F = O. The value of chi-square is calculated using, as observed, proportions in Wright's model; and, as expected numbers, 96

those from Hardy-Weinberg Equilibrium Law. The chi-square value

obtained is given by the following equation:

Therefore, an estimate of F, assuming panmixia and calculating the

expected numbers, gives

The test of the null hypothesis, a=O is the X2 test of significance itself. It should be noted that in the case of two alleles all esti­ mates obtained from these various methods are identical; except for the method of allele reduction which deals with multiple alleles only.

Having estimated the F value for a population, the question arises as to the power of the X2 distribution to detect significant departures from Hardy-Weinberg Equilibrium; hence, significant F values. Ward and Sing (1970) considered this very important problem

2 and derived formulas which expressed the relationship between X , F,

N, and S, which is the statistical power of the statistic. Using A which is a non-centrality parameter of the non-central X2 of Chapman

(1968), they found that A = nF 2 (k-l) which is equal to the central

2 X • The larger A is, the greater the power of the test for a sample with a determined a and a certain size. Therefore, the power of the test will increase as the square of F and will increase linearly with the number of alleles and sample size. Their tables show that for the

F values usually observed in natural populations, F = .10. The size of the sample must be larger than 103 to be at least 90% confident of 97 not committing a type II error, which is accepting an hypothesis F=O, when F10. Therefore, although estimates of F are relatively easy to calculate, the power to detect inbreeding in natural populations has been severely limited due to small sampling experiments. However, separate independent tests may be done on the same population to give the desired power, or, estimates of F for more than one loci may be taken to raise the power of the test. The latter, however, is probably too sensitive to correlations between loci in a population

(Ward and Sing, 1970). These results would, therefore, aid an experimenter to select a power value, B, which would then set lower limits to the sample sizes needed. Their results also aid an experimenter in telling him the degree of certainty that he is not committing a type I error.

11.3. Population Structure

Wahlund in 1928 described the effects of genotypic frequencies in a population with K subdivisions, which are endogamous panmictic groups. The mean and variance of the subdivision gene frequencies are

where q. is the frequency of gene a in the ith subdivision. Therefore, ~ the zygotic proportions of the total population will be as follows:

2 2 AA + 0 il q Aa 2pq' - 20 2 q aa q2 + 0 2 q 98

Hence, the result of a subdivided population is a decrease in the heterozygote proportions, with an increase in the homozygote propor­ tions. Wright in 1943 extended this to the interpretation of F and 2 showed that the 0 = p(l-p)a, where a is the mean inbreeding co­ efficient of the population.

Wright's island model is defined by him as a total population divided into subpopulations, each panmictic within itself, and exchanging a certain proportion of migrants drawn at random from the whole; with each subpopulation having a finite size and an effective number, defined by Crow and Kimura (1963) as "the size in an idealized population that would have the same amount of inbreeding or random genetic drift as the population under consideration." He defines the inbreeding coefficient in such a model as

F = (1-m)2/[2N-(2N-l)(1-m)2], where m = 1 - 12NF/[(2N-l)F+l].

Dobzhansky and Wright (1941) gave an approximation of F, F = 1/[4Nm+l], which is a satisfactory approximation if m is small. However, this model is not likely to be present in nature. As stated by Wright

(1943) the immigrants would not come from a random sample from the total population, but most probably from the adjacent subdivisions.

This, plus the difference in effective m for different loci necessitates a better model.

Wright (1943) developed a model he termed "isolation by distance."

Here it is assumed that there is a complete continuity of distribution, and interbreeding is restricted to small distances, due to only short range migration. He showed that the inbreeding coefficient of individuals in this model population is 99

F ~ I~-l t/[2N - I~-l t] where the ~t can be approximated by the equation,

,K-l Ll t

It should be noted at this time that the subdivided populations are of finite size and subject to the two major forces in finite populations, that of inbreeding and random genetic drift. The first occurs due to the presence of a certain probability that two genes are identical by descent in a finite population. The latter force occurs due to the sampling variance in the process of genetic trans- mission over generations.

Wright studied the effects of mutation, selection, effective size, and long range effects on this model. He found that mutation and long range migration tend to retard the rate of differentiation in large subdivisions, but not in small ones. He also found the amount of adaptive differentiation depended on the size of the sub- division or N, the size of the total population, if the degree of selection differed over the area covered by the population. If these differences are large enough no differentiation can be found in large subdivision sizes or N. However, if it is small, then small sub- divisions are found to have more non-adaptive differentiation.

The application of this Isolation by Distance model to natural populations was shown to be of no use by Spuhler (1961), when he 6 found that the effective population size was 2Xl0 , which is too large for any consanquineous marriages to take place. Morton and 100

Yasuda (1963) pointed out that the fallacy in this model is the

failure of Wright to consider non-randomness of consanguineous

marriages and to consider the leptokurtosis of human migration.

Kimura in 1953 proposed another model of population structure,

which he termed "stepping stone model." In a one-dimensional

stepping stone model (Figure 11.3.1), if migration is restricted to

adjacent colonies, the estimation of the mean inbreeding coefficient

between neighboring steps is at equilibrium,

f - 1+4Nb Il+2m/b O

which if b < < m simplifies to

f o = 1+4N 12mb

where m is the rate of local migration and b, the coefficient of

recall to equilibrium (~illlecot, 1950). Kimura and Weiss (1964) found

that in this model the correlation coefficient between steps is

approximated by

-A P r(p) ex: e , where p is the distance, and A is a constant which is equal to

12m /m . Malecot found a similar relation between the correlation oo l coefficient and x, the number of steps between them,

p(x) = (l+b/m - 1(1+b/m)2-l )x which, if b < < m reduces to

-xl2bm p(x) = e 101

Both Kimura and Weiss' model and Ma1ecot's model give identical results if b < < m. However, if b, the coefficient of recall, which is the sum of mutation, 1inera1ized stabilizing selection, and long range migration, is relatively large, these approximations will not hold.

In the one-dimensional continuous model of Ma1ecot (1967), the coefficient of kinship between neighbors at equilibrium is

1 1+400 • I2b ' where 0 is the density along the line and 0 is the standard deviation of the distance between birthplaces of parent and offspring. Therefore, the value of f(x) is

However, this model is useful only if the birthplace of parents and offspring are known, which is not the case in this study. Although

Malecot's elegant treatment of isolation by distance models may have great utility in human population structure where the migration matrices are defined, they are not useful in studying natural popu1a- tions of animals, in which the migration matrix is impossible to define without extensive trapping and tagging.

Kimura and Weiss (1964) went on to derive formulae for the correlation coefficient between steps in a two dimensional stepping stone model and found that p(x) decreased in proportion to

A~~ - -;- p/vp , 1 e 102 which shows explicitly that the correlation for the two dimensional case decreases more rapidly. For three dimensions, they found that

-/6moo - pip -1 ml r(p) = 7T e which says that the correlation falls off even more rapidly. If migration was not limited to adjacent colonies, the variance of the migration distance per generation may be substituted into the general equation. The correlation will then be

-/2moo r(k) = e m k in the one dimensional case. Therefore, it can be generally stated that when the migration of individuals is much less than the range of the species, random differentiation of gene frequencies results, as hypothesized by Wright's "isolation by distance model". Also, there is a decrease in correlation between divisions with distance, which decreases faster with the degree of dimensionality.

Maruyama (1970) used the one-dimensional stepping stone model of

Kimura and Weiss (1964), however, he considered the problem of finite length; that is, populations with a finite number of colonies rather than infinite. He therefore considered two models, one with absorbing boundaries, terminal colony receiving migrants at the rate of (1/2)m per generation, from the subterminal and the same amount from a hypothetical colony; the other with reflecting boundaries with the rate of migration from the subterminal colony being m. His simulation 103 studies, based on his derivations of the correlation coefficients, showed that as in the infinite model of Kimura and Weiss (1964) the correlation decreased with the number of intervening steps. He also found that the degree of correlation depended on the type of boundary, with those of the reflecting type having larger correla­ tions, which are expected since the rate of migration is twice that of an absorbing boundary.

Wright's hierarchical structure of populations is a model that lends itself to the study of population structure (Wright, 1925). He defines three F statistics, F 'F ' and FIT' F is defined as the IS ST IS probability that two homologous genes in I, an individual in sub­ population S, are derived from the same gene in a common ancestor.

F is the probability that two homologous genes, chosen at random ST from the subpopulation, are both descended from the same gene in the subpopulation. However, Wright's definition in terms of correlations is more useful since now the F values may take on negative values.

F is therefore the average correlation over all subpopulations IS between two uniting gametes relative to those of their own subdivision.

F is the correlation between random gametes within subdivisions, ST relative to the gametes of the total population. FIT then, is the correlation between gametes that unite to produce I relative to the gametes of the total population. The three F statistics are related in the following way:

FIT = FST + (l-FST)FIS or

FST = (FIT - FIS)/(l-FIS)' 104

F is necessarily positive, however, the others Frs' and F do ST rT not have to be. Frs can be negative if there is disassortative mating behavior or if there are heterotic effects at the locus under study.

Wright noted in the British Shorthorns that F 0.249, which he ST = ascribed to the low effective numbers in early years. Since the effective number is actually the harmonic mean of the varying numbers over generations, the bottlenecks will dominate over the larger popu- lation sizes. The relationship between F and immigration is ST F 1/(4Nm + 1). ST = Nei (1965) used the variance and covariance of gene frequencies to obtain measurements of random differentiation of subpopulations.

Hence, he found that the inbreeding coefficient of Wright (1951) was the most convenient way to measure differentiation. He found that an estimate of F is T

where

(fT - f s) Pj (l-Pj ) I-fS

= (fT - f S) PjPk I-fS and f is the inbreeding coefficient of the individuals relative to S the subpopulation and equal to He also obtained a formula for f ' f. ST 105

Therefore, all the f values of Wright's Hierarchical Model can be esti- mated by the mean inbreeding coefficient, and variances and covariances of gene frequencies. He, therefore, deduced that random differentia- tion of subpopulations leads to negative covariances and hence negative correlations of two allele frequencies, of a multiallelic locus,

P.Pk r = J (l-P.) (I-PI) J t

However, he did note that for multiple alleles r may be positive if f is different for different alleles. Li (1969) also found that for T multiple alleles the covariance does not necessarily have to be negative.

Nei used F as a measure of the degree of differentiation ST between subpopulations providing random differentiation has occurred.

These methods of Nei, Wright, Kimura and Weiss to measure population differentiation will be used to determine whether pigeon subpopula- tions are randomly differentiated or differentiated at all.

Since subdivisions are of finite size, we know that they are therefore subject to two related effects, inbreeding, and random gene frequency drift. The first effect was discussed earlier; however, the latter has yet to be discussed. Wright (1943) found that the gene frequency in subdivisions tended to vary about an equilibrium point q in a distribution curve, 106

Li (1955) showed that if the frequency of a gene is q in the parental population, then the frequency in the offspring population of N individuals will vary according to the binomial expansion, which is [peA) + a(a)]2N. The probability that q would have a certain value in the next generation is, if q. = j/2N, (Wright, 1931) ]. 2Nq. (2~) 2N-j qj 2N 2Np.q J J P = (2Nq.) p J J

Therefore, the gene frequency will vary as the result of random sampling in a finite population size. If oq = q.-q defines the J random deviation from the preceding generation, the variance of oq in one generation is thus defined by the equation,

06q = [q(1-q)]/2N.

Since gene frequencies at intermediate levels can reach 0 or 1 due to this random process, but those at 0 and 1 can never deviate to anything but 0 and 1, the ultimate fate of genes in any finite popu- lation will be fixation and extinction of genes after many generations.

This is true if no other pressures besides random drift is acting.

Although inbreeding and subdivision have similar effects, the variance of the random deviation oq in these two are different.

Hence, the sampling variance of q is

06q [q(l-q) (1+F)]/2N for inbreeding; for subdivision, it is

Therefore, it is shown that 06q is smaller by a proportion F for sub­ divisions; for populations with the same size and gene frequency. 107

However, if selection, such as heterotic selection is acting, this will tend to retard the rate of fixation (Robertson, 1962;

Reeve, 1955). Robertson found that the effects of heterozygote advantage depended on the equilibrium gene frequency where the

A effects are greatest for q about 0.50. Equilibrium frequencies out- side the range of 0.2 to 0.8 tended to accelerate the fixation rate; therefore suggesting that at the extremes, the effects of selection and random drift reinforce each other.

Ewens and Thomson (1970) considered the same problem of the interaction between heterozygote advantage and finite population size.

They found that at equilibrium values near 0.5, increasing (8 + 8 ) 1 2 increases the mean fixation time, no matter what the value of pis.

This is expected, since the tendency of a gene to reach its equilib- rium point, which is 0.5, will draw it away from the extreme fre- quencies, and thus retard mean fixation time. They also found, as

Robertson did, that equilibrium values lying outside the range of

0.2 to 0.8 usually led to a decrease in mean fixation time at small population sizes but increased it at a equal to 16 or greater, a = 2N(8 + 8 ), They explained this as due to the fact that when a l 2 values are greater than or equal to 16, the selection is large enough to hold the frequency of the gene near 0.2 and 0.8 therefore, an in- crease in mean fixation time is the result. Therefore, as in

Robertson's results the effects of heterosis were very much depen- dent on the equilibrium gene frequency.

The effect of linkage was found by Gill and Clemmer (1966) to increase the rate of inbreeding. However, Bogyo and Ting (1968) 108

showed that these results were due to randomization errors and con-

trarily, no effects of linkage, without selection, on inbreeding were

found. With truncation selection acting, the result will be an in-

creasing effect on the inbreeding coefficient. With both selection

and linkage acting, an interaction is visualized in later generations,

by not only increasing F.. but also increasing its variance where F.. J.J J.J is the probability of two genes being of identical descent.

These interactions between forces will be used to explain some

of the seemingly variable results observed in these data on pigeon

populations. As in the case for most studies of natural populations,

the effects of subdivision and inbreeding may be masked, due to

counteracting forces, thus giving F values of O. Workman, in 1969,

concluded that in any natural population where different forces are

continually opposing each other, such as heterozygote advantage and

inbreeding, the net result will be an estimation of F not very dif-

ferent from zero; therefore, pointing out the insensitivity of mea-

suring significant departures from F = O. Similarly, it would also be expected for F to be different for different loci, since they must not be acted upon equally by all forces.

11.4. Analysis of Population Structure in Pigeon Populations

The one-dimensional stepping stone model of Kimura and Weiss

(1964) was the model chosen; however, it was used only as an approxi- mation as in Nei and Imaizumi (1966). The first suggestion of random differentiation of pigeon subpopulations was in the tests of homo- geneity of gene frequencies of the electrophoretic loci between loca-

tions; all of which showed significant heterogeneity (Tables 7.3.25, 109

8.3.18, 9.3.15). The p values for all loci were less than 0.005.

The X2 for the shank gene frequencies also showed significant heterogeneity between localities. The X2 value was 44.38 with 5 degrees of freedom, which again gives a p value of less than 0.005.

Therefore, an analysis was performed to test whether these differen- tiations occurred at random.

Nei and Imaizumi (1965) used an approximate equation to obtain

2 2 2 2 the mean sampling variance, 0 • where 0 = 0 + 0 This approxi- op' Po pop' mation was

where n is the mean number of observations for localities. This approximation was used throughout the study. Therefore, by subtract- ing this sampling variance from the observed variance, we have the true genetic variance, and if this is much greater than 06p' then this suggests random variation. Tables 11.4.1, 11.4.2, and 11.4.3 give the genetic and sampling variances of the transferrin, 6PGD, and APR loci respectively. It is noted that all genetic variances are much larger than the sampling variances, therefore, suggesting random differentia- tion for these alleles.

Another method of determining differentiation is to look at f ' ST for, this statistic measures the correlation of two random gametes relative to the total population and hence is a measure of differ- entiation between populations. The values of f obtained are .0081, ST .0001, and .0006, for the transferrin, 6PGD, and APH loci respectively.

The value for the transferrin locus is considerably larger than the 110

values for the other loci. This suggests that the frequencies of the

transferrin locus at the different localities have undergone some

random differentiation.

Another test for random differentiation of gene frequencies is

that of the decrease in correlation of gene frequencies with distance.

According to Kimura and Weiss (1964), Malecot (1950), and Maruyama

(1970), the correlation should decrease with the number of steps in a

one dimensional model. Table 11.4.4. gives the frequencies and lo­

calities involved, plus the correlation coefficients for the trans­

ferrin locus. Steps in this model mean the number of steps away from

one locality to the other. These coefficients support the finding of

differentiation between localities in the previous analysis. There

is a definite decrease in correlation with distance (Figure 11.4.1).

The correlations for the 6PGD and APH loci, however, failed to show any

correlation with distance (Tables 11.4.5. and 11.4.6.) (Figures 11.4.2.

and 11.4.3.). This lack of correlation may be due to large sampling

error, since the number of localities is small, or may be due to non­

random differentiation.

Next, a simple linear regression analysis was done to check for

any significant gene frequency clines. The regression coefficients were 0.014, 0.002, and 0.022 for the transferrin, 6PGD, and APH loci

respectively. The t value for the APH locus was 5.50 and was signifi­

cant at the 0.01 level. The correlation coefficient was also

significant at the 0.05 level, r=0.692 (Table 11.4.7). The regression equation is

Y = 0.842 + .022X, III where distance is the independent variable and gene frequency the dependent variable. Therefore, there does seem to be a gene frequency cline from East to West, with the frequency of the B gene increasing from East to West. The others failed to show any signifi­ cant clines.

The genetic variance for the shank phenotype is almost 7 times larger than the sampling variance; therefore, suggesting random differentiation of frequencies (Table 11.4.8). The f is also quite ST large .0010. However, the correlation between gene frequencies and distance, failed to show any decrease in correlation with distance

(Table 11.4.9) (Figure 11.4.4). However, these estimates are subject to large sampling errors. The regression of gene frequency on dis­ tance was significant at the .005 level (Table 11.4.10). The regression equation is

Y = .345 + .042X.

The regression coefficient suggests a significant gene frequency cline from East to West, with the feathered gene increasing from East to

West.

The pattern allele frequencies also showed marked heterogeneity between localities. Nei (1965) showed that for multiple alleles, if random variation has occurred, the covariance of the allele frequencies will be negative and therefore the correlation coefficient will be T negative, -.0033, -.0071, and -.0026 for the covariances between C T and C, C and c, and C and c, respectively. The correlation coefficients between distance and the different pattern gene frequencies 112 failed to show a decrease in correlation over distance (Table 11.4.12)

(Figure 11.4.5); however, these values are subject to large sampling errors. The correlations of two allele frequencies showed no signi­ ficant differences from the expected r's; therefore again suggesting random differentiation (Table 11.4.13).

The f values calculated from the various allele variances and ST covariances are shown in Table 11.4.11. All of the values are very high, again suggesting a high degree of differentiation between lo­ calities. The f value for q, however, is less than the others. ST This is due to the stability of the C gene in all populations observed.

This was referred to in section 10.3. with the major differentiations T occurring in the C and c alleles. Mainardi (1964) suggests that the checkered allele is maintained by sexual selection in urban areas; while the c gene is maintained by apostatic selection, due to its selective advantage on rocks and in flight in the wild. Mainardi's hypothesis is based on the premise that dominant genes are favored by the assortative mating system. This will be discussed in the section on assortative mating. Therefore, the stability of the C frequency may be due to the balancing of the selective forces and the effects of subdivision. No frequency clines were found for any of the pattern alleles.

From these analyses it can be concluded that the differences found in the allele frequencies for the various loci are mainly due to random genetic drift. There also exists marked differentiation between localities. Two gene frequency clines were found for the APH,

B gene, and shank feathered allele, whether these clines are stable 113 or transient is, of course, unascertainable from these data. However, stable clines have certain characteristics, one of which is the essentiality of different selective conditions between the localities.

There is no evidence, however, that these differences are present.

Therefore, it must be assumed that these clines are transient, and are due to migration or selective accidents. The data reveal that there is migration from the Zoo to St. Andrew's and Bethel. One bird from St. Andrew's belfry was recovered with a Zoo band on it and two birds with Zoo bands were recaptured at Bethel. Therefore, the data do support a theory of limited migration and gene diffusion. It is also evident from the data that the two localities farthest apart, the Zoo and Pier 32, are significantly different in gene frequencies for all of the loci observed.

The Zoo population exhibited a deficiency of heterozygotes in three loci, transferrin, APH, and shank. The inbreeding coefficients computed with a simultaneous estimate of p and q using Yasuda's G-type program (Yasuda, 1968), were .208 ± .043, for transferrin; .189 ± .048 for APH; and, .192 ± .079 for the shank locus. None of these esti­ mates of F are significantly different from each other; therefore, showing the consistency of estimation for all three loci. The 6PGD locus, however, failed to show a significant F, in fact, the follow­ ing negative estimate of F was found: -.050 ± .038, thereby suggest­ ing heterotic effects at the 6PGD locus. Ewens and Johnson (1970) showed in their computer simulation studies that once p reaches its equilibrium frequency, m, it will remain there for very long periods of time, longer than the time required for p to go to m. Therefore, 114 even with inbreeding in the population, equilibrium frequencies can be maintained over long periods of time. However, the frequencies have been shown to have changed significantly from year to year

(Table 11.4.12); therefore, suggesting that they are not in equili­ brium. Jain and Workman (1967) found that F- (X-I) + f(X+l) where - (X+l) + f(X-l) the relative fitnesses of the AA, AB, and BB genotypes are X, 1, and

X, respectively, F will be zero if F = (l-X)/(l+X); therefore, for an f of .20, it would take an X of about .82 or selective disadvantage of about 18% for homozygotes. This is obviously much too high com- pared to previous measurements of heterozygote advantage in other organisms. However, if a system like a balanced lethal system, though not as strong a disadvantage, were present, then an 18% dis- advantage is within reason. Workman in 1969 showed that if the gene frequency is not in equilibrium it would be difficult to attach much meaning to any estimate of F. Hence, this problem is left unanswered for the major part because of the complicated interactions of systematic and non-systematic forces on F.

11.5. Assortative Mating and Inbreeding

Assortative mating is the mating between individuals of similar phenotypes. If the population is large and there exists no sub- divisions, this similarity may be through non-random mating rather than by genetic similarity by common ancestry. However, with sub- division and finite size of populations there will generally be a greater phenotypic similarity between individuals of common ancestry; therefore, the outcome of assortative mating will be similar to those 115 of inbreeding. Since there is a large amount of differentiation among the populations studied due to subdivision, the latter situation is applicable, and therefore will be discussed in this light.

