Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ecological genetics of threespine sticklebacks. (Gasterosteus) Hay, Douglas Edward 1974

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1974_A1 H39.pdf [ 8.56MB ]
Metadata
JSON: 831-1.0099985.json
JSON-LD: 831-1.0099985-ld.json
RDF/XML (Pretty): 831-1.0099985-rdf.xml
RDF/JSON: 831-1.0099985-rdf.json
Turtle: 831-1.0099985-turtle.txt
N-Triples: 831-1.0099985-rdf-ntriples.txt
Original Record: 831-1.0099985-source.json
Full Text
831-1.0099985-fulltext.txt
Citation
831-1.0099985.ris

Full Text

ECOLOGICAL GENETICS OF THREESPINE STICKLEBACKS (Gasterosteus) by DOUGLAS EDWARD HAY B.Sc, University of British Columbia, 1966 M.Sc, University of British Columbia, 1969 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF SESILTDSOPHY in the Department of Zoology We accept this thesis as conforming to the required standard The University of British Columbia Apri l , 1974 In presenting t h i s thesis i n p a r t i a l fulfilment of the requirements for an advanced degree at the University of B r i t i s h Columbia, I agree that the Library s h a l l make i t freely available for reference and study. I further agree that permission for extensive copying of t h i s thesis for scholarly purposes may be granted by the Head of my Department or by his representatives. I t i s understood that copying or publication of t h i s thesis for f i n a n c i a l gain s h a l l not be allowed without my written permission. Department The University of B r i t i s h Columbia Vancouver 8, Canada i i ABSTRACT The threespine stickleback (Gasterosteus aculeatus L.) is a highly varied and widely distributed species of small fish. Throughout much of i t s range there are two distinct forms: a marine or anadromous form, trachurus, with 30-35 lateral bony plates (or scales), and a freshwater form, leiurus, with 7, or fewer, plates. In some areas another form, called semiarmatus, with intermediate plate numbers, i s abundant. Previous studies have emphasized the number of lateral plates as a criterion of phenotypic assessment. However, this criterion i s arbitrary and can be misleading. Therefore, in this study, new c r i t e r i a are described that permit distinction of phenotypes by the presence or absence of certain, individual plates. Leiurus and semiarmatus, previously defined by the number of plates, are re-defined: a l l semiarmatus possess a particular plate that i s absent in leiurus. A l l semiarmatus can be un-equivocally distinguished from leiurus, although there i s considerable phenotypic variation within each group. Phenotypic frequencies were assessed by periodic collections from two populations: a leiurus population and a mixed population of leiurus and semiarmatus. The relative frequency of two different leiurus phenotypes changed markedly between generations but were constant within generations (about one year). In the mixed population, the semiarmatus phenotypes increased in frequency, within generations. Field and laborat-ory tests indicate that the changes in frequency are not due to error in i i i the sampling methods, error in assessment of phenotype, or differential dispersal among the phenotypes. Changes in the frequency of semiarmatus can be explained by differential predation by size-selective predators such as the cutthroat trout CSalmo clarki) and the water scorpion (Ranatra fusca). Both of these predators tend to selectively prey on the smallest sticklebacks available. Early in each generation, leiurus phenotypes are smaller than the semiar- matus phenotypes, but with time, the size difference decreases, and the frequency of semiarmatus increases. This suggests that size-selective predators i n i t i a l l y eliminate more leiurus than semiarmatus, and the effect of this i s an increase on the frequency of semiarmatus each generation. Differences i n certain aspects of reproductive biology were examined as causes of the changes between generations. These included: (1) inheritance of plates; (2) v i a b i l i t i e s of crosses; (3) phenotypic composition of breeding fish; (3) fecundity; (4) assortive mating. The phenotypic composition of breeding f i s h , and the inheri-tance