Since phenotypic similarity usually is accompanied by genetic similarity the effects of inbreeding will come into play, and there- fore there will be an increase in average homozygosity and an increase in total population variance. Therefore, qualitatively the effects of assortative mating should be very similar. However, quantitatively they do differ. Assortative mating leads to less an increase in homozygosity and a larger increase in variance for multifactorial traits (Wright, 1921). This larger increase is due to the causation of gametic phase disequilibrium, which is obvious, since assortative mating increases the frequencies of the extreme phenotypes. 0'Dona1d

(1960) found that complete assortative mating took 1.388 times as long as sib-mating and 4.377 times as long se1fing to achieve the same degree of homozygosity; therefore, the rate of fixation is slower.

However, in the case of partial assortative mating and the phenotypes

AA, Aa, aa are all different, the results are somewhat different. If we assume m, the correlation between mates to be constant in each generation; and p = D + 1/2H, q = R + 1/2H, and D + H + R = 1, it is therefore evident that this type of mating system is equivalent to inbreeding in a subdivided population where k = Fpq, and therefore, 2k 2F m = --= Li (1955) derived that at equilibrium, D = p2 + k, pq+k 1+f" H 2pq - 1k, and R = q2 + k, where k is a positive fraction. Hence, with most systems of partial assortative mating, the population will reach an equilibrium, thereby retaining a certain amount of 116 heterozygosity. Therefore, complete assortative mating will lead to complete homozygosis, while partial assortative mating will lead to an equilibrium condition where a certain amount of heterozygosis will be maintained in a population. As is seen in Li's example, even a mod­ erate amount of assortative mating will lead to a small degree of fixation. O'Donald (1960) found that this was true whether dominance or co-dominance prevailed.

Scudo and Karlin (1969), O'Donald (1960), and Wright (1921) all agree that complete assortative mating with dominance leads to fixa­ tion and no stable equilibrium can be found. Contrary, however to

Mainardi, who believed that assortative mating favors dominant alleles,

Kalmus and Smith (1966) fourrd that if relatively few dominant pheno­ types are introduced into an already assortative mating population of recessive phenotypes, the majority will remain recessive. This same result was found by Sieger and Dixon (1970), who discovered that the ultimate outcome of a new allele, introduced by dispersion, depended on the population size and imprinting coefficient, when the migration rate remains the same. They used a linear stepping stone model of popula­ tion structure and varied the number of mated pairs of birds in five populations, ranging from 125,000 to 15,000, where population number two had all AA phenotypes and the other localities had all aa pheno­ types. Population 2 also had the largest population size. With this model, they found that if the intervening population was of moderate size the A allele would be eliminated, since the movement of A alleles into the terminal population of aa individuals, would be limited.

There did exist a critical amount of inbreeding which would have allowed 117

a stable equilibrium; this amount being dependent on the population

size, and rate of effective migration. The example given by Sieger

and Dixon showed that with a migration rate of about 2% and a popula­

tion size of 15,000 pairs, a stable equilibrium would result if the

degree of assortative mating was 0.50.

It is evident that for a new gene to be established in a randomly

mating population, heterozygote advantage is a necessary requirement.

This, however, was not the case in an assortatively mating population

(Parson,1962). Parson found that the increase in frequency, in a new

gene in a positively assortatively mating population, depends on the

degree of assortative mating found and the dominance relationship be­

tween the new gene and the established gene. In the case of random

mating, in order for the dominant gent to increase, h must be greater

than b; if the recessive gene is introduced, in order for it to in­

crease, h must be greater than a; where a, h, and b are the fitness values for the AA, Aa, and aa genotypes respectively. However, in the

case of complete assortative mating, for gene A to increase, h must be

greater than 2b or a greater than 2b; and for gene a to increase h must

be greater than a or b greater than a. In the case of no dominance and

if gene a is introduced into a population of all AA phenotypes, h must be greater than 2a and b greater than a; in order for gene a to increasa

While for gene A to increase, h must be greater than 2b or a greater

than b. Therefore, heterozygote advantage is not necessary for the fre­ quency of a new gene to increase in a population that practices partial assortative mating, since for partial assortative mating the restrictions will be less severe than those for pure assortative mating. 118

These results show that assortative mating based on two alleles

may not lead to complete homozygosis, but may form a stable equilibrium

between the alleles, providing all other conditions are conducive to

formation of an equilibrium.

It is also seen that only pure assortative mating with no selec-

tion will not lead to an increase or decrease in frequencies. In order

for a stable equilibrium to be found in a population with selection and

complete assortative mating, the following relationships must hold; for h h-a dominance b > h-b and for no dominance h > 2a, 2b (Falk, 1970 and

Parsons,1962). With partial assortative mating and dominance, Falk

(1970) found that computer simulation studies showed that for a stable

equilibrium to be formed, heterozygote superiority must be present.

Leaving the theoretical portion of the introduction into assorta-

tive mating, we shall cite some species in which assortative mating

occurs. Pearson and Lee (1903) found evidence for assortative mating

in man for certain physical characteristics such as height; Cooch and

Beardmore (1959) have described this behavior in the Blue Snow Goose,

Chen caerulescens; O'Donald (1959) found it in the Arctic Skua,

Stercorius parasiticus; and Parsons (1965) found assortative mating

for sterno-pleural chaetae number in Drosophila melanogaster. These

being just a few examples of assortative mating found in Nature.

The basis of, some forms of assortative mating is a behavioral mechanism called imprinting. That is, the attachment of young to

parental characteristics. Heinroth in 1910 was the first to describe

this behavior and later Lorenz (1937) went on to study this behavior,

and develop his theory of the irreversibility of imprinting. Lorenz 119

also theorized that this period of imprinting was relatively short in

the young; and, that the young would direct their sexual behavior at whatever they were imprinted on. Craig (1908) was the first to propose

that the physical characteristics of the parents and their behavior might affect mate selection in the young pigeon.

Derek Goodwin in 1958 found that Columba livia young did imprint on their parents. He found that of 732 wild pairs observed a majority of the pairs were of like phenotype. He also found that raising two young wild type rock pigeons with domestic birds of different colors resulted in the sexual advances of these young being directed toward birds of the same color with which they were reared, rather than towards birds of their own color. Moore (1965) found that red pigeons hatched and reared by one white and one black foster parents, chose only black and white pigeons for mates. Warriner et al. (1963) found that the color of parents strongly influenced mate selection in males, while Kerfoot (1964) showed a similar situation prevails in the female pigeons. Therefore, mate selection in males and females is influenced by imprinting; however, mate selection is male dominant, which means that male preferences determine mate selection.

Mainardi (1964) based on the assumption that sexual selection, assortative mating in which the fitness of phenotypes differ, favors dominant alleles; therefore, hypothesized that imprinting aided the evolution of the different pattern alleles in Columba livia, due to the difference in selection between two ecological niches. In the wild the blue pattern would be favored, due to its camouflaging effect against a rock background, or blue sky background. However, in the 120 urban environment, the sexual selection process overcomes the lowered selective advantage of the blue pattern, and, hence the dominant pattern alleles drives the blue pattern allele to extinction. However,

Seiger (1967) and Kalmus and Smith (1966) found this not to be necessarily true, since the presence of the dominant allele may not lead to the extinction of the recessive allele. Its dominance, however, will lessen the restriction for it to increase its frequency.

Therefore, it can be safely concluded that imprinting and positive assortative mating does occur in pigeons, and, these mechanisms may play an important role in the evolution of the species. The following section will attempt to find evidence for positive assortative mating in the populations studied.

11.6. Analysis of Assortative Mating in the Zoo Population

In June of 1972 recordings of 104 mated pairs of pigeons which resided in the Zoo and adjacent areas were made. The basis for record­ ing two pigeons as a pair was through various aspects of courtship behavior, such as copulation, courtship feeding, courtship preening, and nesting behavior. Plumage phenotype frequencies were computed for all major colors (Table 11.6.1). The frequencies were then grouped into two, white plumage color frequency and AOC, or Any Other Color, frequency.

Table 11.6.2. shows that there is a significant excess of white

X white matings. The X2 value obtained was 13.62 with 2 degrees of freedom. The p value was less than 0.005. This, therefore suggests a significant amount of positive assortative mating present in the 121

population. The consequences of this are two fold. First of all, it was shown by Li (1955), that the results of assortative mating at

equilibrium is identical to inbreeding in subdivided populations, which is evident since assortative mating results in a subdivided popu­

lation. It was also shown, in the previous section, that positive

assortative mating does have similar consequences as inbreeding, how­

ever, it was much slower in rate towards homozygosity. Therefore, it

is concluded that assortative mating in the zoo population added to the

amount of inbreeding found in this population, which helped to explain

the high F values estimated. Secondly, assortative mating in the zoo

population should lead also to a large degree of differentiation

between the zoo and the other localities; which was found and dis­

cussed in the section on population structure.

It should be noted that in the other localities no recordings of mated pairs were ma.de. However, all other localities showed higher

frequencies of checkered patterns than the blue-barred patterns.

Therefore, as Mainardi (1964) observed, high frequencies of checkered

patterns were found in urban populations. However, the frequency of

the blue pattern was still quite high, with the differences between

localities being due to random drift rather than selective disadvantage of the c allele. Furthermore, assortative mating would be more diffi­ cult to find in these populations for the simple reason that blue pigeons, black checkers and blue checkers can come from black checkered pigeons; and blue checkers and blues from blue checkered pigeons. 122

11.7. Homing Behavior and Its Consequences

In the section on the history of the pigeon, it was noted that the homing behavior, the ability to return to a particular territory, of the pigeon was recognized as early as the rein of King Cyrus the

Great, and probably earlier. Therefore, there should be no need to verify the presence of this behavior in Columba livia. However, the consequences of this behavior must be discussed.

It is quite evident that homing behavior should enhance the amount of assortative mating and inbreeding found in a population, for, the ability of a pigeon to home to a particular breeding area would result in birds not only phenotypically similar but also genetically similar to each other. This therefore, leads to an increase in assortative mating and inbreeding within a population. This is obviously the situation in the zoo population, for, the birds have been free-flying for over 36 years and if there was not a tendency for birds bred in the zoo to return to the zoo, the high F values estimated would not be present. This should not be construed to mean that there is no migration occurring, for, it is obvious that there is considerable migration into the Zoo population, due to the large number of colored birds in the population. The predominance of white colored birds also lends evidence for the presence of homing behavior and assortative mating; for the frequency of the white gene should be much less, after over 36 years of migration into the population.

Homing behavior should also lead to greater differentiation between localities. This increase in differentiation is due to the decrease in migration between localities, no matter how close these 123 localities are. The racing homers will home to lofts a few feet away.

Therefore, this behavior seems to be quite strong in Columba 1ivia and reaches its maximum in racing homers. Hence, this strong tendency to return to place of birth, and assortative mating should lead to marked differentiation in different populations of pigeons. It should be pointed out that these two behaviors are not mutually exclusive; for there is a greater probability of mating with an individual, with characteristics upon which one is imprinted, if one returns to the same population.

An example besides the present study which illustrates the dif­ ferentiating effects of homing behavior is the paper by Milne and

Robertson (1965). In one of the breeding sites of the eider duck,

Somateria ~. mo11issima, they found that two-thirds of the population were migratory while the other third were non-migrant and remained in

Forvie all year round, the migrant portion migrated South in the autumn and returned to Forvie in the Spring for breeding. The non­ migrant birds bred on the West side of the moor while the migratory birds nested on the East side. It should also be noted that the pro­ cess of pair formation occurred in the non-migrant population prior to the return of the migratory population. In addition 95% of the migratory birds were paired before returning to the breeding grounds.

Milne and Robertson (1965) found a significant difference in frequencies between these two populations, with respect to two polymorphic alleles found in egg white post-albumins. Therefore, this suggested, as from the observations on migratory and breeding biology, that these populations were to some extent reproductively isolated. 124

11.8. Evolutionary Consequences of Population Structure and

Assortative Mating

Wright in 1966 cited five cases and their values in evolutionary events. The first two cases involved populations too small and too large for any evolution to take place, due to static conditions.

These static conditions are random fixation in small populations, and equilibrium conditions in large populations. The third case is when changed conditions cause "slight and reversible" shifts of gene frequencies to new adaptive peaks. The fourth is the case of popu- lations of intermediate size, where there is continual random shifting of gene frequencies, and alteration of selection coefficients, there- fore, leading to relatively rapid and irreversible changes. However, the absolute rate is slow, due to its dependence on mutation pressure.

The last case is that of a large but subdivided population, quoting

Wright (1966):

"Finally in a large but subdivided population, there is continually shifting differentiation among the local races, even under uniform static conditions, which though intergroup selection, brings about indefinitely continuing irreversible, adaptive and much more rapid evolution of the species as a whole."

Therefore, the last case seems to be the most favorable for the species.

In large populations many different alleles will be present com- pared to two or a few alleles in the fourth case. Random drift in the fifth case would lead to marked differentiation between subdivisions, however, these subdivisions must not be too restrictive such as to prevent diffusion from superior adaptive centers. Wright considers the most favorable case to consist of a three step process: 125

(1) "relatively rapid, random local differentiation," if migration is relatively small (1/2N), (2) this leads to the "crossing of two­ factor saddles and mass selection toward control by the higher selective peaks, and (3) "interlocality selection on the basis of differential population growth and diffusion." The first phase was termed "polyallelic random drift" by Wright (1966). Therefore, in a largely subdivided population, a large amount of genetic variation may be maintained without their having any selective advantage, as is required in mutation theory. Therefore, with this storage of genetic information, a greater number of genotypes will be available for natural selection to experiment with, hence leading to new adaptive peaks and evolution. The above theory appears to explain the rapid speciation of the Hawaiian Drosopholids, which Carson (1968) has incorporated into his theory of "population flush", with the excep­ tion of having complete isolation at a point in time to allow speciation to complete its course of reproductive isolation.

It has been shown in the previous sections that assortative mating will tend to enhance the first phase of the fifth case; there­ fore, leading to a more rapid local differentiation, than if random mating was occurring. It has been suggested by Millicent and Thoday

(1960) and Thoday and Boam (1959) that divergence would proceed without isolation, in a disruptive selection scheme with assortative mating; for, even 25% gene flow between populations led to a large divergence between populations. Therefore the addition of assorta­ tive mating within the mating scheme would lead to very marked dif­ ferentiation. As theorized by O'Donald (1960) and Kalmus and Smith 126

(1966) complete assortative mating would lead to reproductive isola­ tion between populations which were segregating for two alleles.

Hence, since speciation bears a requirement of reproductive isolation, we can conclude that assortative mating can be one of these repro­ ductively isolating mechansims. For if Dobzansky (1951) is right, that a minimum of two complimentary gene differences are necessary for speciation, it is evident that from the simulation studies of the above experimenters a single homozygous allele difference could bring about isolation, if there is strong assortative mating present.

However, sympatric speciation would ensue only if assortative mating is absolute, which is highly improbable in Nature, who is very seldom precise.

Inbreeding within subdivisions also leads to differentiation, for, one of the consequences of subdivision is inbreeding. Therefore, a combination of assortative mating and inbreeding, both of which enhance the effects of subdivision in a large population, will result in a very rapid rate of differentiation between subpopulations, In­ breeding will also aid in the fixation of advantageous gene complexes at a new selective peak.

Hence, with assortative mating, inbreeding, homing behavior and therefore subdivision of a population, a large amount of differentia­ tion will be the result. This will provide the first step needed in a model (case 5) which will lead to intergroup selection and therefore a continuing irreversible and rapid evolution of the species. One can see that these mechanisms will apply especially to polytypic speciation. 127

In the evolution of Columba 1ivia it can be hypothesized that

these mechanisms of assortative mating, inbreeding, homing behavior and subdivision of populations held extremely important roles in the evolution of this species, as suggested by Goodwin (1958), Mainardi

(1964), and Ka1ums and Smith (1966). In the domestication of pigeons, many different breeds of pigeons have been derived from the Rock

Pigeon; for examples see Levi (1965), The Encyclopedia of Pigeon

Breeds. These domestic breeds would have had to withstand large amounts of inbreeding, due to highly selective programs, small N, e assortative mating, as well as systems of matings between close re1a- tives. As in mice (DeFries, 1970), it can be suggested that the pigeon is accustomed to large amounts of inbreeding in the wild, due to the above mechanisms, and therefore can withstand the large amounts of inbreeding built-up in highly inbred lines.

It was shown by O'Dona1d (1960) and Parsons (1962) that hetero- zygote advantage was not a requirement for the establishment of an equilibrium between two alleles in a population. Seiger (1967) and

Sieger and Dixon (1970) found that incomplete assortative mating would lead to balanced po1ymorphisms, independent of the phenomena of heterosis. Therefore, assortative mating would maintain a large array of alleles in a population. These alleles may be for the trait (color), associated traits (behavior), and linked loci. Hence, contrary to randomly mating populations, alleles can reach an equilibrium state without heterozygote advantage, and may therefore lessen the load on the population. For if only heterosis would maintain variation, the load would be too great, as hypothesized by the genetic load theory. 128

Millicent and Thoday (1960) provided evidence for the establishment of polymorphisms in a population undergoing disruptive selection, if given appropriate genetic variation. Therefore, assortative mating could certainly be a means for maintainance of variability in multi­ allelic systems. This, coupled with population subdivision and its conduciveness to polyallelic random drift, may therefore be very important mechanisms in which variability is maintained in a species.

11.9. Summary

Some basic tenets of Hardy-Weinberg Equilibrium Law were dis­ cussed and how departures from these basic tenets result in an in­ crease in homozygosis. Discussion of the different estimates of F was introduced, along with the power to estimate F. The equality of all estimates of F, from a locus with two co-dominant alleles was discussed. A discussion of Ward and Sing's results on the power to estimate F revealed the large probability in committing a type II error in estimating small values of F.

Wahlund's principle was introduced and various models of popu­ lation structure were described. The different estimations of F were also given for each model. The consequences of subdivision were discussed, with the decrease in correlation with distance pointed out as one of the main consequences. The results of Nei's 1965 paper were also discussed. The hierarchical structure model of Wright and its significance was referred to, along with Nei's application of the

F statistics derived by Wright. 129

The consequences of random drift were explained in terms of the variance of the sampling variance. The interactions of random drift and heterozygote advantage in finite populations were discussed.

Other forces interacting with random drift were also discussed. Also the insensitivity of present methods to determine F values other than 0 was pointed out, since the interaction of different forces, which will have both positive and negative effects, would have a net result of F being not very far from zero.

The stepping stone model of Kimura and Weiss was approximated and analyses similar to Nei and Imaizumi (1966) was done. The results revealed that all genetic variances were considerably larger than their sampling variances. A decrease in correlation with distance was found for the transferrin locus but not for the other loci. This difference was discussed in terms of heterosis, gene frequency clines and large sampling error. No significant differences were found between the expected correlation coefficients and the computed correlation coefficients between two alleles of the pattern locus. The covar- iances of these alleles were all negative. The f values for the ST transferrin locus and pattern alleles were also found to be quite large. It was therefore concluded that the marked differentiation bound between localities is due mainly to random genetic drift. Gene frequency clines for the APH locus and shank locus were also found.

Inbreeding coefficients were estimated for all loci. All loci except the 6PGD locus in the zoo p~pulation revealed significant positive F values. The failure oi the 6PGD locus to show a significant F was discussed. 130

The consequences of assortative mating in a population were dis­ cussed. The similarities between assortative mating and inbreeding were pointed out as well as their differences. Partial assortative mating resulting in equilibrium conditions was shown by several experi­ menters to be theoretically possible. The relationship between assor­ tative mating and imprinting was discussed, along with evidences for these mechanisms in pigeons. Analysis of mated pairs in the zoo popu­ lation revealed a significant assortative mating pattern in the white population. The consequences of another behavior, present in pigeons, homing behavior were also discussed in terms of enhancing local dif­ ferentiation.

The last section discussed the evolutionary consequences of all the above forces not considered in the Hardy-Weinberg Equilibrium Law.

The conclusion reached was that assortative mating, inbreeding, and homing behavior greatly enhance population subdivision and therefore, provide the first step in Wright's polyallelic random drift model of evolution. That is, they provide a means to attain relatively rapid random local differentiation which is the initial spark from which crosses to new adaptive peaks and intergroup selection can arise, and hence a more rapid rate of evolution. 131

XII. Electrophoretic Variability in a Single Pigeon Population

12.1. Introduction

Lewontin and Hubby (1966) estimated the amount of allelic varia­ tion found in natural populations of Drosophila pseudoobscura to be about 30% of all electrophoretic loci. Since this study Harris (1966) has observed that 6 out of 18 electrophoretic loci in man are poly­ morphic, therefore, again leading to an estimate of 30%. After these initial studies, many studies have followed to estimate the amount of polymorphisms found in other organisms, and therefore the initial part of this study was to find the number of electrophoretic loci, that are polymorphic in a pigeon population.

12.2. Materials and Methods

All electrophoresis was done on a horizontal electrophoresis starch or acrylamide gel apparatus. The methodology generally follows that of Smithies (1955), therefore, reference to this article will be sufficient in describing this technique. The following are buffers and electrophoretic conditions used for the various proteins and enzymes. In most cases only the conditions which gave the best resolu­ tion are described.

1. Pre-albumins

Ashton System

Gel Buffer - pH 8.1

.008M - Tris-hydroxymethyl-aminomethane

.040M - Citric Acid

420 mls. Gel Buffer + 75 MIs. Electrolyte Buffer 132

Electrolyte Buffer - pH 7.9

.031M Lithium Hydroxide

.194M Boric Acid

Starch Concentration - 13.5%

2. Kristjansson Pre-albumin System (Kristjansson, 1963)

Gel Buffer - pH 7.5

0.0047 Citric Acid

0.14M Tris-hydroxymethyl-aminomethane

Electrolyte - pH 8.7

0.3M Boric Acid

O.lM Sodium Hydroxide

Starch Concentration - l3.5A

The electrophoretic condition~l for the Ashton System were 300 volts for 3 1/2 hours. The size of the gel trays were 25.5 em. long,

12.9 em. wide, and 3 mID. deep. The distances from insert line to the wicks were identical to those described for transferrin. The electrophoretic conditions for the Kristjansson system were an initial voltage of l65V for 15 minutes; inserts removed, 165V for an additional 15 minutes, and then 350V until the borate line was 13 em. from the insert line. The gel tray used was 22.5 em. long, 10.2 em. wide, and 6 mID. in depth. The insert line was 5.8 em. from one end.

The wicks were placed 2.2 em. from the edge of the gel at the insert end and 2.0 em. at the other end. The gels were stained with the same nigrosine stain as in the transferrin system. 133

Albumen

Albumen System - pH 5.6

Gel and Electrolyte Buffers

4.2 gm. Citric Acid

5.0 gm. Tris-hydroxymethyl-aminomethane

Gel Concentration - 13%

Gel Tray - same as in Transferrin

Voltage - 400 V

Current - 50 rnA

Inserts - Whatman No.1

Stain:

Same as in Transferrin

Bands are dark blue on light blue background.