of plates are the most important factors that affect changes in phenotypic frequencies between generations. In the pure leiurus populations; the phenotypic composition of breeders was significantly different from the population at large. This suggests that leiurus phenotypes may differ i n their reproductive potential. The inheritance of the plates (or phenotypes) is especially important as a factor that changes the frequency of semiarmatus, Crosses among leitifus never produce any semiarmatus progeny, but crosses among iv semiarmatus, and between semiarmatus and leiurus, produce many leiurus progeny. Consequently, the frequency of semiarmatus i s higher among breeding fish than i t i s among the young-of-the-year i n the next generation. Leiurus x trachurus crosses produce semiarmatus progeny but semiarmatus is often abundant i n populations where trachurus is absent. In these areas semiarmatus must consistently replicate i t s e l f . Therefore, contrary to previous suggestions, i t is not a phenotype that arises only by introgression or hybridization between leiurus and trachurus. In mixed populations where trachurus is absent, semiarmatus must increase in fre-quency within generations or gradually go extinct. Recurrent extinction of semiarmatus seems a reasonable explanation for the origin of leiurus populations. V TABLE OF CONTENTS Page TITLE PAGE. i . ABSTRACT i i TABLE OF CONTENTS v LIST OF FIGURES v i i i LIST OF TABLES i x LIST OF APPENDICES x i INTRODUCTION 1 GENERAL METHODS AND MATERIALS 8 I. Collecting 8 II . Examining lateral plates 8 II I . Laboratory conditions 8 THE POLYMORPHIC EXPRESSION OF LATERAL PLATES... 10 I. Description of the Polymorphism 10 II . Symmetry of plates 15 III. Development of plates 17 IV. Applicability of phenotypic assessment 18 V. Phenotypic composition of other populations 20 THE COLLECTIONS. 23 I. Analysis of age and growth 23 II. Analysis of phenotypic frequencies - Waterfowl Refuge.. 23 A. Changes within generations 23 B. Changes between generations 30 II I . Analysis of phenotypic frequencies - Hybrid Pond 30 A. Changes within generations 30 B. Changes between generations 36 IV. Changes in phenotypic ratios - summary and conclusions. 38 EXPLANATION OF THE CHANGES 41 I. Hypotheses explaining both kinds of change 41 II . Hypotheses explaining changes within generations 42 II I . Hypothesis explaining changes between generations 42 IV. Non-random sampling 44 A. Waterfowl Refuge 44 B. Hybrid Pond 46 V. Developmental changes i n phenotype 46 VI. Dispersal -r 49 v i Page VII. Selective predation and changes within generations... 50 A. Water scorpion predation. 51 1. Predation experiments.. 53 2. Results 53 a. Analysis by phenotypes 53 b. Predation intensity and prey size 55 c. Size-selective predation and prey avai l a b i l i t y . 58 B. Cutthroat trout predation 58 1. Predation experiments 61 2. Results 62 C. Relevance of the predation experiments 62 D. Explanation of the changes within generation -summary and conclusions 66 VIII. Differential reproduction and changes between generations 68 A. The crosses 68 1. Methods and materials 69 a. Waterfowl Refuge crosses 70 b. Hybrid Pond and other crosses. 70 2. Results of crosses 71 a. Frequency of successful crosses 71 b. V i a b i l i t i e s among successful crosses.. 73 c. Phenotypic analysis of the progeny 76 d. Inheritance of lateral plates 78 e. Hypothesis on the inheritance of plate H. 80 f. A quantitative estimate of inheritance 81 B. Fecundity 83 C. Analysis of breeding populations 85 1. Age of breeding fish 88 2. Phenotypic frequencies during breeding seasons... 91 3. Comparison of breeding and non-breeding fish 91 a. Waterfowl Refuge 91 b. Hybrid Pond. 96 4. Relative abundance of breeding fish 96 D. Habitats of breeding fish 97 IX. Explanation of the changes between generations - a synthesis of previous sections.. 100 A. Predicting phenotypic ratios 103 1. Effects of assortive mating ... 107 2. Effects of differences in fecundity 108 3. Effects of differences in v i a b i l i t y 109 B. Effects of differences in reproductive biology -summary and conclusions......... 