Leucine Aminopeptidase

Law and Munro System (1965)

Gel Buffer - pH 8.6

.04 M- Tris-hydroxymethyl-aminomethane

.003 M- Boric Acid

Electrolyte same as above

300 V were applied to the system with usually 10 rnA per gel current.

The gel dimensions and relevant distances were described in the APH lnaterials and methods section. 70 grams of starch per 500 mls. were used. 134

Stain:

40 mg L-leucyl-S-napthylamide-HCl

50 mg Black K Salt

100 rnIs 0.2M Tris Maleate Buffer, pH 6.5

Purple bands on orange background.

Amylase

Ashton System

Gel Buffer:

.008 M Tris-hydro~yrnethyl-aminomethane

.040 M Citric Acid

+ 50 mls. Electrolyte

Electrolyte

.031 M Lithium Hydroxide

.194 M Boric Acid

Gel - 7% Cyanogum

Inserts - Whatman No.1

The dimensions of the gel plat were the same as above, as well as, the relevant distances, except the gel depth was 2 rom. 400 volts were applied to this system with a current of 25 rnA per gel for 6 hours.

Stain:

Incubate for 45 minutes in 1% soluble starch solution.

Wash excess starch off with distilled H 0. Place gel 2 into an Iodine solution made up of 135

5 bms KI

1 gm 1 2

5 gm NaCl

Bands are transparent in a dark blue background.

Lactic Dehydrogenase

Poulik Buffer System - Poulik (1957)

Gel Buffer

.076 M Tris-hydroxumethyl-aminomethane

.005 M Citric Acid

Electrolyte Buffer

0.3 M. Boric Acid

0.05 M Sodium Hydroxide

10% Sodium Chloride in anodic tray.

Gel Concentration - 13%

Gel Tray - Identical to that for 6PGD

Voltage - 6 volts per centimeter ~ l50V

Current - 25 rnA

Inserts - Whatman No. 3

Stain:

12 mls. 0.5 M DL-Lactate

10 mg. DPN

20 mg. Nitroblue Tetrazolium

5 mg Phenazine Methosulfate

100 mls Tris-Hel pH 8.2 136

Incubate at 37.50°C for 2 to 4 hours. Bands are blue on white background.

Malate Dehydrogenase ­

Gel Buffer - pH 7.0

66.7 mls Electrolyte Buffer diluted to 1 liter

Electrolyte Buffer - pH 7.0

0.155 M Tris-hydroxymethyl-aminomethane

0.043 M Citric Acid

Gel Concentration - 13%

Gel Tran - same as 6 PGD gel tray

Voltage - 100V

Current - 10 rnA per gel

Inserts - Whatman no. 3

Stain:

50 mg. DPN

30 mg. Nitroblue Tetrazolium

2 mg. Phanazine Methosulfate

15 mls. Tris-HCl pH 7.1

70 mls H20 5 mls 7M Na DL-Malate, ph 7.0

Incubate at 37°C for 2 to 4 hours. Bands are blue on white background. 137

Glucose-6-Phosphate Dehydrogenase

Cooper System (Cooper 1969)

Stock Buffer

0.9 M Tris-hydroxymethyl-aminomethane

0.5 M Boric Acid

0.1 M Etylenediamine-tetracetic acid

Gel Buffer

24 mls. Stock buffer

576 mls. H 0 2

Electrolyte Buffer

80 mls. Stock buffer

820 mls. H 0 2

Gel Concentration - 13%

Gel Tray - same as above

Voltage 100 V

Current - 10 rnA per gel

Inserts - Whatman No. 3

Stain:

10 mg. TPN

20 mg Nitroblue Tetrzolium

3 mg Phenazine Methosulfate

1 ml 0.04 M Glucose-6-phosphate sodium salt

100 mls ..05 M Tris-Maleate buffer pH 8.0

Incubate for 2 to 4 hours at 37°C. Bands are blue on white background. 138

Glucose Phosphate Isomerase

Delorenzo and Ruddle System (1969)

Gel Buffer - pH 7.0

Dilute 65 mI. of electrolyte buffer to 1 liter

Electrolyte Buffer

255 mI. 0.2M Na H P0 °H 0 2 4 2

245 mI. 0.2M Na H P0 0 7H 0 2 4 2

Gel concentration - 13%

Gel Tray - same as above

Voltage - 150V

Current - 10 rnA per gel

Inserts - Whatman No. 3

Stain:

10 mg TPN

80 mg NgCl

10 mg Nitroblue Tetrazolium

1 mg Phenazine Methosulfate

5 ml G6PD

160 mg Fructose 6-Phosphate

100 ml Tris-HCl pH 8.0

Incubate for 2 hours at 37°C. Bands are blue on white background.

Phosphoglucomutase

PGM System

Gel Buffer - pH adjusted to 7.4 with NaOH

1/9 dilution of electrolyte buffer 139

Electrolyte Buffer

60.55 gm. Tris-hydroxymethy1-aminomethane

58.04 gm. Maleic Acid per 5 L. 19.01 gm. Ethylenediamine-tetracetic acid

10.15 gm. MgC1 2

Gel Concentration - 13%

Gel Tray - same as in transferrin

Voltage - 100 V

Current - 40 rnA per gel

Inserts - Whatman No.3

Stain

600 mg. Na glucose-l-phosphateo4 H 0 2 2 200 mg. MgC1 06H 0 2 2 10 mg. TPN

80 units G6PD

20 mg. Nitroblue tetrazolium

1 mg. Phenazine Methosulfate

10 mls •.5M Tris-HCl pH 7.1

90 mls. H20 Incubate at 37°C for 6 hours. Bands are blue on white background.

Esterase

Ashton System - see Transferrin

Gel Concentration - 13%

Gel Tray - same as in APH

Voltage - 250 V 140

Current - 50 rnA per gel

Inserts - Whatman No.1

Stain:

2 mls. 1% a-napthyl-acetate

20 mg Fast Blue RR Salt

20 mg. Fast Red TR Salt

100 mIs .•1M Tris-Maleate Buffer pH 6.5

Bands are reddish brown on white background.

Haptoglobin

Haptoglobin System (Dennis)

3:1 serum to hemolysate

Ashton Gel Buffer and electrolyte buffer

Gel Concentration - 12%

Gel Trays - same as in Transferrin

Voltage - 400 V

Current - 90 rnA per gel

Inserts - Whatman No. 17

Stain:

.8 gm. Benzidine

20 mI. Glacial Acetic Acid

0.5 ml. 30% H 0 2 2 400 ml. H 0 2 Bands are blue on white background at first and then turns brown. 141

Hemoglobin

Cooper System (1969)

Buffer same as in 6PGD

Gel Concentration - 13%

Gel Tray - same as in G6PD

Voltage - 300 V

Current - 60 rnA per gel

Inserts - Whatman No. 3

Stain:

same as in haptoglobin

Tetrazolium Oxidase

Gel Buffer same as in 6PGD

Electrolyte Buffer same as in 6PGD

Gel Concentration - 13%

Gel Tray - same as in 6PGD

Voltage - 100 V

Current - 10 rnA per gel

Inserts - Whatman No. 3

Stain:

10 mg. TPN

30 mg. Nitroblue Tetraxolium

5 mg. Phenazine Methosulfate

100 mls. Tris-Maleate Buffer pH 6.5

Bands are white on blue background. 142

Catalase

Poulik Buffer System (Poulik, 1957)

same as in LDH

Gel Concentration - 13% starch

Gel Trays - same as in LDH

Voltage - 300 V

Current - 30 rnA per gel

Inserts - Whatman No. 3

Stain:

Gel soaked in 0.5% H 0 for 2 minutes 2 2 Washed twice in distilled H 0 2 Immersed in a 1% solution of acidified Potassium

Iodide solution

Bands are white on a dark blue background.

12.3. Results and Discussion

The polymorphisms found in the Transferrin, 6PGD, and APR systems have been discussed previously, therefore, only systems other than these three will be described here. The following are descriptions of the electrophoretic patterns observed for the different loci examined:

1. Pre-Albumen-Ashton System-6 bands of equal intensity and

equally distant from each other. No variation was found

in 100 samples examined.

2. Albumen - 1 band with no variants in 100 samples.

3. Leucine Aminopeptidase - 2 bands of equal intensity. No

variants found in 100 samples examined. 143

4. Amylase - 2 major and one minor band. No variants found

in 100 samples examined.

5. Lactic Dehydrogenase - 1 major and 2 minor bands. No

variants found in 200 samples examined.

6. Malate Dehydrogenase - 1 band with no variants found in

50 samples.

7. Glucose -6-Phosphate Dehydrogenase - 1 band with no

variants found in 165 samples.

8. Glucose Phosphate Isomerase - 1 band with no variants

found in 63 samples.

9. Phosphoglucomutase - 3 bands of equal intensity with the

second band being equally distant from the first and third.

No variants found in 48 samples examined.

10. Esterase - 9 phenotypes (Figure 12.3.1.) were observed.

The number of bands in the A region varies from 0-3 bands.

Many variants were found, however the genetics were

ambiguous even with breeding data.

11. Haptoglobin - 1 broad band at the albumin region, and three

bands of equal intensity halfway between the insert line and

the broad band. No variants were found in 100 samples.

12. Hemoglobin - 1 major band and 1 minor band migrating just

below it. No variants were found in 135 samples examined.

13. Tetrazo1ium Oxidase - 4 bands observed migrating below 6PGD

isozymes. All are spaced equally from each other. These

are not alcohol dehydrogenase isozymes. No variants were

found in 200 samples. 144

14. Catalase - 1 band with no variants found in 60 samples.

All electrophoresis on the above proteins were done on hemolysates of red blood cells, except for the pre-albumin, albumin, and esterase loci, which were done on serum. The esterase isozymes were very un­ stable. The isozyme pattern would change after one cycle of freezing and thawing. The breeding data failed to show any simple inheritance such as co-dominance or simple dominance and recessivity.

There is an interesting point about the 6PGD locus. At the

Bethel locality 2 birds were found which showed abnormal patterns.

They were phenotypically typed as AC's, since the slowest of the three bands migrated slower than that of the single band of the B alleles.

The second band in the AC heterozygote migrated in the same position as the BB dimer in the AB heterozygote and was equally distant from the AA climer and CC dimer (Figure 12.3.2). These two birds were migrants from a pigeon fancier's colony, since they were both banded with seamless aluminum bands. Both birds were also red checks. The bands also seemed to indicate that they were from the same colony, since the number series were very close and the name of the club sponsoring the bands were the same. No other localities revealed the presence of this C allele, therefore, again suggesting local differentiation among pigeon populations.

Therefore, out of a total of 17 loci looked at electrophoretically, only 4 were found to be polymorphic; polymorphism being defined as having an allele frequency of 0.01. This would give the proportion of loci which are polymorphic as 0.25, which is somewhat lower than the estimates from man and Drosophila pseudoobscura. The reason may be 145 that the genome of this pigeon population is very sensitive and we11- balanced, therefore, new mutant alleles will be of very little selective advantage and most likely disadvantageous. This will result in a loss of new mutant alleles by random drift, which would be expected in a population of finite size and no selective advantage for an allele.

If the esterase locus was left out due to the ambiguity of the genetics involved, the resulting table will be as follows:

Table 12.3.1.

Proportion of loci out of 16, polymorphic and proportion of genome estimated to be heterozygous in an average individual from the population studied

Population No. of Loci Proportion Proportion Maximum Polymorphic of Loci of Genome Proportion Polymorphic Heterozygous Genome per Heterozygous Individual

Zoo -3 .19 .07 .09

This low value for average proportion of genome heterozygous per individual is very close to the maximum proportion of the genome which could be heterozygous. However, this is a much lower estimate than that of man and D. pseudoobscura, the reason for this may be due to biases, which enter due to the technique itself, which fails to detect all variants; due to not all changes in amino acid sequence leads to charge differences; and due to the choice of the enzymes examined.

For Gillespie and Kojima (1968) found that Group I enzymes, enzymes 146 which are active in energy metabolism, were less polymorphic than

Group II enzymes, which are non-specific enzymes. However, both of these would bias' the sample towards a smaller measurement of average heterozygosity. Therefore, it must be pointed out that the majority of the loci looked at were indeed enzymes actively involved in energy metabolism, and therefore would consequently give lower estimates.

It was also stated by Lewontin and Hubby (1966) that both genetic drift in populations having periodic size reductions and selection against recessive or partially dominant deleterious genes could decrease the amount of genetic variation found in a population. The Zoo population does undergo periodic size reductions, mainly due to the population con­ trol program of the Zoo, which at times has exterminated 1,000 plus birds within a month. The second element which reduces variability is of course feasible if a very delicately balanced genome resists the encroachment of new alleles.

12.4. Summary

Examination of 17 electrophoretic loci revealed that the number of polymorphic loci was 4. Therefore, the proportion of polymorphic loci was .25. A slow variant of the 6PGD locus was observed at one locality but none of the others, which again suggests local differen­ tiation of pigeon subpopulations. The average heterozygosity per individual was found to be 0.07, if the esterase locus was omitted from the analysis, which is three times lower than that found in man and Drosophila pseudoobscura. Possible explanations of this discrepancy are discussed. 147

XIII. Breeding Data

13.1. Introduction

A breeding colony was set up to check for any selective forces

at the three polymorphic electrophoretic loci. All pairs were made

up of pigeons caught at the Zoo. Each bird was classified as to

transferrin genotype and sex and was placed into its respective

group. A table of random numbers was then used to select mates of

the proper genotype. An assumption was made concerning the randomness

of phenotypes of the other loci, within each transferrin phenotype.

The assumption being that they were randomly distributed within

transferrin genotypes.

13.2. Materials and Methods

One hundred pairs of pigeons were placed into 100 breeding cages.

The dimensions of these breeding cages were 15" X 15" X 15". There

were usually 4 breeding cages in a unit, a few had three breeding

cages to a unit. The cages were made of 1" X 2" welded wire fencing.

A plywood framework was made to hold six of the 4 cage units at one

time, since there were three levels a~d each level held two units back

to back. Two rectangular holes were cut at the front of each breeding

cage, to which the feed and water jars were attached by wire jar

holders. The birds were fed and watered every other day. The feed was usually a mixture of pigeon feed (Bell Grain Company) and Pigeon

Checkers (Ralston-Purina) in the ratio of 1 to 3. No loss of birds due to disease was observed. The sole losses observed came mainly from rat infestation. 148

13.3. Results and Discussion

A total of 89 pairs produced young which could be typed. The other 11 pairs were either eaten by rats or were incompatible.

Tables 13.3.1., 13.3.2., and 13.3.3. summarize the results of the breeding experiments. None of the mating types for the transferrin locus revealed any departures from the expected numbers based on the expected segregation frequencies. Similarly, for the 6PGD locus, none of the mating types showed any significant deviations from the expected. However, two of the mating types, 6PGDA/6PGDB X 6PGDA/

6PGDA and 6PGDA/6PGDB X 6PGDA/6PGDB showed trends toward an excess of heterozygote advantage. No significant departures from the expected segregation ratio was found for any of the mating types of the APH locus.

These results show that in this experiment no evidences for negative heterosis, heterosis, or selection against one allele was found. However, the rather small numbers of young involved, ~ 150, is not sufficient to detect any type of selection unless the selec­ tion is rather extreme, which is not the case here. It should be noted that a larger number of offspring and mated pairs may provide the evidence needed to prove some of the trends, suggestive of heterozygote advantage in the 6PGD locus, significant.

13.4. Summary

The breeding data collected from the above experiment failed to reveal any effects of negative heterosis, heterosis, or selection at any of the alleles of the loci looked at. However, there was a 149 suggestion of heterozygote advantage, which may be significant if a larger sample size is obtained. The failure of these experiments to detect any selection does not proclaim that there is no selection at any of these loci, for small numbers such as these would fail to pick up small and even moderate selection coefficients. 150

XIV. Summary

An analysis of six polymorphic loci was made on six different localities, three of which were electrophoretic loci and the remain- ing three morphological loci. Tests for interaction at the various loci revealed the following:

1. The transferrin locus failed to show any interactions

with any of the other five loci.

2. No differences in transferrin phenotype frequencies

were found for the different sexes and age groups.

3. The transferrin phenotype frequencies were stable

over the years of the Zoo study.

4. No interactions were found between the 6PGD locus

and the transferrin, APH, and shank loci.

5. 6PGD phenotypic frequencies were homogeneous for

sex and age.

6. Heterogeneity in 6PGD phenotype frequencies were found

over the years, and was explained by heterotic selection

at this loci In a finite population.

7. The APH loc~ failed to show any interactions with any

of the other loci studied, nor was there any interactions

between this locus and sex and age.

8. A significant heterogeneity in APH gene frequencies was

found and this was discussed in terms of balancing

effects of selection, random drift, and inbreeding. 151

9. The shank loci also failed to exhibit any interactions

with any other phenotypic class.

The zoo locality differed from all the others in that 3 of the

loci analyzed showed significant positive F values. The 6PGD locus

being the only one which failed to exhibit this characteristic,

probably due in part to heterotic effects. No estimates of F could

be obtained from the plumage pattern loci, since it is a 3-3 phenogram,

that is, ¢ = m. All of the F's were homogeneous.

1. Transferrin - f = .208 ± .043

2. Alkaline Phosphatase - f = .189 ± .048

3. Shank Locus - f = .192 ± .079

Different models of population structure were described, along with their formulae for f. The stepping stone model was applied to

the data in the similar way that Nei and lmaizumi applied it to human

blood group data from Japan. T.he results of this analysis are listed

below:

1. Only the transferrin loci showed a decrease in correlation

with distance. Due to the presence of gene frequency clines,

heterosis, and large sampling error, this decrease in

correlation was not found for the other loci.

2. Genetic variances were considerably larger than sampling

variances for all loci.

3. Some of the fST's were quite large.

4. The genetic covariances of the pattern alleles were all

negative. 152

5. Two gene frequency clines for the APR and shank loci were

found, both of which were increasing from East to West.

These results indicate that most of the significant gene frequency differences found were due mainly to random differentiation between localities.

It was hypothesized that assortative mating, inbreeding, imprint­ ing, and homing behavior all enhanced the first step required under

Wright's polyallelic random drift model of evolution. The result being an increase in the evolutionary rate of the species. 153

Appendix 1

Description of Characters Referred to in the Previous Genetic Analysis

Section (Levi, 1969).

Colors:

Blue Pigeon's blue is a grey or bluish grey. Blue is similar

to black in that the pigment is the same, but unlike

black, which is spread, the granules are clumped.

Black - Black pigmented granules spread evenly throughout the

feather structure.

Brown - Brown is similar to black except for the dark brown

pigment.

Red - Red is similar to black and brown except the pigment

is reddish brown.

Silver - Silver is the dilute of blue, which means that the

pigment is the same, but there are fewer and smaller

granules.

Dun - Dilute of black.

Khaki and Drab - Dilute of brown.

Yellow - Dilute of red

White - No pigment

Grizzle - Feathers have a frost-like or frosted appearance,

caused by presence of partly white barbs. a Opal - Pale red-blue tinge to feathers of wing and cover~s

OpaloP - White ground color with colored cheques 154

Smoky - Clumped areas are darkened, light colored base of beak,

ceres, and skin of body; elimination of whitening of the

rump and the two outer tail feathers; and the enhcnce­

ment of the terminal blue edge of the tail.

Milky - Very light ash coloration

Almond - Yellowish ground color with considerable flecking and

splashing of blue, black or brown

Faded - Color slightly lighter than blue

Sandy - Similar to almond but with less flecking

Reduced - Alteration of coloration in shade from juvenile to

adult coloration; closely linked to dilute gene.

Patterns:

Black Check - Spread areas predominate over clumped. Bird

appears almost black.

Blue Chequered - The clumped areas are almost equal to the

spread areas, forming a chequered pattern.

Blue Bar - Two wing bars with the rest of coverts being of

clumped variety.

Barless - No spread areas found on coverts and bird is all blue

except for the barred condition on the tail feathers.

Eye Color:

Bull - has a black appearance due to the lack of pigment on the

anterior surface of the iris. Therefore, this allows the

posterior pigmented layer of the iris to shine through.

Thus, the appearance of blackness. Therefore, the bull

eye is due to the lack of pigment on the outer surface 155

of the iris. Vecchi (1937).

Pearl Eye - The cells of the iris are crowded with granules

which are themselves colorless but prevent the passage

of light, so that when seen by reflected light, it

appears grey-white.

Yellow Eye - This color eye has a network of branching cells

crowded with small yellow granules lying on the outer

surface of the iris.

Orange and Red Eye - The granules are yellow but the red color

is caused by the presence of red blood vessels. There­

fore, the ruby color is caused by the vessels covering

the yellow pigment.

Albino Eye - pink in color with no pigment granules. This eye

eliminates pigment from the choroid layer and inner

surface of the iris, and, thus has little effect on

outer iris pigment.

Feather Structure and Ornaments:

Crest - Feathers at back of head either extend towards the rear

or are reversed and stand erect.

Hood - Neck feathers form a whorl pattern and stand erect to

give a hooded appearance.

Silky - An abnormal feather condition caused by the barbs being

unconnected by the barbules. This gives it a soft

appearance due to its lack of strength and elasticity

and the tendency of barbs to twist. 156

Scraggly - An abnormal plumage condition similar to silky but

feathers are stiffer and the skin is tight, thick and

crusty.

Porcupine - An abnormal plumage condition caused by the failure

of the vexillum (vane) to unfold and thus remains in its

sheath.

Types of Feathered Legs:

Muffed - Feathers of the tarsi and toes grow to great lengths.

Grouse - The feathers are quite short but very wide.

Slippered - Feathers of the tarsi and toe are short and close­

fitting, and most developed on the toe.

Physiological Traits:

Ataxia Behaviorally the bird exhibits nodding and swaying of

the head and neck, unsteady gait, falling on the side,

flying irregularly and sometimes backward; in general,

complete disequilibrium. Brain is arrested in

juvenile state.

Clumsy - Behaviorally they grope their way while walking and

fly into objects. Their vision is markedly impaired

caused by an abnormality in the rods and cones of the

eyes. TABLE 7.3.1. -- Heterogeneity chi-square test for homogeneity of samples over all areas­ Transferrin locus

TFRA/TFRA TFRA/TFRB TFRB/TFRB 2 Obser- Expec- Obser- Expec- Obser- Expec- X d.L P Area n ved ted ved ted ved te~ AA np2 AB 2npq BB nq

I 24 2 1.17 7 8.66 15 14.17 .26 1 >.500

II 44 7 4.69 15 19.67 22 19.67 1.77 1 >.100

III 196 37 28.58 76 92.84 83 74.56 5.91 1 <.025

IV 13 1 0.60 4 4.80 8 7.60 .04 1 >.750

V 16 4 2.13 4 7.74 8 6.12 2.55 1 >.100

VI 40 10 4.44 7 18.11 23 17.44 13.44 1 >.005

VII 23 0 1.22 11 8.56 12 13.22 .91 1 >.250 2 Sum of X 's 23.97 7 <.005 2 Pooled X 's 356 61 42.50 124 161. 00 171 152.45 18.81 1 <.005

Heterogeneity X2 . 5.61 6 >.500 6 ...... IJ1 158

TABLE 7.3.2. -- Contingency table of transferrin phenotypes within sexes

Sex TfrA/TfrA TfrA/TfrB TfrB/TfrB Observed Observed Observed

Male 22 71 107

Female 26 65 94

2 X2 = .14 p > .900

TABLE 7.3.3. -- Contingency table of transferrin phenotypes within ages

Age TfrA/TfrA TfrA/TfrB TfrB/TfrB Observed Observed Observed

Young 27 43 78

Mature 47 135 199

2 = 3.95 p > .100 X2 159

TABLE 7.3.4. -- Contingency table of transferrin phenotypes within shank phenotypes

Shank TfrA/TfrA . TfrA/TfrB TfrB/TfrB Phenotype Observed Observed Observed

Feathered 1 9 12

Non-feathered 27 92 151

2 - .91 p > .500 X2

TABLE 7.3.5. -- Contingency table of transferrin phenotypes within the white and colored populations

Plumage TfrA/TfrA TfrA/TfrB TfrB/TfrB Color Observed Observed Observed

White 44 105 167

Colored 6 14 13

2 = p > .250 X2 2.62 160

TABLE 7.3.6. -- Contingency table of transferrin phenotypes within eyecolor phenotypes

Eyecolor TfrA/TfrA TfrA/TfrB TfrB/TfrB Observed Observed Observed

Red 0 6 7

Brown 16 55 102

Pearl 0 2 5

Orange 5 12 18

Yellow 3 9 15

2 Xa = 4.24 p > .750

TABLE 7.3.7. -- 2X 2 contingency table of transferrin homozygote and heterozygote frequencies within the sexes

Tfr Homozygotes Heterozygotes

Male 129 71

Female 120 65

2 Xl contingency = .000 161

TABLE 7.3.8. -- 2 X 2 contingency table of homozygote and heterozygote frequencies within ages the transferrin locus

Tfr Homozygotes Heterozygotes

Young 105 43

Old 246 135

2 Xl contingency with Yates = 1.667 p > .100

TABLE 7.3.9. -- 2 X 2 contingency table of homozygote and heterozygote frequencies within shank phenotypes -- the transferrin locus

Tfr Homozygotes Heterozygotes

Feathered 13 9

Non-feathered 178 92

2 Xl contingency = .170 162

TABLE 7.3.10. -- 2 X 2 contingency table of homozygote and heterozygote frequencies within the white and colored populations -- the transferrin locus

Tfr Homozygotes Heterozygotes

White 220 105

Colored 19 14

2 Xl contingency = .962 TABLE 7.3.11. -- Woolf's analysis to test for differences in homozygote and heterozygote frequencies within sexes -- the transferrin locus

System Hale Female 2 Homo- Hetero- Homo- Hetero- X = hI< d.L X P zygotes zygotes zygotes zygotes Hk (h) (k) (H) (K)

Transferrin 103 56 83 53 l.17 1 0.44 >.500

6-Phosphog1uconic 90 69 85 51 0.78 1 l.05 >.250 Acid Dehydrogenase

Alkaline 124 35 106 30 l.00 1 0.00 Phosphatase

..... 0\ W TABLE 7.3.12. -- 1voolf's analysis to test for differences in homozygote and heterozygote frequencies within ages -- the transferrin, 6-phosphogluconate dehydrogenase, and alkaline phosphatase loci

Young Mature 2 System Homo- Hetero- Homo- Hetero- X=hK d.f. X P zygotes zygotes zygotes zygotes Hk (h) (k) (H) (K)

Transferrin 71 28 184 109 1.50 1 2.57 >.100

6-Phosphogluconic 65 34 176 117 1.27 1 0.97 >.250 Acid Dehydrogenase

Alkaline 124 35 106 30 1.00 1 0.00 Phosphs.case

..... C1' ~ TABLE 7.3.13. -- Woolf's analysis to test for differences in homozygote and heterozygote frequencies between the white and colored populations -- the transferrin, 6-phosphogluconate dehydrogenase, and alkaline phosphatase loci

White Colored 2 System Homo­ Hetero­ Homo- Hetero- x=hK d.f. X P zygotes zygotes zygotes zygotes Hk h k H K

Transferrin 166 84 33 16 .958 1 0.02 .900

6-Phosphogluconic 155 95 32 17 .867 1 0.19 >.500 Acid Dehydrogenase

Alkaline 204 46 39 10 1.140 1 0.11 >.500 Phosphatase

I-' CJ'\ In TABLE 7.3.14. -- Woolf's analysis of 2 X 2 comparisons of the eyeco1or phenotypes~ to test for differences in homozygote and heterozygote frequencies -- the transferrin locus

2 Eyeco1or X Eyeco1or Homo­ Hetero­ Homo­ Hetero­ X=hK d.f. X P zygotes zygotes zygotes zygotes Hk (h) (k) (H) (K)

1. Red X Brown 6 5 96 47 0.59 1 0.71 >.250

2. Red X Pearl 6 5 5 1 0.24 1 1.30 >.250

3. Red X Orange 6 5 19 9 0.57 1 0.60 >.250

4. Red X Yellow 6 5 14 7 0.60 1 0.45 .250

5. Brown X Pearl 96 47 5 1 0.41 1 0.65 >.250

6. Brown X Orange 96 47 19 9 0.96 1 0.01 >.900

7. Brown X Yellow 96 47 14 7 1.02 1 0.00 >.900

8. Pearl X Orange 5 1 19 9 2.36 1 0.55 >.250

9. Pearl X Yellow 5 1 14 7 2.50 1 0.59 >.250

10. Orange X Yellow 19 9 14 7 1.06 1 0.01 >.900

I-' C' 0\ 167

TABLE 7.3.15. -- Contingency table to test first order interactions between the transferrin and 6­ Phosphogluconate dehydrogenase loci

6PGD TfrA/TfrA TfrA/TfrB TfrB/TfrB Phenotypes Observed Observed Observed

6PGDA/6PGDA 40 95 151

6PGDA/6PGDB 25 73 98

6PGDB/6PGDB 4 6 15

X~ = 2.10 p>.500

TABLE 7.3.16. -- Contingency table to test first order interactions between the transferrin and alka­ line phosphatase loci

APH TfrA/TfrA TfrA/TfrB TfrB/TfrB Phenotypes Observed Observed Observed

APHA/APHA 4 10 11

APHA/APHB 20 44 58

APHB/APHB 42 116 181

2 = 2.33 p>.250 X4 168

TABLE 7.3.17. -- Contingency table of transferrin phenotypes over the years 1969-1972

Year TfrA/TfrA TfrA/TfrB TfrB/TfrB Observed Observed Observed

1969 35 67 90

1970 7 15 21

1971 34 97 162

1972 5 16 26

2 6.05 p > .250 X6 = TABLE 7.3.18. -- Contingency chi-square values and their associated p values, to test for homogeneity of transferrin phenotype frequencies between sexes within localities

Transferrin 2 Locality Male Female d.L X P AA AB BB AA AB BB Observed Observed

St. Andrew's 5 10 18 2 10 16 2 1.00 >.500

Bethel 2 19 12 0 10 9 2 1.57 >.250

Pier 27 1 12 9 4 9 11 2 2.35 >.250

Pier 32 6 11 3 8 13 6 2 0.42 >.750

..... \0'" TABLE 7.3.19. -- Contingency chi-square values and their associated p values, used for testing the homogeneity of transferrin phenotypes between ages within localities

Transferrin 2 Locality Young Mature d. f. X P AA AB BB AA AB BB Observed Observed

St. Andrew's 1 2 1 7 20 33 2 1.47 >.250

Bethel 2 5 5 2 27 21 2 2.69 >.250

Pier 27 4 8 10 2 16 15 2 2.22 >.250

Pier 32 1 10 7 15 21 10 2 5.48 .10>p>.05

...... I--' o TABLE 7.3.20. -- Contingency chi-square values and their associated p values, used to test first order interactions between the transferrin locus and the shank locus, within their respective localities

Shank ++ + - Transferrin 2 Locality AA AB BB AA AB BB AA AB BB d.f. X P Observed Observed Observed

St. Andrew's 1 2 0 2 10 14 5 10 20 4 4.69 >.250

Bethel 0 3 3 0 9 8 4 20 15 4 2.68 >.500

Pier 27 1 3 2 2 6 11 3 14 13 4 1. 75 >.750

Pier 32 3 3 5 6 3 6 7 25 6 4 12.07 .025>p>.010

I-' "'-J I-' 172

TABLE 7.3.21. -- Contingency table of transferrin pheno­ types within the orange eyeco1or and within localities

Locality Orange Eyeco1or TfrA/TfrA TfrA/TfrB TfrB/TfrB Observed Observed Observed

St. Andrew's 5 16 26

Bethel 0 18 14

Pier 27 1 8 7

Pier 32 8 3 2

2 35.83 p<.OO5 X6 = TABLE 7.3.22. -- Woolf's analysis to test for differences in homozygote and hetero­ zygote frequencies between the sexes, and within localities -- the transferrin locus

Transferrin Male Female 2 Locality Homo- Hetero- Homo- Hetero- X=hK dL X P zygotes zygotes zygotes zygotes Hk

1. St. Andrew's 23 10 18 10 1.28 1 .20 >.500

2. Bethel 24 9 10 9 2.40 1 2.11 >.100

3. Pier 27 10 12 15 9 .50 1 1.33 >.100

4. Pier 32 9 11 14 13 .760 1 .22 >.500

Deviation X~ = 0.00

Heterogeneity X; = 3.85 p>.250

I-' '-l W TABLE 7.3.23. -- Woolf's analysis to test for differences in homozygote and hetero­ zygote frequencies between ages, and within localities -- the transferrin locus

Transferrin Young Mature 2 Locality Homo- Hetero- Homo- Hetero- X=hK d.L X P zygotes zygotes zygotes zygotes Hk

1. St. Andrew's 2 2 40 20 .500 1 0.45 .500

2. Bethel 7 5 23 27 1.643 1 0.58 >.250

3. Pier 27 14 8 17 16 1.647 1 0.78 >.250

4. Pier 32 8 10 25 21 .672 1 0.51 >.250

Deviation xi = 0.00 . 2 Heterogenelty X = 2.32 p>.500 3

...... I-' ~ TABLE 7.3.24. -- Heterogeneity chi-square to test for heterogeneity between the samples from the different localities -- the transferrin locus

TfrA/TfrA TfrA/TfrB TfrB/TfrB 2 Obser- Expec- Obser- Expec- Obser- Expec- X d.f. P Locality n ved ted ved ted ved ted AA np2 AB 2npq BB nq2 - Zoo 575 81 55.30 195 246.39 299 273.31 24.31 1 <.005

St. Andrew's 64 8 5.24 21 26.51 35 32.24 2.08 1 >.100

Bethel 63 4 6.24 32 27.52 27 27.24 1.16 1 >.250

Ha1dron 21 1 1.26 8 7.66 12 12.18 .07 1 >.750

Pier 27 59 8 6.67 24 26.67 27 25.67 .30 1 >.500

Pier 32 64 16 15.38 31 32.24 17 16.38 .02 1 >.900 2 Sum of X 's 27.94 6 <.005 2 Pooled X 's 846 118 84.60 311 372.24 417 389.16 21.21 1

Heterogeneity 6.73 5 >.100 2 X5

t-' "'-J IJ1 176

TABLE 7.3.25. -- Test of homogeneity of gene frequencies for the different localities -- the transferrin locus

Location Total A Genes p (x/2n) px Genes (2n) (x)

Zoo 1150 357 .310 110.813

St. Andrew's 128 37 .289 10.693

Bethel 126 40 .317 12.696

Waldron 42 10 .238 2.380

Pier 27 118 40 .339 13.556

Pier 32 128 63 .525 33.075

Totals 1692 l:x=547 l:px=183.213 p=.3232 2 X -l:px- pl:x =29.370 p<.005 q=.6768 5 pq 177

TABLE 7.3.26. -- Contingency chi-square values and their associated p values, used in testing the homogeneity of transferrin phenotypes between two localities

2 Locality X Locality d. f. X P

1. Zoo X St. Andrew's 2 0.20 >.900

2. Zoo X Bethel 2 8.03 .025>p>.010

3. Zoo X Waldron 2 1.49 >.250

4. Zoo X Pier 27 2 1.12 >.500

5. Zoo X Pier 32 2 15.49 <.005

6. St. Andrew's X Bethel 2 4.64 >.100

7. St. Andrew's X Waldron 2 1.04 >.500

8. St. Andrew's X Pier 27 2 1.03 >.500

9. St. Andrew's X Pier 32 2 10.82 <.005

10. Bethel X Waldron 2 1.29 >.500

11. Bethel X Pier 27 2 2.35 >.250

12. Bethel X Pier 32 2 9.48 • OlO>p>. 005

13. Waldron X Pier 27 2 1.50 >.250

14. Waldron X Pier 32 2 7.94 .025>p>.010

15. Pier 27 X Pier 32 2 5.64 .100>p>.050 178

TABLE 7.3.27. -- Contingency table to test the homogeneity of transferrin phenotypes between localities

Locality TfrA/TfrA TfrA/TfrB TfrB/TfrB Observed Observed Observed

Zoo 81 195 299

St. Andrew's 8 21 35

Bethel 4 32 27

Waldron 1 8 12

Pier 27 8 24 27

Pier 32 16 31 17

2 X = 25.94 p<.OO5 10 179

TABLE 7.3.28. -- Table of estimates of p, q, and a for the transferrin locus gotten from Yasuda's G-type program

Locality p±crp q±cr q a±cr a

Zoo 0.3l0±.O15 0.690±.015 0.208±.043

St. Andrew's O. 290±' 044 0.710±.044 0.202±.130

Bethel 0.3l7±.038 0.683±.038 -0 .172±.115

Waldron 0.238±.064 O. 762±. 064 -0.050±.209

Pier 27 0.339±.042 0.66l±.042 0.089±.128

Pier 32 0.492±.045 0.508±' 045 0.03l±.124 TABLE 7.3.29. -- Woolf's analysis of homozygote and heterozygote frequencies between two localities -- the transferrin locus

2 Locality X Locality Homo- Hetero- Homo- Hetero- X=hK d.L X P zygotes zygotes zygotes zygotes Hk (h) (k) (H) (K)

1. Zoo X St. Andrew's 380 195 43 21 0.95 1 0.03 >.750 2. Zoo X Bethel 380 195 31 32 2.01 1 6.85 .010>p>.005 3. Zoo X Waldron 380 195 13 8 1.20 1 0.16 >.500 4. Zoo X Pier 27 380 195 35 24 1.08 1 1.08 >.250 5. Zoo X Pier 32 380 195 33 31 1.83 1 5.20 .025>p>.010 6. St. Andrew's X Bethel 43 21 31 32 2.11 1 4.17 .005>p>.025 7. St. Andrew's X Waldron 43 21 13 8 1.26 1 0.20 >.500 8. St. Andrew's X Pier 27 43 21 35 24 1.40 1 0.82 >.250 9. St. Andrew's X Pier 32 43 21 33 31 1.92 1 3.21 >.050 10. Bethel X Waldron 31 32 13 8 0.60 1 1.01 >.250 11. Bethel X Pier 27 31 32 35 24 0.66 1 1.25 >.250 12. Bethel X Pier 32 31 32 33 31 0.91 1 0.07 >.750 13. Waldron X Pier 27 13 8 35 24 1.11 1 0.04 >.750 14. Waldron X Pier 32 13 8 33 31 1.52 1 0.68 >.250 15. Pier 27 X Pier 32 35 24 33 31 1.36 1 0.75 >.250

Deviation xi = (Lwy)2/LW = 19.98 p<.005 2 2 2 ..... Heterogeneity X = Lwy - (Lwy) /LW = 32.63 p<.005 00 14 0 TABLE 8.3.1. -- Heterogeneity chi-square test of samples taken from different areas -­ the 6PGD locus

6PGDA/6PGDA 6PGDA/6PGDB 6PGDB/6PGDB 2 Obser- Expec- Obser- Expec- Obser- Expec- X d.L P Area n ved ted ved ted ved ted AA np2 AB 2npq BB nq2

I 31 21 21. 74 10 8.52 0 0.74 .19 1 >.500

II 73 52 51. 74 19 19.51 2 1. 74 .04 1 >.750

III 220 112 120.67 102 84.66 6 14.67 8.46 1 <.005

IV 13 7 7.60 6 4.80 0 0.60 .12 1 >.500

V 18 9 7.23 5 8.54 4 2.23 2.03 1 >.250

VI 61 37 34.60 18 22.81 6 3.59 1.93 1 >.250

VII 23 18 18.22 5 4.55 0 0.22 .35 1 >.500 2 Sum of X 's 13.07 7 >.050 2 Pooled X 's 439 256 261.03 165 154.97 18 23.00 1.85 1 >.100

Heterogeneity 2 X6 11.24 6 .10>p>.05

coI-' I-' 182

TABLE 8.3.2. -- Table of pair-wise comparisons of hetero­ zygote and homozygote frequencies, between areas within the zoo -- the 6PGD locus

2 Locality X Locality X=hK d.L X P Hk

1. IX II 1.64 1 0.31 >.500 2. XIII 3.89 1 2.83 >.050 3. X IV 2.40 1 0.66 >.250 4. XV 1.21 1 0.05 >.750 5. X VI 2.40 1 0.66 >.250 6. XVII 1.60 1 0.23 >.500 7. II X III 2.38 1 4.07* .05>p>.025 8. X IV 2.44 1 1. 22 >.250 9. XV 1.47 1 0.21 >.500 10. X VI 0.74 1 0.26 >.500 11. XVII 0.98 1 0.00 12. III X IV 0.62 1 0.42 >.500 13. XV 0.62 1 0.42 >.500 14. X VI 0.31 1 6.30* .025>p>.01O 15. XVII 0.41 1 2.09 >.250 16. IV XV 0.7- 1 0.08 >.750 17. X VI 0.38 1 1. 47 >.100 18. XVII 0.50 1 0.60 >.250 19. VX VI 0.51 1 0.64 >.250 20 XVII 0.67 1 0.19 >.500 21. VI XVII 1.31 1 0.14 >.500

2 2 Heterogeneity X Deviation Xl = 0.00 20 183

TABLE 8.3.3. -- Contingency table to test for homogeneity of 6PGD phenotype frequencies within the sexes -- Zoo

Sex 6PGDA/6PGDA 6PGDA/6PGDA 6PGDA/6PGDA Observed Observed Observed

Male 116 86 10

Female 113 67 10

2 = 1.19 p > .500 X2

TABLE 8.3.4. -- Contingency table of 6PGD phenotype frequencies within ages -- zoo

Age 6PGDA!6PGDA 6PGDA/6PGDB 6PGDB/6PGDB Observed Observed Observed

Young 117 68 9

Mature 228 150 19

2 .46 p > .750 X2 = 184

TABLE 8.3.5. -- Contingency table to test interaction between the 6PGD and shank loci -- zoo

Shank 6PGDA/6PGDA 6PGDA/6PGDB 6PGDB/6PGDB Phenotype Observed Observed Observed

Feathered 18 9 3

Non-Feathered 192 119 15

2 1. 91 p > .250 X2 =

TABLE 8.3.6. -- Contingency table of 6PGD phenotype frequencies within the white and colored population -- zoo

Plumage 6PGDA/6PGDA 6PGDA/6PGDB 6PGDB/6PGDB Color Observed Observed Observed

White 219 129 017

Colored 023 014 001

2 = .34 p > .750 X2 185

TABLE 8.3.7. -- Contingency table of 6PGD phenotype frequencies within eyecolor phenotypes -- zoo

Eyecolor 6PGDA/6PGDA 6PGDA/6PGDB 6PGDB/6PGDB Observed Observed Observed

Red 8 4 2

Brown 104 62 9

Pearl 5 2 1

Orange 19 8 1

Yellow 14 8 1

2 4.14 p > .750 X8

TABLE 8.3.8. -- 2 X 2 contingency table of homozygote and heterozygote frequencies within the sexes -- zoo

6PGD Homozygotes Heterozygotes

Male 126 86

Female 123 67

2 Xl contingency = .980 186

TABLE 8.3.9. -- 2 X 2 contingency table of homozygote and heterozygote 6PGD frequencies within ages -- zoo

6PGD Homozygotes Heterozygotes

Young 126 68

Old 247 150

2 Xl contingency with Yates = .306

TABLE 8.3.10. -- 2 X 2 contingency table of homozygote and heterozygote 6PGD frequencies within shank phenotypes -- zoo

6PGD Homozygotes Heterozygotes

Feathered 21 9

Non-feathered 207 119

2 Xl contingency = .261 187

TABLE 8.3.11. -- 2 X 2 contingency table of homozygote and heterozygote 6PGD frequencies within the white and colored population -- zoo

6PGD Homozygotes Heterozygotes

White 236 129

Colored 24 14

2 Xl contingency = .000 TABLE 8.3.12. -- Woolf's analysis to test for differences in homozygote and heterozygote frequencies between shank phenotypes -- the transferrin, 6PGD, and APR systems

Shank Phenotype Feathered Non-feathered 2 X Romo- Retero- Romo- Retero- X=hK d.f. X P System zygotes zygotes zygotes zygotes Rk

Shank X Transferrin 8 7 III 56 0.58 1 1.03 >.250

Shank X 6PGD 12 3 101 66 2.61 1 2.09 >.100

Shank X APR 14 1 137 30 3.06 1 1.13 >.250

..... ex> ex> TABLE 8.3.13. -- Woolf's analysis of pair-wise comparisons of homozygote and heterozygote 6PGD frequencies between two eyecolor phenotypes -- zoo

2 Eyecolor X Eyecolor Homo- Hetero- Homo- Hetero- X=hK d.f. X P zygotes zygotes zygotes zygotes Hk h k H K

1. Red X Brown 7 4 89 54 1.06 1 0.01 >.900

2. Red X Pearl 7 4 4 2 .88 1 0.02 .900

3. Red X Orange 7 4 16 12 1.31 1 0.14 >.500

4. Red X Yellow 7 4 12 9 1.31 1 0.13 >.500

5. Brown X Pearl 89 54 4 2 .82 0.05 >.750

6. Brown X Orange 89 54 16 12 1.23 0.26 >.500

7. Brown X Yellow 89 54 12 9 1.23 0.20 >.500

8. Pearl X Orange 4 2 16 12 1.50 0.18 >.500

9. Pearl X Yellow 4 2 12 9 1.50 0.17 >.500

10. Orange X Yellow 16 12 12 9 1.00 0.00

.... 00 1.0 190

TABLE 8.3.14. -- Contingency table to test for first order interactions between the 6PGD and alkaline phosphatase loci