110 DISCUSSION 112 CONCLUSIONS 128 ACKNOWLEDGMENTS REFERENCES APPENDIX v i i i LIST OF'FIGURES Page 1. The study areas . -. «• 6 2. Plate pattern in Gasterosteus 12 3. Plate patterns of leiurus populations 13 4. Plate patterns of the trachurus and semiarmatus phenotypes.... 14 5. The number of lateral plates of the G and H phenotypes 19 6. Age and growth of six generations in the Hybrid Pond 24 7. Age and growth of five generations in the Waterfowl Refuge.... 25 8. Frequencies of the G and H phenotypes in the Waterfowl Refuge. 26 9. Frequencies of the G, H, and I phenotypes in the Hybrid Pond.. 35 10. Frequencies of phenotypes between generation 40 11. Size selection predation by Ranatra 57 12. Size selective predation and prey availability 60 13. Cutthroat trout predation tests 64 14. Relative size and frequency of semiarmatus in the*]Hybrid Pond. 67 15. Phenotypic composition of the progeny 77 16. Predicted phenotypic ratios 82 17. Fecundities of the phenotypes 86 ix LIST OF TABLES Page I Plate numbers and plate patterns from the Waterfowl Refuge. . .... 16 II Phenotypic composition of other populations 21 III Phenotypic frequencies i n the Waterfowl Refuge. Analysis of changes within generations 27 IV Phenotypic frequencies in the Waterfowl Refuge. Analysis of changes between generations 31 V Phenotypic frequencies in the Hybrid Pond. Analysis of changes within generations 32 VI Phenotypic frequencies in the Hybrid Pond. Analysis of changes between generations 37 VII".: Sampling tests in the Waterfowl Refuge.... 45 VIII Sampling tests in the Hybrid Pond 47 IX Phenotypic analysis of Ranatra predation experiments 54 X Size analysis of Ranatra predation experiments... 56 XI Size selective predation and prey availability 59 XII Results of cutthroat trout predation experiments 63 XIII Relative sizes of leiurus and semiarmatus phenotypes in the Hybrid Pond 65 XIV Frequency of successful crosses 72 XV V i a b i l i t i e s of successful crosses 74 XVI Phenotypic composition of the progeny 84 XVII Analysis of fecundities 87 XVIII Age distribution during the breeding season in the Waterfowl Refuge , 89 XIX Age distribution during the breeding season in the Hybrid Pond 90 X Page XXT Analysis of breeding populations in the Waterfowl Refuge 92 XXI Analysis of breeding populations in the Hybrid Pond 93 XXII Comparison of phenotypic frequencies between breeding and non-breeding fish in the Waterfowl Refuge 94 XXIII Comparision of phenotypic frequencies between breeding and non-breeding fi s h in the Hybrid Pond 95 XXIV Habitat analysis of leiurus phenotypes 99 x i APPENDIX Page I Analysis of the collections.. 132 II Results of the crosses 134 1 INTRODUCTION The simultaneous investigation of the ecology and genetics of an organism is an effective way of studying i t s evolution. This inter-disciplinary approach, called ecological genetics, i s seen in the classic works of Cain and Sheppard (1954) on land snails (Cepaea), Kettlewell (1956) on moths (Biston), and Ford (1971) on butterflies (Maniola). In the present study a combination of ecological and genetic methods is used to study natural selection and evolution in threespine sticklebacks (Gasterosteus aculeatus L.). In many ways Gasterosteus i s an ideal organism for such invest-igations. It exhibits considerable phenotypic variation that i s largely genetic in origin (Munzing, 1959; Lindsey, 1962; Hagen, 1967). It usually resides in shallow water where i t is easily captured and observed. It is easy to breed and maintain under laboratory conditions. Gasterosteus i s widely distributed along coastal areas in temperate and sub-arctic regions of the northern hemisphere. It lives both in marine and freshwater habitats, but i t s penetration into inland waters is limited and rarely exceeds 200 km. This distribution suggests that freshwater habitats were originally populated by migrants from the sea. Most freshwater populations are morphologically distinct from the marine, or anadromous populations. There are a number of important differences between the two groups, but the most striking difference i s in the number of lateral bony plates (modified scales). 