APH 6PGDA/6PGDA 6PGDA/6PGDB 6PGDB/6PGDB Phenotypes Observed Observed Observed

APHA/APHA 16 12 1

APHA/APHB 57 43 5

APHB/APHB 207 145 19

.29 p>.990

TABLE 8.3.15. -- Contingency table of 6PGD phenotype frequencies over the years 1969-1972

Year 6PGDA!6PGDA 6PGDA/6PGDB 6PGDB/6PGDB Observed Observed Observed

1969 143 91 12

1970 54 15 4

1971 140 116 12

1972 26 19 2

12.92* p < .050 TABLE 8.3.16. -- Heterogeneity chi-square test of the underlying genetic hypothesis -- 6PGD locus

6PGDA/6PGDA 6PGDA/6PGDB 6PGDB/6PGDB 2 Obser. Expec- Obser- Expec- Obser- Expec- X d.f. P Locality n ved ted ved ted ved ted AA np2 AB 2npq BB nq2

Zoo 634 363 368.64 241 229.73 30 35.64 1.32 1 .250

St. Andre\v's 58 47 45.68 9 11.64 2 .68 1.40 1 p>.750

Bethel 62 55 54.22 6 7.54 1 .23 .47 1 p>.900

Ha1dron 22 19 19.07 3 2.86 0 .07 2.71 1 .900

Pier 27 58 49 48.39 8 9.22 1 .39 .09 1 p>.900

Pier 32 60 52 52.23 8 7.53 0 .24 .30 1 p>.900 2 Sum of X 's 3.79 6 p<.500 2 Pooled X 's 585 587.11 275 273.88 34 33.01 .01 1

Heterogeneity 3.78 5 p<.500 2 X5

I-' \0 I-' 192

TABLE 8.3.17. -- Estimates of p, q, and a for the different localities, obtained from Yasuda's G-type program

Locality p±a q±a a±a p q a

Zoo O.763±.O12 O.237±.O12 -O.OSO±.038

St. Andrew's O.888±.O32 O.112±.O32 O.220±.180

Bethel O.93S±.024 O.06S±.024 O.198±.211

Waldron O.932±.O38 O.O68±.O38 no estimate

Pier 27 O.914±.O28 O.O86±'O28 O.12S±.180

Pier 32 O.933±.O32 O.O67±.O32 no estimate 193

TABLE 8.3.18. -- Test of homogeneity of 6PGD gene frequencies between localities

Location Total A P px Genes (2n) Genes (x) (xj2n)

Zoo 1268 967 .763 737.434

St. Andrew's 116 103 .888 91.454

Bethel 124 116 .935 108.506

Waldron 44 41 .932 38.204

Pier 27 '6 106 .914 96.852

Pier 32 120 112 .933 104.530

Totals 1788 2:x==1445 1176.980 p .8081 = 2 X 2:px - p2:x 59.8412 p <.005 5 = = q = .1919 pq 194

2 TABLE 8.3.19. -- Contingency X values and their associated p values of 6PGD phenotype frequencies between two localities

2 Locality X Locality d.L X P

1. Zoo X St. Andrew's 2 12.70 <.005

2. Zoo X Bethel 2 23.31 <.005

3. Zoo X Waldron 2 7.53 .025>p>.010

4. Zoo X Pier 27 2 16.35 <.005

5. Zoo X Pier 32 2 20.07 <.005

6. St. Andrew's X Bethel 2 1.43 >.250

7. St. Andrew's X Waldron 2 0.85 >.500

8. St. Andrew's X Pier 27 2 .43 >.750

9. St. Andrew's X Pier 32 2 2.28 >.250

10. Bethel X Waldron 2 0.60 >.500

11. Bethel X Pier 27 2 0.50 >.750

12. Bethel X Pier 32 2 1. 34 >.500

13. Waldron X Pier 27 2 0.39 >.750

14. Waldron X Pier 32 2 0.001 >.995

15. Pier 27 X Pier 32 2 1.05 >.500 TABLE 8.3.20. -- Woolf's analysis of pair-wise comparisons of homozygote and heterozygote 6PGD frequencies between two localities

2 Locality X Locality Homo- Hetero- Homo- Hetero- X=hK d.f. X P zygotes zygotes zygotes zygotes Hk (h) (k) (H) (K)

1. Zoo X St. Andrew's 393 241 49 9 0.30 1 10.52 <.005 2. Zoo X Bethel 393 241 56 6 0.17 1 15.92 <.005 3. Zoo X Waldron 393 241 19 3 0.26 1 4.69 ,050>p>.025 4. Zoo X Pier 27 393 241 50 8 0.26 1 11.90 <.005 5. Zoo X Pier 32 393 241 52 8 0.25 1 12.67 <.005 6. St. Andrew's X Haldron 49 9 19 3 0.86 1 0.04 >.750 7. St. Andrew's X Bethel 49 9 56 6 0.58 1 0.92 >.250 8. St. Andrew's X Pier 27 49 9 50 8 0.87 1 0.07 >.750 9. St. Andrew's X Pier 32 49 9 52 8 0.83 1 0.11 >.500 10. Bethel X Haldron 56 6 19 3 1.47 1 0.26 >.500 11. Bethel X Pier 27 56 6 50 8 1.49 1 0.49 >.250 12. Bethel X Pier 32 56 6 52 8 1.44 1 0.40 >.500 13. Haldron X Pier 27 19 3 50 8 1.01 1 0.00 .000 14. Haldron X Pier 32 19 3 52 8 0.97 1 0.00 .000 15. Pier 27 X Pier 32 50 8 52 8 0.96 1 0.01 >.900

..... \0 \Jl TABLE 8.3.21. -- Contingency chi-square values and their associated p values 6PGD phenotype frequencies between the sexes, with the data subdivided by localities

Dehydrogenase 6-Phosphogluconic Acid 2 Locality Males Females d.£. X P AA AB BB AA AB BB Observed Observed

St. Andrew's 31 2 o 22 5 1 2 3.43 >.100

Bethel 29 3 o 14 2 1 2 2.03 >.250

Pier 27 20 3 o 19 3 o 2 0.00 >.995

Pier 32 16 2 o 21 5 o 2 0.52 >.750

...... \0 0\ TABLE 8.3.22. -- Contingency chi-square values and their associated p values of 6PGD phenotype frequencies between ages, with the data subdivided by localities

6-Phosphogluconic Acid Dehydrogenase 2 Locality Young Mature d.L x P AA AB BB AA AB BB Observed Observed

St. Andrew; s 3 1 o 52 7 1 2 0.66 >.500

Bethel 11 2 o 42 4 1 2 0.78 >.500

Pier 27 19 4 o 29 3 o 2 0.78 >.500

Pier 32 14 2 o 38 6 o 2 0.01 >.900

..... \0 '-I TABLE 8.3.23. -- Contingency chi-square values and their associated p values of 6PGD phenotype frequencies between shank phenotypes, with the data subdivided by localities

Shank -t+ +- 6-Phosphogluconic Acid Dehydrogenase 2 Locality AA AB BB AA AB BB AA AB BB d.L X p Observed Observed Observed

St. Andrew's 3 o a 25 1 a 27 7 1 4 5.04 >.250

Bethel 6 o o 13 2 1 34 4 o 4 3.65 >.250

Pier 27 5 1 o 16 3 o 26 3 1 4 1.23 >.750

Pier 32 10 1 o 12 3 o 30 4 o 4 0.82 >.900

...... \C OJ TABLE 8.3.24. -- Woolf's analysis to test for differences in homozygote and heterozygote frequencies between the sexes, with the data subdivided by localities

Acid Dehydrogenase 6-Phosphogluconic 2 Locality Homo- Hetero­ Homo- Hetero­ X=hK d. f. X P zygotes zygotes zygotes zygotes Hk Hale Female

1. St. Andrew's 31 2 23 5 3.36 1 1.90 >.100

2. Bethel 29 3 15 2 1.29 1 0.07 >.750

3. Pier 27 20 3 19 3 1.05 1 0.00

4. Pier 32 16 2 21 5 1. 91 1 0.51 >.250

Deviation xi = 0.00 Heterogeneity X~ = 2.49 p>.500

..... \C \C TABLE 8.3.25. -- Woolf's analysis to test for differences in homozygote and heterozygote frequencies between ages, with the data subdivided by localities

6-Phosphog1uconic Acid Dehydrogenase Young Mature 2 Locality Homo- Hetero- Homo- Hetero­ X-hK d.f. X P zygotes zygotes zygotes zygotes Hk

1. St. Andrew's 3 1 53 7 .396 1 0.57 >.250

2. Bethel 11 2 43 4 .512 1 0.52 >.250

3. Pier 27 19 4 29 3 .491 1 0.75 >.250

4. Pier 32 14 2 38 6 1.105 1 0.01 >.900

Deviation xi = 0.00

Heterogeneity X; = 1.86 p>.500

oN o TABLE 9.3.1. -- Heterogeneity chi-square test of samples taken from different areas within the zoo -- alkaline phosphatase locus

2 Obser- Expec- Obser- Expec- Obser- Expec- X d.L P Area n ved ted ved ted ved ted AA np2 AB 2npq BB nq2

I 26 1 .41 5 6.18 20 19.41 .00 1 >.900

II 83 5 1.67 11 20.66 67 60.67 10.97 1 <.005

III 194 12 8.17 49 63.67 133 122.17 5.09 1 <.005

IV 11 0 .14 3 2.71 8 8.14 .06 1 >.750

V 11 0 .28 4 3.43 7 7.29 .02 1 .900

VI 49 5 1.08 5 12.84 39 35.08 15.31 1 <.005

VII 18 2 .29 1 4.43 15 13.28 7.21 1 .01>p>.005 2 Sum of X 's 38.66 7 <.005 2 Pooled X 's 392 25 12.07 78 112.15 289 267.78 27.63 1 <.005

Heterogeneity 2 X6 11.03 6 .01>p>.05

N 0 I-' 202

TABLE 9.3.2. -- Woolf's analysis of pair-wise comparison of APH homozygote and heterozygote frequencies between areas within the zoo

Location X Location X=hK d. f. p Hk

1. IX II 0.13 1 2.45 >.100 2. XIII 1.48 1 0.24 >.500 3. X IV 2.00 1 0.43 >.500 4. XV 1.33 1 0.06 >.750 5. X VI 0.44 1 0.66 >.500 6. XVII 0.31 1 0.82 >.500 7. II X III 11.14 1 5.44* .025>p>.010 8. X IV 15.00 1 4.78* .050>p>.025 9. XV 3.33 1 1.03 >.250 10. X VI 3.33 1 1.03 >.250 11. X VII 2.31 1 0.33 >.500 12. III X IV 1.34 0.17 >.500 13. XV 0.90 0.02 >.900 14. X VI 0.30 3.59 >.100 15. XVII 0.21 2.23 >.100 16. VI XV 0.67 1 0.14 >.500 17. X VI 0.22 1 2.60 >.100 18. XVII 0.15 1 2.22 >.100 19. VX VI 0.33 1 1.16 >.250 20. XVII 0.23 1 1. 23 >.250 21. VI X VII 0.69 1 0.09 >.250

2 2 Deviation X = 0.00 Heterogene1.ty. X 20 203

TABLE 9.3.3. -- Contingency table of APH phenotype frequencies within the sexes -- zoo

Sex APHA/APHA APHA/APHB APHB/APHB Observed Observed Observed

Male 10 53 140

Female 8 37 138

2.05 p>.250

TABLE 9.3.4. -- Contingency table of APH phenotype frequencies within ages -- zoo

Age APHA/APHA APHA/APHB APHB/APHB Observed Observed Observed

Young 7 35 133

Mature 18 90 278

2 X = .97 p >.500 2 204

TABLE 9.3.5. -- Contingency table of APH phenotype frequencies within shank phenotypes -- zoo

Shank APHA/APHA APHA/APHB APHB/APHB Phenotype Observed Observed Observed

Feathered 3 7 23

Non-feathered 26 62 248

2 0.262 p <.750 X2 =

TABLE 9.3.6. -- 2 X 2 contingency table of homozygote and heterozygote APH frequencies within the sexes -­ zoo

APH Homozygotes Heterozygotes

Hale 160 43

Female 160 33

2 Xl contingency = .812 205

TABLE 9.3.7. -- 2 X 2 contingency table of homozygote and heterozygote APH frequencies within ages -- zoo

APH Homozygotes Heterozygotes

Young 150 25

Old 309 74

2 Xl contingency with Yates = 1.752

TABLE 9.3.8. -- 2 X 2 contingency table of homozygote and heterozygote APH frequencies within shank phenotypes zoo

APH Homozygotes Heterozygotes

Feathered 26 7

Non-feathered 274 62

2 Xl contingency with Yates = .024 206

TABLE 9.3.9. -- Contingency table APH phenotype frequencies within the white and colored populations -- zoo

Plumage APHA/APHA APHA/APHB APHB/APHB Color Observed Observed Observed

White 18 78 258

Colored 2 7 25

.07 p >.950

TABLE 9.3.10. -- Contingency table of APH phenotype frequencies within the different eyecolor phenotypes -- zoo

Eyecolor APHA/APHA APHA/APHB APHB/APHB Observed Observed Observed

Red 2 2 9

Brown 19 36 143

Pearl 0 1 7

Orange 3 8 25

Yellow 4 8 18

2 = 3.97 p >.750 X8 TABLE 9.3.11. -- \voolf's analysis of pair-wise comparisons of homozygote and heterozygote APH frequencies between two eyecolor phenotypes -- zoo

------2 ,P Eyecolor X Eyecolor Homo- Hetero- Homo- Hetero- X=hK d. -'. X zygotes zygotes zygotes zygotes Hk (h) (k) (H) (K)

1. Red X Brmm 8 2 115 28 0.97 1 0.00 .000

2. Red X Pearl 8 2 5 1 0.80 1 0.03 >.750

3. Red X Orange 8 2 21 7 1.33 1 0.10 >.750

4. Red X Yellow 8 2 14 7 2.00 1 0.57 >.250

5. Brown X Pearl 115 28 5 1 0.82 1 0.03 >.750

6. Brown X Orange 115 28 21 7 1.36 1 0.42 >.500

7. Brown X Yellow 115 28 14 7 2.05 1 2.00 >.100

8. Pearl X Orange 5 1 21 7 1.67 1 0.19 >.500

9. Pearl X Yellow 5 1 14 7 2.50 1 0.59 >.250

10. Orange X Yellow 21 7 14 7 1.50 1 0.41 >.500

No -...J 2 TABLE 9.3.12. -- Heterogeneity X test to test the homogeneity of the underlying genetic hypothesis over all the localities -- APH locus

APHB/APHB APHA/APHA APHA/APHB 2 Obser- Expec- Obser- Expec- Obser- Expec- X d.L P Locality n ved te2 ved ted ved te2 AA np AB 2npq BB nq

Zoo 603 33 17.42 137 170.17 433 415.41 21.15 1 p <.005

St. Andrew's 63 0 0 0 0 63 63 0 1

Bethel 57 1 0.46 8 9.35 48 47.19 0.85 1 p >.250

Waldron 14 0 0.17 3 2.74 11 11.09 0.19 1 p >.SOO

Pier 27 S6 1 0.20 4 6.32 51 49.48 4.10 1 p <.OSO

Pier 32 58 2 0.28 4 7.55 52 50.17 12.30 1 p <.005 2' Sum of X s 38.59 6 p <.005 2 Xl pooled 851 37 156 658 36.24 p <.005

Heterogzneity 2.35 5 p >.750 Xs

No 00 209

TABLE 9.3.l3~ -- Estimates of p, q, and a of the APH locus for the different localities, obtained from Yasuda's G-type program

Locality p±a q±a a±a p q a

Zoo 0.168±0.018 0.832±0.018 0.189±.048

St. Andrew's 0.000 1.000 no estimate

Bethel 0.088±.028 0.9l2±.028 0.123±.180

H'aldron 0.110±.059 0.890±.059 no estimate

Pier 27 0.054±.024 0.946±.024 0.296±.257

Pier 32 0.069±.028 0.93l±.028 0.463±.227 210

TABLE 9.3.14. -- Test of homogeneity of APR gene frequencies from the different localities

Location Total BP Genes (2n) Genes (x) (x/2n) px

Zoo 1206 1003 0.832 832.49

St. Andrew's 126 126 1.000 126.00

Bethel 114 104 .912 94.64

Waldron 28 25 .893 22.25

Pier 27 112 106 .946 99.64.

Pier 32 116 108 .931 100.44

Totals 1702 1472 1275.46

l:px-pl:x p = .865 X~ = 18.68 p<.005 p q q = .135 q q 211

TABLE 9.3.15. -- Contingency chi-square values and their associated p values of APH phenotype frequencies between a pair of localities

2 Locality X Locality d.£. X P

1. Zoo X St. Andrew's 2 23.85 <.005

2. Zoo X Bethel 2 4.28 >.100

3. Zoo X Waldron 2 .86 >.500

4. Zoo X Pier 27 2 9.75 .025>p>.010

5. Zoo X Pier 32 2 8.92 .025>p>.010

6. St. Andrew's X Bethel 2 10.75 <.005

7. St. Andrew's X Waldron 2 14.05 <.005

8. St. Andrew's X Pier 27 2 5.87 >.050

9. St. Andrew's X Pier 32 2 6.86 .050>p>.025

10. Bethel X Waldron 2 0.69 >.500 n. Bethel X Pier 27 2 1.42 >.250

12. Bethel X Pier 32 2 1.82 >.250

13. Waldron X Pier 27 2 2.73 >.250

14. Waldron X Pier 32 2 3.09 >.100

15. Pier 27 X Pier 32 2 .31 >.750 TABLE 9.3.16. -- Woolf's analysis of pair-wise comparisons of homozygote and heterozygote APR frequencies between two localities

Locality X Locality Romo- Retero- Romo- Retero- 2 X=hK d. f. P zygotes zygotes zygotes zygotes - X Rk (h) (k) (R) (K)

1. Zoo X Pier 32 466 137 54 4 0.25 1 6.84 <.010 2. Zoo X Bethel 466 137 49 8 0.56 1 2.23 >.100 3. Zoo X Waldron 466 137 11 3 0.93 1 0.01 >.900 4. Zoo X Pier 27 466 137 52 4 0.26 1 6.45 .025>p>.010 5. Zoo X St. Andrew's 6. Pier 32 X Bethel 54 4 49 8 2.20 1 1.51 >.100 7. Pier 32 X 1oJa1dron 54 4 11 3 3.68 1 2.45 >.100 8. Pier 32 X Pier 27 54 4 52 4 1.03 1 0.00 9. Bethel X Waldron 49 8 11 3 1.67 1 0.46 >.250 10. Bethel X Pier 27 49 8 52 4 0.47 1 1.37 >.100 11. Waldron X Pier 27 11 3 52 4 0.28 1 2.31 >.100 12. Bethel X St. Andrew's 49 8 01 0 13. Pier 32 X St. Andrew's 54 4 01 0 14. Waldron X St. Andrew's 11 3 01 0 15. Pier 27 X St. Andrew's 52 4 01 0

N I-' N TABLE 9.3.17. -- Woolf's analysis to test for differences in homozygote and heterozygote APH frequencies between the sexes, with the data subdivided by localities

Alkaline Phosphatase Locality Hale Female 2 Homo- Hetero- Homo- Hetero- X=hK d.f. X P zygotes zygotes zygotes zygotes 1IK (h) (k) (H) (K)

Bethel 26 5 15 2 .69 1 .17 >.500

Pier 27 18 2 21 2 .86 1 .02 .900

Pier 32 16 3 23 1 .23 1 1.48 >.100

Deviation xi = 0.00

Heterogeneity X; = 1.67

.....N Vol TABLE 9.3.18. -- Woolf's analysis to test for differences in homozygote and heterozygote APH frequencies between ages, with data from the Bethel location

Alkaline Phosphatase Mature Locality Young 2 Homo- Hetero- Homo- Hetero- X=hK d.L X P zygotes zygotes zygotes zygotes ilk (h) (k) (H) (K)

Bethel 7 1 41 7 1.19 1 .02 .900

.-.N .s::-- 215

2 TABLE 9.3.19. -- Contingency X valaes and their associated p values of APH phenotype frequencies within the sexes, with data subdivided by localities

2 Sex APHA/APHA APHA/APHB APHB/APHB P X2

Bethel

Male 1 5 27 1.29 > .500 Female a 2 15

Pier 27

Male 1 2 17 1.22 > .500 Female a 2 21 Pier 32

Male 1 3 15 1.77 > .250 Female 1 1 22 216

2 TABLE 9.3.20. -- Contingency X values and their associated p values of APH phenotype frequencies within ages, with data subdivided by localities

2 Age APHA/APHA APHA/APHB APHB/APHB P X2

Bethel

Young 0 1 7 .20 > .900 Mature 1 7 40

Pier 27

Young 0 0 21 3.88 > .100 Mature 1 4 25

Pier 32

Young 0 0 16 2.55 > .250 Mature 2 4 36 217

2 TABLE 9.3.21. -- Contingency X values and their associated p values of APH phenotype frequencies within shank phenotypes, with data subdivided by localities

Shank 2 Phenotype APHA/APHA APHA/APHB APHB/APHB P X4

Bethel

++ 0 0 4 + - 0 3 14 9.93* <.050

1 5 29

Pier 27

++ 0 1 5 + - 1 2 14 3.85 >.100

0 1 27

Pier 32

++ 0 0 11 + - 2 4 28 9.72* <.050

0 0 13 218

TABLE 9.3.22. -- Contingency table to test for homogeneity of APH phenotype frequencies over the years 1969 - 1972 -­ zoo

Year APHA/APHA APHA/APHB APHB/APHB Observed Observed Observed

1969 10 33 140

1970 7 14 81

1971 14 77 180

1972 2 13 32

2 = 12.88 .050

TABLE 10.1.1. -- 2 X 2 contingency table to test for inter­ action between shank phenotype and sex -- zoo

Shanks Sex Feathered Non-Feathered Observed Observed

Male 14 160

Female 21 141

= 1.68* p>.250

TABLE 10.1.2. -- 2 X 2 contingency table to test for inter­ action between shank phenotype and age -- zoo

Shanks Age Feathered Non-Feathered Observed Observed

Young 15 139

Mature 35 297

.01* p = .995

*2x2 with Yate's correction 220

TABLE 10.1.3. -- Contingency table of shank phenotype frequencies over the years 1969 - 1972 zoo