2 At present, marine and freshwater forms are recognized as different subspecies: G.a. aculeatus L. (the marine form), G.a. micro- cephalus Girard (the Pacific Coast freshwater form), and G.a. leiurus L. or G.a. gymnurus L. (the European freshwater form). The freshwater forms of Europe and North America are morphologically similar but they are provisionally regarded as different subspecies (Miller and Hubbs, 1969). The validity of this subspecific status i s , however, in doubt. Hagen (1967) found evidence of reproductive isolation between marine and freshwater forms in a small coastal stream in British Columbia and he advocates that two species be recognized: G. trachurus, the marine fi s h , and G^ aculeatus, the freshwater fish. (Most investigators in Europe arid North America simply refer to the marine form as trachurus and the freshwater form as leiurus; these terms are used only for con-venience and have nothing to do with zoological nomencalture.) Hagen's evidence for reproductive isolation is convincing. The marine and freshwater forms have different habitats and they are found in different parts of the river: low-plated fish (leiurus) l i v e in up-stream areas and high-plated fish (trachurus) li v e in downstream areas. Transfer experiments indicate that neither form can survive in the hab-i t a t of the other. Hagen (1967) points out that the two forms differ in many mor-phological characteristics but he considers the number of lateral plates to be diagnostic of each species. He defines the marine fi s h (trachurus) as having more than 30 plates and the freshwater fish (leiurus) as hav-ing 7, or fewer plates. Defined in this way, the distributions of 3 leiurus and trachurus are mutually exclusive except for a short, inter-mediate section of the river where they overlap. In this area, one also finds fish having between 7 and 30 plates. These fish are often called semiarmatus, but the term has no taxonomic significance. Munzing (1959) and Hagen (1967) demonstrated that the semiar-matus phenotype can be produced from leiurus x trachurus crosses in the laboratory. Hagen (1967) regards semiarmatus as a hybrid that i s poorly adapted to either leiurus or trachurus habitats, although he found no direct evidence of semiarmatus i n v i a b i l i t y . However, semiarmatus i s sometimes abundant and widespread i n other populations or river systems. Miller and Hubbs (1969) regard this as evidence of interbreeding and introgression between the marine and freshwater forms, and for this reason they recommend retention of the subspecific status. Although both taxonomic interpretations disagree about the significance of lateral plate number, they agree that plate number is an accurate and meaningful assessment of phenotype. I think that count-ing plates can often be a useful assessment of phenotype, but i t can also be misleading. Plates vary considerably i n size, shape, and position relative to other structures. The most anterior plates are often very small and d i f f i c u l t to see. Some larger, more posterior plates are consistently associated with the pelvic girdle or the bony supports for the dorsal spines, and in trachurus and some semiarmatus the plates on the caudal peduncle are modified to form a keel. There-fore, in the present study, I present alternate c r i t e r i a for phenotypic assessment that is based on the presence or absence of certain specific plates. These plates are recognized by their size and position relative to other structures. This method of phenotypic assessment allows an un-equivocal distinction between semiarmatus and "leiurus: a l l semiarmatus possess a particular plate that is absent in leiurus. There i s a series of leiurus phenotypes and a series of semiarmatus phenotypes. The study began when I observed that the frequencies of certain phenotypes changed with time in two different populations: a leiurus population, and a mixed population of leiurus and semiarmatus. From these observations, I outlined a study with two basic objectives: (1) to document the changes in phenotypic frequency; and (2) to explain, by experiment and observation, the reasons for the changes. Dobzhansky (1951) defines evolution as a change in the genetic composition of populations. Therefore, understanding the reasons for change is fundamental to understanding the evolution of an organism. I feel that this study, which attempts to explain the reasons for pheno-typic change, i s relevant to understanding the significance of the leiurus and semiarmatus phenotypes in the evolution of Gasterosteus. My use of the term phenotype i s controversial. The reader i s encouraged to substitute the term morph for phenotype at his pleasure. 