Year Feathered Non-Feathered Observed Observed

1969 9 82

1970 26 150

1971 14 157

1972 1 46

2 8.044 X3 = p< ·.050 221

TABLE 10.1.4. -- Contingency table of shank phenotype frequencies within localities

Shank Phenotype Locality ++ +-

1. Zoo 15 53 130

2. St. Andrew's 4 36 33

3. Bethel 6 40 17

4. Waldron 2 8 10

5. Pier 27 6 30 21

6. Pier 32 12 39 15

2 XlO = 60.862 p<.005 222

TABLE 10.1.5. -- Test of homogeneity of shank allele frequencies between localities

Location Total A genes P px Genes (2N) (X) (x/2N)

Zoo 396 83 .210 17.430

St. Andrew's 146 44 .301 13.244

Bethel 126 52 .413 21. 476

Waldron 40 12 .300 3.600

Pier 27 114 42 .368 15.456

Pier 32 132 63 .477 30.051

Totals 954 LX = 296 LPx = 101. 257

p = .310 x~ = Epx-pEx/pq = .214 = 44.379 p<.005 q = .690 223

TABLE 10.1.6. -- Contingency table of shank phenotype frequencies within localities, with the zoo excluded

Locality Shank Phenotypes ++ + ­ Observed Observed Observed

St. Andrew's 4 33 36

Bethel 6 17 40

Pier 27 6 21 30

Pier 32 12 15 39

2 = 13.09 .050 > p > .025 X6 224

TABLE 10.1.7. -- Estimates of p, q, and a of the shank locus with separate estimates for each locality

Locality p ± crp q ± crq a ± cra

Zoo .2l0±.022 .790±.022 .192±.079

St. Andrew's .30l±.034 .699±.034 -.171±.105

Bethel .413±.036 .587±.036 -.310±.116 lValdron .300±.074 .700±.074 .048±.227

Pier 27 .368±.042 .632±.042 -.13l±.128

Pier 32 .477±.039 .533±.039 -.184±.12l TABLE 10.1.8. -- Contingency chi-square values and their associated p values of shank phenotype frequencies within the sexes, with the data subdivided by localities

Sex 2 Hales Females d. f. X P Locality Shanks Observed -t+ +- -- -t+ +---

St. Andrew's 1 17 19 2 15 16 2 0.49 > .750

Bethel 3 11 19 1 4 14 2 1.35 > .500

Pier 27 4 7 11 2 11 10 2 1.58 > .250

Pier 32 3 2 16 6 7 15 2 2.87 > .100

N N lJ1 TABLE 10.1.9. -- Contingency chi-square values and their associated p values of shank phenotype frequencies within ages, with the data subdivided by localities

Age 2 Locality Young Mature d.f. X P Shank Phenotypes Observed +++- - - ++ + -

St. Andrew's 1 1 2 3 32 34 2 3.33 >.100

Bethel 2 3 8 4 14 32 2 0.69 >.500

Pier 27 1 6 14 5 14 15 2 3.00 >.100

Pier 32 5 4 9 7 11 30 2 1.60 >.250

N N 0\ TABLE 10.1.10. -- Woolf's analysis to test for differences in homozygote and heterozygote frequencies between the sexes, with the data subdivided by localities

Shank Locality Male Female 2 Homo- Hetero­ Homo- Hetero­ X=hK d.L X P zygotes zygotes zygotes zygotes Hk

1. St. Andrew's 20 17 18 5 .980 1 0.00

2. Bethel 22 11 15 4 .533 1 0.87 >.250

3. Pier 27 15 7 22 11 1.070 1 0.01 >.900

4. Pier 32 19 2 21 7 3.170 1 1. 79 >.100

Deviation X~ = 0.00

Heterogeneity X; = 2.68 p>.250

N N "-.J TABLE 10.2.1 -- Maximum likelihood frequencies of the alleles of the red plumage locus for each sex with the data subdivided by localities

A B + b Red Wild Brown (p±a ) (q±a ) (r±a.) Locality p q r Hale Female Male Female Hale Female Hale Female Hale Female Male Female

Zoo 14 8 25 16 2 2 .37±.06 .31±.05 .46±.08 .61±.05 .17±.08 .08±.OS

St. 1 2 34 28 0 1 .0l±.01 .07±.02 .99±.01 .90±.03 0 .03±.02 Andrew's

Bethel 2 1 30 24 0 2 .03±.02 .04±.03 .97±.02 .89±.03 0 .07±.03

Waldron

Pier 27 1 0 26 29 0 0 .02±.02 0 .98±.02 0 0 0

Pier 32 0 0 29 35 0 0 . 0 0 1.0 1.0 0 0

N N (Xl 229

TABLE 10.2.2 -- Haldane's test for deviations in phenotype frequencies from the expected in a dia11e1ic sex-linked locus with dominance

2 1 p Dilute Allele Frequency Xl 011 110- d p

A- a = 40 .06±.04 .55 -650 ±249.19 -2.15 <.005 aa d = 1

Aw e = 26 aw f = 0 230

TABLE 10.2.3. -- Contingency chi-square values and their associated p values of red phenotypes within the sexes, with the data subdivided by localities

Red Phenotype 2 Locality Red Wild Brown d.£. X P (blue) Sex

1. Zoo

Male 14 25 2 Female 8 16 2 2 .27 >.750

2. St. Adnrew's

Male 1 34 0 Female 2 28 1 2 1.67 >.250

3. Bethel

Male 2 30 0 Female 1 24 2 2 2.59 >.250

4. Waldron

Male 1 8 0 Female 1 11 0 1 .289 >.500

5. Pier 27

Male 1 26 0 Female 0 29 0 0 .001 >.900

6. Pier 32

l1ale 0 29 0 Female 0 35 0 0 231

TABLE 10.2.4. -- Maximum likelihood estimates of the pattern a11e frequencies for each locality

T Locality C C c (p±a ) (q±a ) (r±a ) p q r

Zoo .17±.03 .2S±.OS .SS±.OS

St. Andrew's .40±.OS .2S±.06 .3S±.06

Bethel .27±.OS .24±.OS .49±.O6

\.;ra1dron .24±.O7 .20±.OS .S6±.10

Pier 27 .39±.OS .20±.OS .41±.O6

Pier 32 .33±.OS .36±.O6 .31±.O6 TABLE 10.2.5. -- Contingency chi-square values with their associated p values of pattern phenotypes within two localities

T T C _ C- CC C _ C- CC Locality X Locality Black Blue Blue Black Blue Blue 2 Check Check Bar Check Check Bar d.f. X P Observed Observed Observed Observed Observed Observed

1. Zoo X St. Andrew's 15 9 16 39 14 16 2 10.51 .010>p>.005

2. Zoo X Bethel 15 9 16 31 16 13 2 4.00 >.100

3. Zoo X '.Ja1dron 15 9 16 8 5 6 2 0.39 >.750

4. Zoo X Pier 27 15 9 16 35 14 10 2 7.09 .05>p>.025

5. Zoo X Pier 32 15 9 16 34 23 7 2 12.12 <.005

6. St. Andrew's X Bethel 39 14 8 31 16 13 2 2.23 >.250

7. St. Andrew's X \va1dron 39 14 8 8 5 6 2 4.07 >.100

8. St. Andrew's X Pier 27 39 14 8 35 14 10 2 .41 >.750

9. St. Andrew's X Pier 32 39 14 8 34 23 7 2 2.53 >.250

10. Bethel X Waldron 31 16 13 8 5 6 2 .86 >.500

11. Bethel X Pier 27 31 16 13 35 14 10 2 .76 >.500

12. Bethel X Pier 32 31 16 13 34 23 7 2 3.07 >.100

13. Waldron X Pier 27 8 5 6 35 14 10 2 2.31 >.250

N 14. Waldron X Pier 32 8 5 6 34 23 10 2 4.74 .100>p>.050 w N 15. Pier 27 X Pier 32 35 14 10 34 23 10 2 2.53 >.250 TABLE 10.2.6. -- Contingency table of 5 major plumage color phenotypes within the localities, zoo excluded

Locality Plumage Color Red Blue Bar Blue Check Black Check Brown Check Observed Observed Observed Observed Observed

St. Andrew's 3 8 12 39 1

Bethel 3 8 15 19 2

Pier 27 1 9 10 24 0

Pier 32 0 7 17 24 0

2 X12 = 14.12 p>.250

N W W 234

TABLE 10.3.1. -- Contingency table to test for interaction between eyecolor and plumage color phenotypes

Plumage Eyeco1or Color Red Brown Pearl Orange Yellow Observed Observed Observed Observed Observed

White 13 210 3 16 11

Colored 8 12 8 12 14

2 = 89.68 p <.005 X4

Table 10.3.2. -- Contingency table of eyeco1or phenotype frequencies over the years 1969-1972, the numbers in parentheses are phenotype frequencies

Year Red Brown Pearl Orange Yellow Observed Observed Observed Observed Observed

1970 15 (.11) 87 (.67) 5 (.04) 6 (.05) 17 (.13)

1971 7 (.04) 121 (.72) 4 (.02) 28 (.17) 8 (.05)

1972 2 (.04) 31 (.66) 3 (.06) 7 (.15) 4 (.09)

2 Xa = 23.83 p <.010 TABLE 10.3.3 -- Chi-square values and their associated p values of eyecolor phenotype frequencies within the sexes, with data subdivided by localities

Male Female Locality Eyecolor Observed Eyecolor Observed 2 Red Brown Pearl Orange Yellow Red Brown Pearl Orange Yellow d.L X p

St. Andrew's 1 0 1 34 1 1 3 5 20 2 4 9.17 .10>p>.050

Bethel 3 1 5 22 2 1 2 6 8 2 4 4.52 >.250

Pier 27 0 1 10 4 8 0 1 3 5 5 4 2.53 >.500

Pier 32 0 0 0 5 11 1 1 1 5 13 4 2.54 >.500

~ w V1 236

TABLE 10.3.4. -- Contingency table to test for homogeneity of eyecolor phenotypes between localities

Locality Eyecolor Red Brown Pearl Orange Yellow Observed Observed Observed Observed Observed

St. Andrew's 2 3 6 54 3

Bethel 5 3 11 32 4

Pier 27 0 1 4 17 9

Pier 32 1 1 3 13 26

66.60 p <.005 237

TABLE 10.4.1. -- Contingency table to test for differences in male and female frequencies between localities

Locality Hale Female

Zoo 212 190

St. Andrew's 33 28

Bethel 32 17

Pier 27 23 22

Pier 32 18 26

2 Xs = 5.66 p >.250

Table 10.4.2. -- Contingency table to test for differences in young and mature frequencies between localities

Locality Young Mature

Zoo 194 397

St. Andrew's 4 60

Bethel 13 47

Pier 27 23 32

Pier 32 16 44

2 X = 25.33, p < .005 4 238

TABLE 11.4.1. -- Variances and f values for the transferrin locus

Variance Genetic Sampling

.0018 .0008 .0081 .1528 .1597 .1596

Table 11.4.2. -- Variances and f values for the 6-phosphog1uconate dehydrogenase locus

Variance Genetic Sampling

.0115 .0007 .0001 -.0078 -.0076 .0078

Table 11.4.3. -- Variances and f values for the alkaline phosphatase locus

Variance Genetic Sampling

.0021 .0008 .0006 .1725 .1730 .1947 239

TABLE 11.4.4. -- Correlation coefficients over the different distances measured by steps for the transferrin locus

Steps Locality Locality Pi Pj r

One Zoo X St. Andrew's .310 .289

St. Andrew's X Bethel .289 .317 r 1=·345 Bethel X ~oJa1dron .317 .238

Ha1dron X Pier 27 .238 .339

Pier 27 X Pier 32 .339 .525

Two Zoo X Bethel .310 .317

St. Andrew's X Ha1dron .289 .238 r 2=-·267 Bethel X Pier 27 .317 .339

Ha1dron X Pier 32 .238 .525

Three Zoo X Ha1dron .310 .238

St. Andrew's X Pier 27 .289 .339 r =-·500 3 Bethel X Pier 32 .317 .525

Four Zoo X Pier 27 .310 .339

St. Andrew's X Pier 32 .289 .525 r =-·667 4 240

TABLE 11.4.5. -- Correlation coefficients over the different distances measured by steps for the 6-phosphog1uconate locus

Steps Locality Locality Pi Pj r

One Zoo X St. Andrew's .763 .888

St. Andrew's X Bethel .888 .935 r =·889 1 Bethel X lVa1dron .935 .932

Waldron X Pier 27 .932 .914

Pier 27 X Pier 32 .914 .933

Two Zoo X Bethel .763 .935

St. Andrew's X Waldron .888 .932

Bethel X Pier 27 .935 .914 r 2=-·385 lo1a1dron X Pier 32 .932 .933

Three Zoo X Waldron .763 .932

St. Andrew's X Pier 27 .888 .914 r =-·100 3 Bethel X Pier 32 .935 .933

Zoo X Pier 27 .763 .914 Four r 4=·750 St. Andrew's X Pier 32 .888 .933 241

TABLE 11.4.6. -- Correlation coefficients over the different distances measured by steps for the alkaline phosphatase locus

Steps Locality Locality Pi Pj r

One Zoo X St. Andrew's .168 0

St. Andrew's X Bethel 0 .090

Bethel X Waldron .090 .110 r =-·762 l Waldron X Pier 27 .110 .060

Pier 27 X Pier 32 .060 .070

Two Zoo X Bethel .168 .090

St. Andrew's X Waldron 0 .110 r 2=-·459 Bethel X Pier 27 .090 .060

'valdron X Pier 32 .110 .070

Three Zoo X 'valdron .168 .110

St. Andrew's X Pier 27 0 .060 r 3=-.971

Bethel X Pier 32 .090 .070

Four Zoo X Pier 27 .168 .060

St. Andrew's X Pier 32 0 .070 r =-·500 4 TABLE 11.4.7. -- Regression and correlation coefficients, and t values, for the transferrin, 6-phosphogluconate dehydrogenase, and alkaline phosphatase loci

Regression Correlation SS Regression System Coefficient Coefficient y.x b Equation b a r t = b/Sb t=r/(l-r) 2/N-2

Transferrin 0.014 0.2863 0.254 0.087 .009 Y=0.286 + 0.014X 1.562 ---0.682

6-Phosphogluconic 0.002 0.776 0.000 0.249 .027 Y=0.776 + 0.002X Acid Dehydrogenase .070

Alkaline Phosphatase 0.022 0.842 0.692 0.037 .004 Y=0.842 + 0.022X 5.500** 4.49**

r-J ~. r-J 243

TABLE 11.4.8. -- Variances and f values for the shank locus

Location f P S

Zoo .193 .210

St. Andrew's .172 .301

Bethel -.309 .413

Waldron .048 .300

Pier 27 -.132 .368

Pier 32 -.184 .477

Variance Genetic Sampling

.010 .002 .001 -.026 -.025 -.009 244

TABLE 11-•.4.9. -- Correlation coefficients over the different distances measured by steps for the shank locus

Steps Locality Locality Pi Pj r

One Zoo X St. Andrew's .210 .301

St. Andrew's X Bethel .301 .413

Bethel X Waldron .413 .300 r =·232l l Waldron X Pier 27 .300 .368

Pier 27 X Pier 32 .368 .477

Two Zoo X Bethel .210 .413

St. Andrew's X Waldron .301 .300

Bethel X Pier 27 .368 .413 r 2=-·2574 Waldron X Pier 32 .300 .477

Three Zoo X Waldron .210 .300

St. Andrew's X Pier 27 .301 .368 r 3=·998 Waldron X Pier 32 .413 .477

Four Zoo X Pier 27 .210 .368 r 4=·980 St. Andrew's X Pier 32 .301 .477 TABLE 11.4.10. -- Regression table giving the regression and correlation coefficients and their t values

System Regression Correlation Coefficient -y Coefficient b a r Sy.x Sb t d. f.

Shank .042 .345 .763 .056 .006 7.000** 4 .005>p>.001

Regression Equation

Y = .345 = .042X

~ ~ lJ1 246

TABLE 11.4.11. -- Variances and covariances found for the alleles of the pattern locus

Gene Variance Frequencies or f ST Concerned Covariance

p .0078 .0358

q .0029 .0151

r .0104 .0427

p.q. -.0033 .0397

p.r. -.0071 .0528

q.r. -.0026 .0238 247

TABLE 11.4.12. -- Correlation coefficients over the different distances measured by steps for the alleles of the pattern locus

Steps Locality X Locality pi pj qi qj ri rj Correlation Coefficient

One Zoo X St. Andrew's .17 .40 .25 .25 .58 .35 r =-.638 p St. Andrew's X Bethel .40 .27 .25 .24 .35 .49

Bethel X Waldron .27 .24 .24 .20 .49 .56 r =-.369 q Ha1dron X Pier 27 .24 .39 .20 .20 .56 .41

Pier 27 X Pier 32 .39 .31 .20 .36 .41 .33 r =-.241 r Two Zoo X Bethel .17 .27 .25 .24 .58 .49 r =-.272 P St. Andrew's X Ha1dron .40 .24 .25 .20 .35 .56

Bethel X Pier 27 .27 .39 .24 .20 .49 .41 r =-.926 q Ha1dron X Pier 32 .24 .31 .20 .36 .56 .33 r =-.649 r Three Zoo X Waldron .17 .24 .25 .20 .58 .56 r = l.00 p St. Andrew's X Pier 27 .40 .39 .25 .20 .35 .41 r =-.141 q Bethel X Pier 32 .27 .31 .24 .36 .49 .33 r =.542 r Four Zoo X Pier 27 .17 .39 .25 .20 .58 .41 r =-1. 00 p St. Andrew's X Pier 32 .40 .31 .25 .36 .35 .33 r = 0 q r = 1.00 r 248

TABLE 11.4.13 -- Observed and expected correlation coefficients between alleles of the pattern locus and their associated Z values to test for dif- ferences between observed and expected

Gene Frequencies Expected Observed Z P Concerned r r

p.q. -.407 -.694 -.714 .269

p.r. -.584 -.788 -.708 .260

q.r. -.504 -.473 .068 .473 2.49

TABLE 11.6.1. -- Plumage color phenotype frequencies in the zoo population

Plumage Color n Freq.

White 554 .841

Red 20 .030

Blue Bar 16 .024

Blue Check 9 .014

Black Check 15 .023

Brown 4 .006

Grizzle 40 .061

Dunn 1 .001

Total 659 250

TABLE 11.6.2. -- Observed and expected frzquencies of the various matings used in calculating the X obtained

Pair Phenotype Hhite X \Vhi te \Vhite X AOC Colored X Colored

Observed 87 15 2

Expected 73.52 27.85 2.64

2 13.62 p<.005 X2 = 251

TABLE 13.3.1. -- The different mating types and numbers and phenotypes of offspring produced from each type -- the transferrin locus

Offspring 2 Hating Type n Totals d.f. X P Male X Female AA AB BB

TfrA/TfrAX TfrA/TfrA 0 0 0 a a a .00 .000

TfrA/TfrAX TfrA/TfrB 3 a 5 a 1 .58 >.250 TfrA/TfrBX TfrA/TfrA 1 2 a a J TfrA/TfrBX TfrB/TfrB 14 a 9 17 26 1 1. 88 >.100 TfrB/TfrBX TfrA/TfrB 18 a 13 17 30 1 .30 >.500

TfrA/TfrBX TfrA/TfrB 13 6 10 6 22 2 .18 >.900 TfrA/TfrAX TfrB/TfrB 10 a 20 a 20 a 0 .000

TfrB/TfrB TfrA/TfrA 3 a 6 a 6 a a .000 TfrB/TfrB TfrB/TfrB 22 a 37 a 37 a a .000 TABLE 13.3.2. -- The different mating types and numbers and phenotypes of offspring produced from each type -- 6-phosphogluconate dehydrogenase locus

Hating Types Offspring 2 Hale Female n 6PGDA/6PGDA 6PGDA/6PGDB 6PGDB/6PGDB Totals d .f. X P Obs. Exp. Obs. Exp. Obs. Exp.

6PGDA!6PGDA 6PGDA/6PGDA 16 31 31.00 00 00.00 a 0.00 31

6PGDA/6PGDA 6PGDA/6PGDB 14 10 13.00 16 13.00 a 0.00 26 1 0.96 >.250 6PGDA/6PGDB 6PGDA/6PGDA 23 13 17.50 22 17.50 a 0.00 35 1 1. 83 >.100

6PGDA/6PGDB 6PGDA/6PGDB 18 6 8.50 19 17.00 9 8.50 34 2 1.04 >.500

6PGDA/6PGDB 6PGDB/6PGDB 5 a 0.00 5 3.50 2 3.50 7 1 0.56 >.250 6PGDB/6PGDB 6PGDA/6PGDB 3 a 0.00 4 3.00 2 3.00 6 1 0.16 >.500 6PGDA/6PGDA 6PGDB/6PGDB 3 a 0.00 6 6.00 a 0.00 6 6PGDB/6PGDB 6PGDA/6PGDA 2 a 0.00 4 4.00 a 0.00 4

Totals 84 60 76 149

N \J1 N TABLE 14.3.3. -- The different mating types and numbers and phenotypes of offspring produced from each type -- the alkaline phosphatase locus

Offspring Mating Types 2 Hale Female n APHA/APHA APHA/APHB APHB/APHB Totals d.£. X P Obs. Exp. Obs. Exp. Obs. Exp.

APHB/APHB APHB/APHB 26 0 0.0 0 00.00 43 43.00 43

APHB/APHB APHA/APHB 12 0 0.0 8 10.50 13 10.50 21 1 0.76 >.250

APHA/APHB APHB/APHB 17 0 0.0 16 15.50 14 15.50 30 1 0.04 >.750

APHA/APHA APHB/APHB 6 0 0.0 9 9.00 0 00.00 9

APHB/APHB APHA/APHA 7 0 0.0 12 12.00 0 00.00 12

APHA/APHA APHA/APHB 2 1 2.0 3 2.00 0 00.00 4 1 0.26 >.500

APHA/APHB APHA/APHA 4 5 4.0 3 4.00 0 00.00 8 1 0.12 >.500

APHA/APHB APHA/APHB 13 7 5.25 9 10.50 5 5.25 21 2 0.80 >.500

APHA/APHA APHA/APHA 2 4 4.0 0 00.00 4

Totals 89 7 60 75 152

N VI W 254

Figure 2.1. A Tiselius electrophoresis cell and electrode assembly (Reprinted from Shaw, 1969) 255

A

Figure 2.2. E1etrophoretic pattern (descending) of human plasma. A-albumin; a , a , y-g1obu1ins; -fibrinogen; 1 2 S, E-sa1t boundary (Reprinted from Shaw, 1969) 256

E;~:!rco~:Tl";S:': -----...

Figure 2.3. The counteracting effects of electroosmosis and electrophoresis (Reprinted from Shaw, 1969) 257

Figure 2.4. Horizontal Starch Gel Electrophoresis apparatus. A-gel; B-sample; C-gel plate; D-saran wrap or petroleum jelly seal; E-sponge wicks; and F-connecting plugs. '. ,-...... ""\ ; . . .' - ;\:(.;. :.':' . ',:.; .; -" '. ''''.:.'. '. 'f . .':.: \".7...:. .'~" "'" . "..~.. /'n : ,'<>./ : .. ::. .2/ I ~'.. ~l ~ ~.I . '.~" '., . '~'" 1.\'C"/./,~i,.. ... II: >''':''-~""0:;"'.-...... •... <'/".... \ \ I' .'. ~ ;.; <::"f ...... i. '. ..'.,c-] I v",. . '. .. '.,.:.'.'. .1 ~ ~,~ >'u'''1~~"".'1: .. .. ,... .•• , - ~. ~,. U -~- ~~.J..-z' ~, '1' . .< • '"'''''' '. '--:::.J .., /' "" :". ',' .' "." .'.: ';\~'- ,- ., (~~.' ; ...:---=---:-.""'''--~d\ , r"~ur "--'""1 ,....'.~_.. ------' " '.1'"''<.'"~L. . (\ '. 1J ~ <~. "'I'tl'. ' c,- ZOO. ---..~- :,,~ r. , . ., '..' Z, .\ \ ;, A'~=---v.' ':-/\ (> '-- ==-:c.6.~·C.. ... /. ,{ 0.· \ " <' .1. '.;' ;;! .~ ·~,""~·~ :~';:'~ ~ H" '< (..,. \\ ~~=-==~~.~~_\"',.-;..~1'1.. --"~-.... ~...'i··-OOj ,. '.". J k. . \. .., .. I11 '.:: _ .\\.'0 '". ., < /.'''7~------=_~., '" ~ .•. " ',. '\"------.,,.. . . •". . \ .. \i' '~%:'~'\\~',r; ,,...,, "'--' _. ('~J"'~'/.. ::; \,'.\~:. ", ~• '" .' " \\.\ £."~""'C'E /.<.. .. -" J >'. -...'''~ . '.\\ '. .' .~. \~'\.\~J :'. ,\. r.. l .1,1•, ~ \ L . '.:'-.'·'·"'X'.~- 1-.i"@:0'''''- '#/7; ,\""\,-,..\%•.\. \\\\...,~ \\ /; 11. . . " h ",'. ,I j r:::::T1\ I -.::: -..... ; .. : c\. -/., , '\\ \.\'. OJ' • • 'I '..--r:::., 1 - '.'':_. . ;\ i 1'...... I I~I .~\ ,-I:)-';i/::,:~:.,~'~~'~'-l", -,~ /:;~,• I\~'I\\:'"" '..'."Y.~\ ~ .<'. '~','" .•~I)I~" • ~. f~ ; r I0.'\\ ,; "",c-':-" ,. \, ..-'\ '"\\\.....\ /'8.11, . •. ,0.• '.i,...... "::'.''. 1:lji' .. • < " • "', "\:". . •. '. , C '1/ ~ qlj' ':..:J.... " l.-l~\., " , . -", 'r \, ~·~-~0·,. •••• -' I~ _. II '._. I .j • '.\-• "'''''.dj/,,,", CJ Om", :un.>:. "j ~i 7 ~'\ -:" 'I .,...... :II '. 1'1 '"t.....:. ,' ".':,\<\-\~~"",'. , . ---.\\ll..~' '. , ".,c,.. " ...'. " . "'," Ii (. ~~<:~1 //~"""''\', .<~~~! ~~.r<7.- ~j-~_. ~x~. j".h. . iii:. I! I / <'~.,,; ". /)':"',ill I, , ", '«' / /'" .'VI .'C.--..--'- . "" f. '.. J I. . i j '.,""~','" /. /• ~'.,'''",- •. ,v,••-2..~~"" '..c I. I J, .""·'\~«)Y,.f!J~;:"/. <-, ., .~~':'$ ~.",.~ ....• ': ' .•!, : !II'~ f . '~",,';;).,~ "";,-...;,~ C:>~ ~ . " -'. ' ,~,/: j " I ~e.,.. ,.. t ''IIfill d . "'''':~ .. "~~- __<..~.__. ~. ;...,~~;,,:". j . . '\'~" <:·~:.~.~··:·jw'?~.,.. ·4 f': I J ,. . .,.. ,• """'-r ...... ~""'-_. _ -' . ••. ~.····.·J:: '...... ' • '. '. '.•' . j I ". ".., "': '. .. . _..,. 7/ '. ~ . _~"""__ ,. ""'" '"I . , """",-....",~.,.,."". . I I • x' .. ! 1' . .: . 'Ilr' ----.'• .' ~ ".'". " ..N .. ,'.' .. '• 1 .neV 1thin the Zoo cotlp ...... " Figure 4.1.1. A map of the Honolulu Zoo depicting the .1 trapping areas, w1th1n'. 'he• Zoo complex SA ND ISLAND

~-----

"

""'- "- "- " "-, o Ind/c:llc3 Fler No

GRA?HIC SCALE IN I'eer

Figure 4.5.1. Map of a portion of the Honolulu Harbor complex 260

1'''{!·'\ltH~ln· • • • >- \lLu~11l D 0 D

~ ---~ - -}- lr"nnh'rrlnll - I""l'rt I.llllC Tfrl~/tfrr.-

Figure 7.1.1. Diagramatic representation of the transferrin phenotypes r-l '"N

_ • ::!""4."

Glu',)'''l,C'''' GLUCOSE ( ::-I..~ t'r'O"':;:',~• ,,--- GS • .'~ .. -. ~..,; ~. '.: AT? I .,.~~ -• -",,-.- 0.4 ,) I).. '\I .-~·,.~.,.V ~ I G: .. ~-:·"'':'''. HeIC_l"'J~e· ~i.J~ ft~ J ;_\ ... ··,;""p· .. '.is .... ADf:' ~'; t-tzO -----/' '\- I ~~', 'PN~ 4.,:; "'ules" • '- • Gl) COlilfn--..G-l-P ----~G,6·~~----.------.-.---. t r"'~\:"~'~',o"c'ac'une- to tl ~· ... .:.\;'~-:.;I\J(\'nt' oJCIJ G·6 P ~~~J~'C"itMSf'''' Phosohohelose I~C"T\tra\e - :0:- I/'~ '!' ;..,~ • 1 ( t.o P G Jctly,,·o.~~r ..ae F-6'P ~\.... : f'''~", AiP '~ .~ ,,\.~~.rt. __ ,,,, .. R· 5 p .- -~~~~~~ t·L ~, p Phosp~~fructck.,ncse ~ r.t) "~~"'tras~ A 1 l'''.~l''';)!' ~"e'~~~I"'~f'J\.~ ADp.1 ~'b P• E 4 P Trcr:.s- rs i P 1,~n ...Il.e"ola.. ~ t lG~p 5 F-l:6- P .,. l '-a ,jJ1Qo:.t 4---. ' ------... '(r r. AJdelesr1 ~ ...I~-.___-- Tr~n5"~I:..ll------"..he r---p-'-'-'-P-M-'-'j-O-,,- ..------OHAP ------. GAP

iso",.,ooo ~P~, OPN +_ -Hb

DPfHf - me'hemOQio~)lr; ffdue'ose· Fod:::d~ Kno • ., el'll.,me Jtflclercl.' --"'Hb ...... O.-d.3"' nr \~t1\1r J'e Jd(l~d ...... - T,c<. ~CI•.,." 2. 3.QoG mutase- ... / t,3-0PG +- - -~ ·r"':. 00:',,,,,, •., 't',cu'~".~1o --- I ""P 2,3-0PG ~___ (-P'lcsphoQlyceric I1Cld kinose

2,3-0PG pI\O$p/lOl= ~ ~ ATP 3-PG Kreb1 cycle .., 3·PG mulo.e 1 I AOP ATP OPNH OPN

2'PG~ PEP~-L..- PyruviC acid '---'<~. " lot'ic aeid P,rUw i( ceid Leetle aCtd ~lnr:~e" de~1C'rogencse Figure 8. 1.1. Embden-Meyerhoff and pentose-phosphate pathways (Re­ printed frOB Unis and Yasmineh, 1969). Pathways are shown proceed- ing in one direction for simplicity. 262

T-Pattern Black Checker

Blue Checker

Blue Bar

f. livia Standard

Figure 10.2.1. The three plumage pattern types found 263

Figure 11.3.1. One dimensional stepping stone model of Kimura and Weiss (Reprinted from Kimura and Weiss, 1964) 264

1.0

.9

.8

.7

.6

.5

.4 0 I-l .3 +J l:l Q) "M .2 u 'M 4-l 4-l .1 Q) 0 u 0 l:l 0 'M _ +J .1 til .-l Q) - I-l .2 I-l 0 o u_.3

-.4

-.5 o

-.6 o -.7

-.8

-.9

-1.0 1 2 3 4 Steps

Figure 11.4.1. Graph illustrating the correlation of Tfr allele frequencies with distance (steps) 265

1.0

.9 0 .8

.7 0 .6

.5

.4

I-l .3 -l-l l:: Q) .2 OM (J OM 4-l .1 4-l Q) u0 a l:: 0 OM -.1 0 -l-l CIl .-l Q) -.2 I-l I-l u0 -.3 0 -.4

-.5

-.6

-.7

-.8

-.9

-1.0

1 2 3 4 Steps

Figure 11.4.2. Graph illustrating the correlation of 6PGD allele frequencies with distance 266

1.0

.9

.8

.7

.6

.5

.4

~ .3

.I-J l=: Q) .2 .,-l C) .,-l ~ .1 ~ Q) 0 u 0 l=: 0 ',-l -.1 .I-J III r-f Q) -.2 ~ ~ 0 u -.3 -.4 o -.5 o

-.6

-.7 0 -.8

-.9 o -1. 0

1 2 3 4 Steps

Figure 11.4.3. Graph illustrating the correlation of APH allele frequencies with distance 267 268

1.0 0 0

.9

.8 .7

.6

.5

.4

P- .3 I-l +J Q Q) .2 'M U 'M 4-l .1 4-l Q) u0 a Q 0 'M -.1 +Jro r-l Q) -.2 I-l I-l 0 u0 -.3 -.4

-.5

-.6 0 -.7

-.8

-.9

-1.0

1 2 3 4 Steps

Figure 11.4.5. Graph illustrating the correlation of pattern allele, CT, frequency with distance 269

1.0

.9

.8 .7

.6

.5

.4

C1' .3 l-l .w c::: .2 Q) 'M (J 'M .1 4-l 4-l Q) o 0 0 u c::: 0 -.1 'M o .w Cll r-l -.2 Q) l-l l-l 0 -.3 u 0 -.4

-.5

-.6 -.7

-.8

-.9 o -1.0

1 2 3 4 Steps

Figure 11.4.6. Graph illustrating the correlation of pattern allele, C, frequency with distance 270

1.0 0

.9

.8

.7

.6 0 .5

.4

l-l l-l .3 .w c.: QJ .2 .,-l tJ .,-l .1 4-l 4-l QJ 0 U c.: 0 -.1 ',-l .w C1l r-l QJ -.2 l-l 0 l-l 0 u -.3 -.4

-.5

-.6 0 -.7

-.8

-.9

-1.0

1 2 3 4 Steps

Figure 11. 4.7. Graph illustrating the correlation of pattern allele, c, frequency with distance r--l "N

];.", ••••••••• ------­~------c------:1;1.«'0

Figure 12.3.1. Diagramatic representation of the 9 esterase phenotypes found 272

Figure 12.3.2. Diagramatic representation of the 6 possible 6-Phosphog1uconic Dehydrogenase phenotypes 273

LITERATURE CITED

Aasa, R. and Aisen P., 1968. An electron paramagnetic resonance study of the iron and copper complexes of transferrin. J. BioI. Chern. 243: 2399-3404.

Aasa, R., Malstrom, B. G., Saltman, P., and T. Vangard, 1963. The specific binding or iron and copper to transferrin and conalbumin. Biochem. Biophys. Acta. 75: 203.

Ashton, G. C., 1957. Serum protein differences in cattle by starch gel electrophoresis. Nature, Lond., 182: 193-194.

____, 1958. Further S-globulin phenotypes in sheep. Nature, Lond., 182: 1101-1102.

1959. S-globulin alleles in Zebu cattle. Nature, London, 184: 1135-1136.

______, 1960. Thread protein and S-globulin polymorphism in the serum proteins of pigs. Nature, London, 186: 991-992.

,1965. Serum transferrin D alleles in Australian cattle. ------Aust. J. BioI. Sci. 18: 665-670.

Ashton, G. C. and A. W. H. Braden, 1961. Serum B-globulin poly­ morphism in mice. Aust. {. BioI. Sci. 14: 248-251.

Ashton, G. C. and E. I. Me Dougall, 1958. Beta-globulin polymorphism in cattle, sheep, and goats. Nature, London, 182: 1786-1789.

Baker, C. M. A. and C.Manwell, 1967. Molecular genetics of avian proteins VIII. Egg white proteins of the migratory quail, Coturnix coturnix -- New concepts of hybrid vigour. Compo Biochem. Physiol. 23: 21-47.

Bartlett, G. R. and A. W. Schafer, 1961. phosphorylated carbchydrate intermediates of the human erythrocyte during storage in acid citrate dextrose. II Effect of the addition of inosine late in storage. {. Clin. Invest. 40: 1185-1193.

Beckman, L., and F. M. Johnson, 1964. Variations in larval alkaline phosphatase controlled by APH alleles in Drosophila melanogaster. Genetics 49: 829-835.

Bell, D. J., 1960. Tissue components of the domestic fowl. 4: Plasma alkaline phosphatase activity. Biochem. J. 75: 224.

______, 1964. Plasma acid phosphatase activity and bone dystrophies in the fowl. Camp. Biochem. Physiol. 12: 417-427. 274

______, 1971. Plasma eyzymes in Physiology and Biochemistry of the Domestic Fowl. 2: 965.

Bender, K. and S. Ohno, 1968. Duplication of the autosomally inherited 6-Phosphog1uconate dehydrogenase gene locus in tetraploid species of Cyprinid fish. Biochem. Genetics, 2: 101-107.

Bernfie1d, P., Donahue, V. ~1. and F. Homburger, 1953. Characteristic individual electrophoretic patterns in humans. Proc. Soc. ~. Bio1. and Med. New York 83: 429-434.

Bessmertnaja, S. J., 1928. }mterials on the genetics of pigeons. (translated title from Russian). Transactions of the Lab. of ~~. Biology, Zoopark (Moscow), 4: 410-438.

Bogyo, T. P. and S. W. Ting, 1968. Effect of selection and linkage on inbreeding. Aust.~. Bio1. Sci. 21: 45-53.

Bo1, C. J. A. C., 1926. Genetische analyse van k1euren, veerpatronen, tinten, en afteekeningen bij postduiven. Genetica, 8: 45-154.

Bond, C. J., 1919. On certain factors concerned in the production of eye colour in birds. Jour. of Genetics 9: 69-81.

Bonhote, J. L., 1911. On the inheritance of the web-footed character in pigeons. Zoological Society of London, Proceedings, pp. 14-19.

______, and F. W. Smalley, 1911. On colour and colour pattern i~heritance in pigeons. Zoological Society of London, proceed­ ings, pp. 601-619.

Bo~~an, J. E., P. E. Carson, and Frischer, H., 1966. Genetics of starch-gel electrophoretic variants of human 6-Phosphogluconic dehydrogenase populations and family studies in the United States and Mexico. Nature, London, 210: 811.

Personal interview with Paul Breese, former Honolulu Zoo ------=--:--Director, 1972.

Boyer, S. H. and W. J. Young, 1960. S-globu1in polymorphism in chimpanzees. Nature, London, 187: 1035-1036.

Braend, M., 1964a. Genetic studies on serum transferrins in reindeer. Hereditas, 52: 181-188.

Braend, ~1., 1964b. Serum types of Norwegian horses. Nord. Vet. Med. 16: 363-373.

Braend, ~1. and C. Stormont, 1964. Studies on hemoglobin and trans­ ferrin types in horses. Nord. Vet. Med. 16: 31-37. 275

Brown, R. V. and J. H. Manley, 1970. Serum alkaline phosphatase inheritance in the pigeon. Anim. Blood Grps. Biochem. Genet. 1: 43-45.

Cannings, C. and A. W. F. Edwards, 1969. Expected genotypic frequencies in a small sample: Deviation from Hardy-Weinberg Equilibrium. Am. Jour. Hum. Genet. 21: 245-247.

Carson, H. L. 1968. The population flush and its genetic consequences. Population Biology and Evolution, Syracuse University Press, Syracuse, N. Y., U. S. A.

Carter, N. C., Gould, S. R., Parr, C. W., and P. H. Walter, 1966. Differential inhibition of human red cell phosphog1uconate­ dehydrogenase. Acta. Genet. 18: 109.

Chamberlain, E., 1907. The Homing Pigeon, Manchester, England: The Homing Pigeon Pub. Co.

Chapman, D. C., 1968. Counted data, pp. 417-427 In International Encyclopedia of Social Sciences. Vol. 3, Macmillan, New York.

Christe, W., and C. Wriedt, 1923. Die Vererbung von Zeichnungen, Farben, und anderen Charakteren bei Tauben. Zeitschrift f. induktive Abstammungsund Vererbungs1ehre 32: 233-298.

____, 1927. Schok1ade ein neuer gesch1echtsgebundener Farbencharakter bein Tauben. Zeitschrift f. induktive Abstammungs-und Vererbungs1ehre 43: 391-392.

Cole, L. J., 1912. A case of sex-linked inheritance in the domestic pigeon. Science 36: 190-192.

,1914. Studies on inheritance in pigeons: I. Hereditary ---~ relations of the principal colors. Rhode Island State College Agric. Exper. Station Bull. 158.

___~, 1939. The origin of the domestic pigeon. Proceedings of the Seventh World's Poultry Congress and Exposition, Cleveland, Ohio, pp. 462-466. ---

and L. E. Hawkins, 1930. Studies on inheritance in pigeons --~-:-- VIII. 'Porcupine' pigeons. Jour. of Heredity 21: 51-60.

and W. F. Hollander, 1939. The inheritance of silky plumage ----:--in the domestic pigeon. Jour. of Heredity 30: 197-201.

and F. J. Kelley, 1919. Studies on inheritance in pigeons. ------:--III. Description and linkage relations of two sex-linked characters. Genetics 4: 183-203. 276

Cole, L. J. and R. D. Owen, 1944. Naked pigeons. Jour. of Heredity, 35: 3-7.

Common, R. H., 1936. Serum phosphatases in the domestic fowl. J. Agr. Sci. 26: 492.

Cooch, F. G., and J. A. Beardmore, 1959. Assortative mating and reciprocal differences in the Blue-Snow Goose complex. Nature 183: 1833-1834.

Cooper, D. W., Irwin, M. R., and W. H. Stone, 1969. Inherited varia­ tion in the dehydrogenases of doves (Streptope1ia.) II. Autosomal inheritance of glucose-6-phosphate dehydrogeonase. Genetics, 62: 607-617.

-----;-::--.,1969. Inherited variation in the dehydrogenases of doves (Strepiope1ia.) I. Studies in 6-phosphog1uconate dehydrogenase. Genetics, 62: 597-606.

Cooper, M. and G. G. Sharman, 1964. Transferrin variation in Kangaroo. Nature, London, 203: 1094.

Cotterman, ~. W.~ 1954. Estimation of gene frequencies in nonexperi­ mental populations. Chapt. 35, p. 449-465. In O. Kempthorne, T. A. Bancroft, J. W. Gowen, and J. L. Lush (eds.). Statistics and Mathematics in Biology. Iowa State College Press, Ames.

Craig, W. W., 1908. The voices of pigeons regarded as a means of social control. Amer. J. Sociol. 14: 86-100.

Dahlberg, G., 1929. Inbreeding in man. Genetics 14: 421-454.

Darwin, C., 1897. The variation of animals and plants under domestication. Second ed., Vol. 2, New York: D. Appleton and Co. (First ed., 1868; second ed., first printing, 1875.)

___~' C., 1900. The origin £f species ~ means of natural selection. Sixth ed. New York: A. L. Burt Co., (First ed. -1859. )

Davis, B., Sa1tman, P. and S. Benson, 1962. The stability constants of the iron-transferrin complex. Biochem. biophys. Res. Commun. 8: 56.

Delorenzo, R. J., F. H. Ruddle, 1969. Genetic control of two electrophoretic variants of glucosephosphate isomerase in the mouse (Mus musculus) 3iochem. Genet. J: 151.

Dobzhansky, T. 1951. Genetics and the Origin of Species. 2nd rev. ed. Columbian University Press, New York. 277

Doncaster, L., 1912. Notes on inheritance of colors and other characters in pigeons. Jour. of Genetics 2: 89-98.

E-C Bulletin, 1971. Determination of alkaline phosphatase isoenzymes. Electrophoresis counter-current bulletin 5: 1-2.

Ewens, W. J. and G. Thomson, 1970. Heterozygote selective advantage. Ann. Hum. Genet. 33: 365-376.

Falk, C. T., 1970. The combined effects of positive assortative mating and selection. Heredity 27: 125-136.

Fildes, R. A. and C. W. Parr, 1963. Human red cell phosphogluconate dehydrogenase. Nature, 200: 890.

Fisher, R. A., 1922. On the mathematical foundations of theoretical statistics. Phi. Trans. Roy. Soc. Lond. SeT. A. 222: 390-368.

Frelinger, J. A., 1968. Maintenance of transferrin polymorphism in pigeons. Genetics 68: 521.

Gahne, B., 1963. Genetic variation of phosphatase in cattle serum. Nature 199: 305.

______and J. Rendel, 1961. Blood and serum groups in Reindeer compared with those in cattle. Nature, London, 192: 529-530.

Ghigi, A., 1908. Sviluppo e comparsa di caratteri sessuli secondari in alcuni ucelli. R. Accad. Scieze, 1st. di Bologna, 15 Mar.

Gill, J. L. and B. A. Clemmer, 1966. Effects of selection and linkage on degree of inbreeding. Aust. J. BioI. Sci. 13: 307-317.

Gillespie, J. H. and K. I. Kojima, 1968. The degree of polymorphisms in enzymes involved in energy production compared to that in nonspecific enzymes in two Drosophila ananassae populations. Proc. N.A.S. 61: 582-585.

Gitlin, D. and J. Kitzes, 1967. Synthesis of serum albumin, embryo­ specific a-globulin and conalbumin by the chick yolk sac. Biochem. Biophys. Acta 147: 334-340.