5 THE STUDY AREAS The investigation was carried out on the L i t t l e Campbell River, a small river in the extreme southwest corner of British Columbia (Fig.l). The river is about 25 km long and i s characterized by a variety of habitats (see Hagen, 1967, for a detailed description). Two areas, about 14 km apart, were selected for detailed study. One area, the Hybrid Pond, about 5 km from the mouth of the river, (pond M of Hagen, 1967) i s isolated from the river for most of the year. On occasions of extremely heavy rain and high water in the winter and early spring, a small channel connects the pond to the river. The pond i s about 0.4 hectares in area, and has extensive vegetation in the shallow water around the perimeter. The center of the pond i s deeper (2 to 3m). The pond supports an abundant and varied insect and amphibian fauna as well as sticklebacks. Coho salmon smolts (Oncbrhyhchus kisutch) and cutthroat trout (Salmb clarki) are occasionally captured in the pond. The second study area, the Waterfowl Refuge, is a bird sanc-tuary about 19 km from the mouth of the river (14 km above the Hybrid Pond). This area i s i n the headwaters of the river (Hagen's station E) and i s characterized by slow turbid water, heavy vegetation, and shallow depths (maximum about 2 m). The width of the river at this point rarely exceeds 6 m. In addition to an aquatic insect and amphibian fauna, several other fish species occur i n the area. These include: coho salmon, cutthroat trout, prickly sculpins (Cbttus asper), and finescale suckers Fig. 1. The study areas. 7 (Catostomus catbstomus). Two additional areas were used for other purposes. An aband-oned gravel pit called Deck's Pond, located between the two main study areas, is several hectares in area, f i l l e d with clear water, and supports an abundant stickleback population. The other area consists of irrigation and drainage ditches in Ladner, B.C. These ditches also support dense populations of sticklebacks. Some preserved stickleback collections were also used in this study. These collections were made from lakes on Vancouver Island and streams i n the v i c i n i t y of Vancouver. 8 GENERAL METHODS AND MATERIALS I. Collecting Collections of sticklebacks were usually made with a large aluminum dip net (0.3 cm mesh). The net opening was at right angles to a main shaft about 2.5 m long. The net was thrown out from the shore, sunk in the water (perhaps two to three m away), and pulled towards the shore along the bottom. One haul could cover three m^  of the bottom and strain about one m^  of water. The collections were either preserved in the f i e l d (in 10% formalin) or transported alive to the laboratory. II. Examining lateral plates Phenotypes were determined by the position and structure of certain individual plates. A description of the plates, and the criterion for phenotypic recognition follows later. Staining with aliz a r i n red dye (Aliztgirin red S, C.L. 58005, J.T. Baker Chemical Company, Phillipsburg, N.J., U.S.A.), which has an a f f i n i t y for calcified structures, helped me identify the plates. The fish were fixed in 10% formalin, l i g h t l y rinsed in water, and then immersed in the staining solution ( a 2 N solution of sodium hydroxide containing alizer i n dye). The concentration of alizer i n dye in the staining solution was varied according to the number of f i s h to be stained. III. Laboratory conditions Maintenance procedures varied (see later sections). Many different aquaria were available, ranging from 9 to 4000 l i t e r s . 9 Temperatures in the laboratory varied between 16 and 21''C., but were generally around 19 C. Photoperiod for most aquaria was fixed at 16 hours light: 8 hours dark. Fish were fed Tubifex worms and frozen brine shrimp (Artemia). Small fish were fed on liv e brine shrimp nauplii. 10 THE POLYMORPHIC EXPRESSION OF LATERAL PLATES I. Description of the polymorphism The polymorphism that is the basis of this study involves certain lateral bony plates that are present in some individuals and absent i n others. Although there i s a considerable literature*- dealing with plates as taxonomically significant meristic characters, they have never been regarded as characters exhibiting a discrete, discontinuous polymorphism (i.e., l i k e bands on a snail or spots on a butterfly). For this reason, a detailed description of the polymorphism follows. Figure 2 is a diagram of a stickleback with nine lateral plates. Each plate is assigned a letter: A i s the smallest, most an-terior plate, and I i s the most posterior plate. The key to the recog-nition of the polymorphism, or pattern, i s the consistency of occurrence and position of the central plates. Plate D i s anterior to the base of the f i r s t dorsal spine and the anterior end of the pelvic girdle; plates F and G adjoin the second dorsal spine base and the pelvic girdle. Plates E, F, and G are always among the largest of the plates, and are readily distinguished by their relationship to the dorsal spines and pelvic girdle. Occasional individuals with large D and H plates are also found, so that size alone i s not an absolute criterion of plate pattern. Plates A, B, and C are not associated with other ossified structures. Plate C i s more common than B, and plate A is rare i n pure leiurus populations. Gaps in the series are uncommon. For instance, 11 when plates C and E are present, plate D Is invariably present. Figure 3 shows the phenotypic patterns found in leiurus populations. Most 3- and 4-plated fish (Figs. 3a-b) possess plates EFG and DEFG respectively. The two common 5-plated f i s h (Figs. 3c-d) are either CDEFG or DEFGH. Similarily, 6-plated fish (Figs. 3e-f) can be either BCDEFG or CDEFGH. Seven-plated fish (Fig. 3g) are usually BCDEFGH. In Fig. 4 the stipled structures represent plates found in individuals with intermediate to high plate numbers (8 to 35). Plates A and B, omitted for simplicity, may or may not be present in these forms. Figure 4a shows a fully-plated individual with 31 plates. The most posterior plates from a keel on the caudal peduncle, typical of the anadromous trachurus form. The phenotypes in Figs. 4b-e are typical of fish with intermediate plate numbers. Hagen (1967) refers to these types as hybrids; Heuts (1947) and Munzing (1959, 1963) c a l l them semiarmatus, but give the name with no taxonomic significance. The range of variation in this form (semiarmatus) i s extensive. Frequently, individuals are fully-plated except for a gap immediately anterior to the caudal keel (Fig. 4b), or the most posterior plates (including the caudal keel) may be absent (Fig. 4c). Fish with lower plate numbers are shown in Figs. 4d-e. A keel, or partial keel, may be present. Figure 4f shows an individual with seven plates. The most posterior plate (plate I) distinguishes this phenotype from a 7-plated leiurus, where the most posterior plate i s plate H. It i s proposed First dorsal spine and base Pectoral fin (removed) Second dorsal spine and base * Nine lateral plates (Atol) Pelvic spine t o Fig. 2. Plate pattern in Gasterost eus. 13 (a) R a t e s E . F . G (b) P l a t e s D . E . F . G (c) P l a t e s C . D . E . F . G (d) P l a t e s D , E , F , G , H (e) P l a t e s C . D . E . F . G . H (f) P l a t e s B . C . D . E . F . G (g) P l a t e s B , C , D , E , F , G , H Fig. 3. Plate patterns of leiurus populations. 14 (a) 25 plates posterior to plate H The most posterior plates form a keel on the caudal peduncle (b) 17 plotes posterior to plote H. The last 4 plates form a keel. (c) 15 plates posterior to plate H. No keel. (d) 7 plates posterior to plate H. The last 4 plates form a keel. (e) 6 plates posterior to plate H. No keel. (f) I (one) plote posterior to plate H. Fig. 4. Plate patterns of trachurus (a) and the semiarmatus phenotypes (b to f ) . 