Goodman, M., Kulkarni, A., Poul, K. E. and E. Reklys, 1965. Species and geographic differences in the transferrin polymorphism of the macaques. Science, 147: 884-886.

Goodman, M. and R. C. Wolf, 1963. Inheritance of serum transferrins in rhesus monkeys. Nature, London, 197: 1128. 278

Goodwin, D., 1958. The existence and causation of colour preference in the pairing of feral and domestic pigeons. Bull, Brit. Ornith. Club. 78: 136-139.

Graefe, C. F., 1951. First report on a new sex-linked color. National Pigeon Association Year Book, 1951, pp. 39-40.

Haldane, J. B. S., 1963. Tests for sex-linked inheritGnce on popula­ tion samples. Ann. Hum. Genet., London 27: 107-111.

Hardy, W. B., 1899. On the coagulation of protein by electricity. J. Physiol., London, 24: 288.

Harris, H., 1966. Enzyme polymorphism in man. Proc. R. Soc. B. 164: 268.

Hawkins, L. E., 1931. Studies on inheritance in pigeons X. Relation of chocolate to black and dominant red. Genetics J6: 547-573.

Heinroth, 0., 1910. Beitrage zur Biologie, namentlich Ethologie und Psychologie der Anatinen. Verh. V. Intern. Ornitho1. Kongress, pp. 589-702.

Hickman, C. G., and O. Smithies, 1957. Evidence for inherited dif­ frences in the serum proteins of cattle. Proc. Genet. Soc. Can. 2: 39.

Hodson, A. W., Latner, A. L. and L. Raine, 1962. Isozymes of alkaline phosphatase. C1in, chim. Acta 7: 225.

Hollander W. F., 1937. Cross breeding for better market squabs. Amer. Pigeon Jour. 26: 229 (Aug.).

,1938(a). Inheritance of certain 'blue-black' patterns ---~ and 'bleached' colorations in the domestic pigeon. Genetics 23: 12-23.

,1938(b). A sex difference in linkage intensity of three ----autosomal factors in the domestic pigeon. Genetics 23: 24-27.

_____, 1938(c). Clumsy pigeons. Jour. ~ Heredity 29: 65-66 •

• 1944. Mosaic effects in domestic birds. Quarterly ------Review of Biology 19: 285-307.

,1948. The Yeagley experiments. Amer. Racing Pigeon News. ---~- Nov., pp. 1-3.

______-' 1954. Pigeon toes are perfectly simple. Amer. Pigeon Jour. 43: 110. 279

Hollander, W. F. and L. J. Cole, 1940. Somatic mosaics in the domestic pigeon. Genetics 25: 16-40.

and R. D. Owen, 1939. Iris pigmentation in domestic ----pigeons. Genetica 21: 408-419

Horlacher, W. R., 1930. Studies on inheritance of pigeons, VII. Inheritance of red and black color patterns in pigeons. Genetics 15: 312-346.

Hunter, R. L. and M. S. Burstone, 1958. The zymogram as a tool for characterizing enzyme substrate specificity. J. Histochem. Cytochem. 6: 396.

Jamieson, A., 1965. The genetics of transferrins in cattle. Heredity, 20: 419-441.

Jones, H. D. C and D. J. Perkins, 1965. Metal-ion binding of human transferrin. Biochem, Biophys. Acta 100: 122.

Jones, S. V. H., 1922. Studies on inheritance in pigeons IV. Checks and bars and other modifications of black. Genetics 7: 466­ 507. Also partly reprinted in Amer. Pigeon Keeper, Vol. 26, no. 2 (Feb.), pp. 18-22, with editorial comment.

Kalmus, H. and S. M. Smith, 1966. Some evolutionary consequences of pegmatypic mating systems (Imprinting). Am. Nat. 100: 619-635.

Katz, J. H. and J. H. Jand1, 1964. In "Iron Metabolism" (F. Gross, ed.) pp. 103-117. Springer-Verlag, Berlin, Gottingen and Heidelberg.

Kazazian, H. H., 1966. Molecular size studies on 6-phosphogluconate dehydrogenase. Nature 212: 197.

, Young, W. J. and B. Childs, 1965. X-linked 6-phospho­ ----gluconate dehydrogenase in Drosophila. Subnit associations. Science, 150: 160.

Kerfoot, E. M. 1964. Some aspects of mate selection in domestic pigeons. Ph.D. Thesis, Univ. Oklahoma.

Khattab, A. G. H., Watson, J. H. and R. F. E. Axford, 1964. Associa­ tions between serum transferrin polymorphism and disturbed segregation ratios in Welsh mountain sheep. Anim. Prod. 6: 207-213.

Kimura, M. 1953. "Stepping stone" model of population. Ann. Rept. Nat. Inst. Genetics, Japan 3: 62-63. 280

Kimura, M. and G. H. Weiss, 1964. The stepping stone model of popu­ lation structure and the decrease of genetic correlation with distance. Genetics 49: 561-576.

Komatsu, S. K. and R. E. Fenney, 1967. Role of tyrosyl groups in metal binding properties of transferrins. Biochemistry 6: 1136.

Kristjansson, F. K., 1960(a). Inheritance of a serum protein in swine. Science 131: 1681.

,1960(b). Genetic control of two blood serum proteins in ----swine. Can. {. Genet. Cytol. 2: 295-300.

,1963. Genetic control of two pre-albumins in pigs. ----Genetics, 48: 1059-1063.

----, and J. D. Cipera, 1963. The effect of sialidase on pig ~ransferrins. Can. {. Biochem. Physiol. 41: 2523-2527.

Kunkel, H. G. and R. J. Slater, 1952. Lipoprotein patterns of serum obtained by zone electrophoresis. J. Clin. Invest. 31: 677.

~_~~ and A. Tiselius, 1951. Electrophoresis of proteins on filter paper. J. Gen. Physiol. 35: 89.

Langdon, S. H., 1931. The mythology of all races. Vol. V., Semitic, Boston: Marshall Jones Co., Inc.

Law, G. R. J., 1967. Alkaline phosphatase and leucine aminopeptidase association in plasma of the chicken. Science 156: 1106-1107.

---,...... - and S. S. Munro, 1965. Inheritance of two alkaline phos- phatase variants in fowl plasma. Science 149: 1518.

Levene, H., 1949. On a matching problem arising in genetics. Ann. Math. Stat. 20: 91-94.

Levi, W. M., 1969. The Pigeon. South Carolina: Levi Pub. Co.

and W. F. Hollander, 1939. Structural anomalies and color mosaics observed in a colony of domestic pigeons. JOUL of Heredity 30: 453-457.

Lewontin, R. C. and J. L. Hubby, 1966. A molecular approach to the study of genec heterozygosity in natural populations. II. Amount of variation and degree of heterozygosity in natural populations of Drosophila pseudoobscura. Genetics 54: 595-609.

Li. C. C., 1955. Population Genetics. The University of Chicago Press, Chicago and London. 281

Li, C. C., 1969. Population subdivision with respect to multiple alleles. Ann. Hum. Genet. 33: 23-29.

and D. G. Hurvitz, 1953. Some methods of estimating the -----:---:- inbreeding coefficient. ~. Hum. Genet. 5: 107-117.

Lloyd-Jones, 0., 1915. Studies on inheritance in pigeons II. A microscopial and chemical study of the feather pigments. Jour. Exp. Zoology, 18: 453-509.

Loisel, G., 1903. Recherches de statistique sur ladescendance des pigeons voyageurs (note preliminaire.) Assoc. Franciase pour L'Avancement de Science, Compte Rendu de la 32me Sessione, Angers, 1903. Seconde Partie: Notes et Memoires (Paris) pp. 760-764.

,1904. Recherches de statistique sur ladescendance des ----pigeons voyageurs. 6me. Congres International de Zoologie, Compte Rendu des Seances 1904, Berne, pp. 663-672.

,1905. Etudes sur l'heredite de la coloration du plumage ----chez les pigeons voyageurs. Comptes Rendus, Societe de Biologie (Paris) 58: 465-468.

Lorenz, K., 1937. The companion in the bird's world. Auk 54: 245-273.

Lowe, V. P. W. and E. I. McDougall, 1961. Serum S-protein types in red deer and other species and their stability in the presence of bacteria. Nature, London, 192: 983-984.

Lyell, J. C., 1881. Fancy pigeons. Second ed. 1887. Third ed. 1897. London: L. Upcott Gill.

Mainardi, D. 1964. Effeto evolutio della selezione sessuale basata su imprinting in Columba livia. Rivista Italiana di Ornitologica. 34: 213-216.

Malecot, G., 1948. Les Mathematique de l'Heredite. Masson, Paris.

,1950. Quelques Schemas Probabilistes sur La Variabilite -----:o- des Populations Naturelles. Ann. de l'Universite de Lyon, 13. Section A. In Sciences Mathematigues ~ Astronomie (Paris, Masson et C). pp. 37-60.

, 1959. Les modeles stochastiques en genetique de population. ----Publications de l'Institut de Statistique de l'Universite de Paris. 8: 173-210. 282

Manwell, C. and C. M. A. Baker, 1968. Genetic vAriation of isocitrate, malate, 6-phosphogluconate dehydrogenases in snails of the genus Cepea introgressive hybridization, polymorphism and pollution? Compo Biochem. Physiol. 26: 195-210.

______~, and C. M. A. Baker, 1970. Molecular Biology and the Origin of Species. Heterosis, Protein Polymorphism and Animal Breeding. University of Washington Press, Seattle.

Martin, C. M. and J. H. Jandl, 1959. Inhibition of virus multiplica­ tion by transferrin. J. Clin. Invest. 38: 1024.

Maruyama, T., 1970. Analysis of population structure. I. One­ dimensional stepping-stone models of finite length. _A_n_n. _H_un_l. Genet., London, 34: 201-219.

Mc Daniel, L. S. and H. L. Chute, 1961. Enzyme activity levels in chicken plasma. Am.~. Vet. Res. 22: 99.

Mc Donald, M. R., and O. Riddle, 1945. The effect of reproduction and estrogen administration on the partition of calcium, phosphorous, and nitrogen in pigeon plasma. ~. BioI. Chem. 159: 445.

Metzelaar, J., 1924. The inheritance of red in pigeons. Amer. Pigeon Keeper, Vol. 26, no. 3 (Mar.), pp. 15-16.

,1924. The inheritance of red-barred in pigeons. Amer. ------=--:--Pigeon Keeper, Vol. 26, no. 4 (April), pp. 18-19.

______, 1924. The khaki dragoon. Amer. Pigeon Keeper, Vol. 26, no. 9 (Aug.) p. 22.

,1926. Color breeding in pigeon plumage. Geneva, Illinois: ------:--American Pigeon Keeper.

,1928. Further experiments in inheritance of color in ------domestic pigeons. Univ. of Michigan, Museum of Zoology, Occassional papers no. 194.

Michaelis, L., 1909. Elektrische Uberfuhrung von Fermenten. J. Biochem. Z. 16: 81-86.

Miller, W. J., and W. J. Wagen, 1955. Sexing matur~ Columbiforrnes by cloacal structures. The Auk, 72: 279-285.

Millicent, E. and J. M. Thoday, 1960. Gene flow and divergence under disruptive selection. Science 131: 1311-1312.

Milne, H. and F. W. Robertson, 1965. Polymorphisms in egg albumen protein and behaviour in the eider duck. Nature 205: 367-369. 283

Moore, J. A., 1965. The effect of dimorphic parents on mate selection in pigeons. Ph.D. Thesis, Univ. Oklahoma.

Morgan, T. H., 1911. Notes on crosses between different races of pigeons. Biological Bulletin 21: 215-221.

Morton, N. E. and N. Yasuda, 1962. The genetical structure of human populations. In Les Depacements humains, Entretien de Monaco en Sciences Humaines. J. Sutterm Ed., I. Hachette. pp. 186-203.

Morton, M., Miki, C. and S. Yee, 1968. Bioassay of population structure under isolation by distance. Amer.~. Hum. Genet. 20: 411-419.

Mueller, J. 0., Smithies, 0., and M. R. Irwin, 1962. Transferrin variation in Co1umbidae. Genetics, 47: 1385-1392.

Nee1, J. V. and W. J. Schull, 1954. Human Heredity, the University of Chicago Press, Chicago, Illinois.

Nei, M., 1965. Variation and covariation of gene frequencies in sub­ divided populations. Evolution, 19: 256-258.

----:--- and Y. Imaizumi, 1966. Genetic structure of human popu1a- tions, 1. Local differentiation of blood group gene frequencies in Japan. Heredity, Vol. 21, pp. 9-35.

0'Dona1d P., 1959. Possibility of assortive mating in the Arctic Skua. Nature 183: 1210-1211.

,1960(a). Assortative mating in a population in which two -----alleles are segregating. Heredity 15: 389-396.

,1960(b). Inbreeding as a result of imprinting, Heredity --':""':15=-:- 79-85.

Ogden, A. L., Morton, J. R., Gilmour, D. G., and E. M. Mc Dermid, 1962. Inherited variants in the transferrins and cona1bumins of the chicken. Nature, London, 195: 1026-1028.

Ornstein, L., 1964. Disc electrophoresis -- I. Background and theory. Ann. N. Y. Acad. Sci. 123: 321-349.

Parr, C. W. 1966. Erythrocyte phosphog1uconate dehydrogenase poly­ morphism. Nature, London, 210: 487.

Parsons, P. A., 1962. The initial increase of a new gene under positive assortative mating. Heredity, 7: 267-276.

-----,.--,1965. Assortative mating for a metrical characteristic in Drosophila. Heredity, 20: 161-167. 284

Pearson, K. and A. Lee, 1903. On the laws of inheritance in man. I. Inheritance of physical characters. Biometrika, 2: 357-462.

Pou1ik, M. C. 1967. Starch gel electrophoresis in a discontinuous system of buffers. Nature, London, 180: 1477-1479.

Raymond, S. and B. S. Weintraub, 1959. Acry1amide gel as a support­ ing medium for zone electrophoresis. Science 130: 711.

Reeve, E. C. R., 1955. Inbreeding with homozygotes at a disadvantage. Ann. Hum. Genet., London, 21: 277-288.

Rende1, J., and C. Stormont, 1964. Variants of ovine alkaline phosphatases and their association with the R-O blood groups. Proc. Soc. expo Bio1. Med., 115: 853.

Riddle, O. and W. F. Hollander, 1943. Scraggly plumage and ataxia, two inherited characteristics in the pigeon. Jour. of Heredity 34: 167-172.

Robertson, A., 1962. Selection for heterozygotes in small populations. Genetics 47: 1291-1300.

Sa1tman, P. and P. J. Charley, 1960. Th~ regulation of iron metabolism by equilibrium binding and chelation, in Seven M. J. and Johnson, L. A. (eds.) Metal Binding in Medicine, p. 241. Lippincott, Philade1phia.

Schade, A. L. and L. Caroline 1946. An iron-binding component in human blood plasma. Science, N. Y. 104: 340-341.

, Oyama, J., Reinhart, R. W. and J. R. Miller, 1954. Bound ----iron and unsaturated iron-binding capacity of serum; rapid and reliable quantitative determination. Proc. Soc. expo Bio1., 87: 443. ------

, Reinhart, R. and H. Levy, 1949. Carbon dioxide and oxygen ----in complex formation with iron and siderophilin, the iron-building component of human plasma. Arch. Biochem. Biophys. 20: 170. .. Schmidt, D.O., 1962. Uber den erb1ichen Po1ymorphismus des roten B1uffarbstoffes bei Rindern der Hohenrassen. Z. Tier zucht, Zuchtbio1. 79: 286-290.

Scudo, F. M. and S. Karlin, 1969. Assortative mating based on pheno­ type. I. Two alleles with dominance. Genetics 63: 479-499.

Seiger, M. B., 1967. A computer simulation study of the influence of imprinting on population structure. Amer. Natur., 101: 47-57. 285

Seiger, M. B. and R. D. Dixon, 1970. A computer simulation study of the effects of two behavioural traits on the genetic structure of semi-isolated populations. Evolution 24: 90-97.

Selander, R. K., 1970. Behavior and genetic variation in natural populations. Am. Zoologist 10: 53-66.

Shaw, C. R. 1965. Electrophoretic variations in enzymes. Science 149: 936.

Shaw, D. J., 1969. Electrophoresis. Academic Press, New York and London.

Shreffler, D. C., 1960. Genetic control of serum transferrin type in mice. Proc. natn. Acad. Sci. U.S.A., 46: 1378-1384.

Singe, R. L. M. and A. Tiselius, 1950. Fractionation of hydrolysis products by amulose by electrokinetic ultra filtration in an agar-agar jelly. Biochem. J. 46: xli (abs.)

Smithies, 0., 1955. Zone electrophoresis in starch gels: group variations in the serum proteins of normal human adults. Biochem. J. 61: 629-644.

, and O. Hiller, 1959. The genetic control of transferrins -----in humans. Biochem. J. 72: 121-126.

Snedecor, G. W. and W. G. Cochran, 1967. Statistical Methods. The Iowa State University Press, Ames.

Soderberg, E., 1927. Ein beitrag zur genetik der spitzkappe und der rundkappe bei rassetauben. Hereditas 8: 363-366.

Spuhler, J. 1961. Communication to Tenth Pacific Science Congress, Honolulu.

Staples-Browne, R., 1905. Note on heredity in pigeons. Zoological Society of London, proceedings pp. 550-558.

______, 1908. On the inheritance of colour in domestic pigeons, with special reference to reversion. Zoological Society of London, proceedings, pp. 67-104.

,1912. Second report on the inheritance of colours in ----:--pigeons, together with an account of some experiments on the crossing of certain races of doves, with special reference to sex-limited inheritance. Jour. of Genetics 2: 131-162.

Steele, D. G., 1928. Genetic relations of chocolate brown plumage color in the domestic pigeon. Proceedings Society Exper. Biology and Medicine 25: 477-478. 286

Steele, D. G., 1931. Studies on inheritance in pigeons, IX. The chocolate-brown plumage color. Genetics 16: 532-546

Steele, R. G. D. and J. H. Torrie, 1960. Principles and Procedures of Statistics with Special Reference to the Biological Sciences. McGraw-Hill Book Company, Inc., New York, Toronto, London.

Stutts, E. C., Briles, W. E. and H. o. Kunkel, 1957. Plasma alkaline phosphatase activity in mature inbred chickens. Poultry Sci. 36: 269.

Tanabe, Y. and F. H. Wilcos, 1960, Effects of age, sex, and line on serum alkaline phosphatase of the chicken. Proc. Soc. Expt1. Bio1. and Med. 103: 68.

Thoday, J. M. and T. B. Bown, 1959. Effects of disruptive selection II. Polymorphism and divergence without isolation. Heredity, 13: 205-218.

Thuline, H. C., Morrow, A. C., Norby, D. E. and A. G. Motulsky, 1967. Autosomal phosphogluconic dehydrogenase polymorphism in toe cat. Science 157: 431.

Tiselius, A., 1937. A new apparatus for electrophoretic analysis of colloidal mixtures. Trans. Faraday Soc. 33: 524-531.

Tonoue, T. and K. Matusumoto, 1961. Serum alkaline phosphatase response to the injection of thyroxine in young chickens. Poult. Sci. 40: 206-212.

Vecchi, A., 1935. RicheLr.he sull'eredita nei co1ombi domestici. Eredita di alcuni caratteri del colobo pavonce1l. Bol. Zool. (Torino) 6: 227-243.

,1937. SuI colore dell'iride nei colombi domestici e sulla ----sua eriditarieta. Archivio Zoologico Ital. 24: 91-112.

Wahlund, S., 1928. Zusammensetzung von populationen und korrela­ tionser-scheinungen von standpunkt der vererbungslehre aus betrachtet. Hereditas 2: 65-106.

Ward, R. H. and C. F. Sing, 1970. A consideration of the power of the X2 test to detect inbreeding effects in natural popula­ tions. The Am. Naturalist. 104: 355-366.

Warriner, C. C., Lemon, W. B., and L. S. Ray, 1963. Early experience as a variable in mate selection. Anim. Behav. 11: 221-224.

Whitman, C. 0., 1919. Posthumous works. Vol I. Orthogenetic evolution in pigeons, O. Riddle Ed. Carnegie Institution of Washington, Publication 257. 287

Wilcox, F. H., 1966. Genetics, Princeton. 53: 799-805.

Williams, J. 1962. A comparison of conalbumin and transferrin in the domestic fowl. Biochem. J. 83: 355-364.

Windle, J. J., Wiersoma, A. K. and J. R. Clark, 1963. Investiga­ tion of the iron and copper complexes of avian conalbumins and human transferrins by electron paramagnetic resonance. Biochemistry (Wash.), 2: 1341-5.

Woolf, B. 1955. On estimating the relation between blood group and disease. Ann. Hum. Genet. 14: 251-253.

Workman, P. L., 1969. The analysis of simple genetic polymorphisms. Hum. BioI., 41: 97-114.

Wright, S., 1921. System of mating. Genetics 6: 111-178.

,1922. Coefficients of inbreeding and relationship. Am. ----Naturalist, 56: 330-338.

,1931. Evolution in Mendelian pcpu1ations. Genetics 8: --9=7--159.

----,1943. Isolation by distance. Genetics 16: 97-159. ____, 1946. Isolation by distance under diverse system of mating. Genetics 31: 39-59.

,1948. Genetics of populations. Encyclopedia Britannica ----~1~0-: lll-A-D-112.

______, 1951. The genetical structure of population. Ann. Eug., London 15: 323-354.

______, 1965. The interpretation of population structure by F­ statistics with special regard to systems of mating. Evolution 19: 395-420.

------,1966. Polyallelic random drift in relation to evolution. ~.!.~.~., 55: 1075-1081.

______, 1969. Evolution and the Genetics of Populations. 1l The Theory £i Gene Frequencies. The University of Chicago Press, Chicago and London.

Wright, T. D. and A. D. Hasler, 1967. An electrophoretic analysis of the effects of isolation and homing behaviour upon the serum proteins of the white bass (Roccus chrysops) in Wisconsin. Am. Nat., Vol. 101: 401-413. 288

Wright, S. and H. C. McPhee, 1925. An approximate method of calcula­ tion of coefficients of inbreeding and relationship from livestock pedigrees. ~. Agr. Res. 31: 377-383.

Yasuda, N. 1966. The genetical structure of Northeastern Brazil. Ph.D. thesis, University of Hawaii.

----,1968. An extension of Wah1und's principle to evaluate mating type frequency. Amer.~. Hum. Genet. 20: 1-23.

,1968. Estimation of the inbreeding-coefficient from ----phenotype frequencies by a method of maximum likelihood scoring. Biometrics 24: 914-935.

Yoshioka, R., Fujii, T. and Y. Ito, 1966. An electrophoretic resolution and densitometric determination of Apo-transferrin and iron-bound transferrin. Biochem. biophys. Res. commun. 24: 203.