15 here that this phenotype i s a semiarmatus phenotype, and that plate I is diagnostic of semiarmatus. Such individuals are not found in "pure" leiurus populations such as the upstream Waterfowl Refuge, but are common in areas where mixing or hybridization occurs between leiurus and trachurus. On this point I disagree with Hagen's (1967) arbitrary desig-nation of a l l specimens with 7 or fewer plates as the leiurus phenotype. Instead, I define as leiurus a l l fish without plates posterior to H. The genetic basis for this definition, and i t s applicability and generality, are discussed later. I I . Symmetry of plates Plate numbers and plate patterns are often different between opposite sides of the body. The assessment of this asymmetry depends on the kind of phenotypic analysis used. For instance, Table I shows the phenotypic assessment of a leiurus population in which'.the plate pattern and plate number of the l e f t side of the fish are arrayed against the plate pattern and plate number of the right side. No individual differs by more than two in the number of plates on opposite sides of the body. However, differences in plate patterns do not differ by more than one in the anterior direction and one in the posterior direction. For instance, i f the most anterior plate on one side i s plate C, then the most anterior plate on the other side w i l l be B, C, or D. If D is the most anterior plate on one side, then the most anterior plate on the other side w i l l be C, D, or E. Similarly, i f G is the most posterior plate on one side then the most posterior plate on the other side w i l l be F, G, or H. 16 Table I. Plate numbers and plate patterns from the Waterfowl Refuge. The numbers and pattern of plates on the right side of the body are arrayed against the number and pattern on the l e f t side of the body. The plate number does not differ by more than two between each side. The number in each square is the number of individuals of each phenotypic combination. PLATES ON RIGHT Posterior Plate = G SIDE OF BODY Posterior Plate = H o : w (5: (6: (5) (6) (7) TOTAL (3) (4) C5) C6) EFG DEFG CDEFG BCDEFG C5) C6) (7) DEFGH CDEFGH BCDEFGH TOTAL E F a 2 ; 3 1 6 G 5 151: 27 14 3 200 C D E F G 30 119 10 8 21 1 189 B C D E F G 11 ' 5 3 3 22 D E F G H 20 4 42 10 76 C D E F G H 10 19 4 18 65 5 121 B C D E F G H 3 7 9 19 7 214 180 22 86 109 18 633 17 Occasional f i s h do not conform to this pattern. Exceptions, however, are usually due to various anatomical deformities or severe parasitic infections, that obliterate plate pattern. Asymmetry i s more pronounced in the posterior plates of semiarmatus individuals (plates between I and the caudal keel plates). Here differences of more than two between sides are common. III. Development of plates Plates develop sequentially. The f i r s t plates to appear in young fish are the three largest: E, F, and G. These are closely followed by plate H, (posteriorly, i f present) and plate D (anteriorly). Generally, the most anterior plates (A, B, and C) are expressed later than plate H. Plate C is expressed before B, and plate B before A. Expression of the other plates (posterior to plate H) in semiarmatus,' and trachurus generally proceeds from the most anterior towards the most posterior. Plate I, i f i t i s expressed at a l l , i s present on even very small f i s h . Because some plates are formed later than others, one cannot determine the phenotype of very young fi s h . I therefore decided to use the presence or absence of plates H and I to distinguish the following three phenotypes: 1. Phenotype G: In this phenotype plates H and I are absent from both sides of the body. I c a l l this phenotype G because plate G i s the most posterior plate. Infrequently plate G i s missing from one or both sides 18 so that F i s the posterior plate. Such fish are s t i l l scored as pheno-type G. 2. Phenotype H: In this phenotype plate H i s present on one or both sides of the body, but plate I i s absent. (Plate G i s invariably present.) 3. Phenotype I: In this phenotype plate I i s present on one or both sides of the body. Plates posterior to plate I may also be present. (Plates G and H are invariably present.) This phenotype includes a l l the semiarmatus or hybrid categories referred to in other studies (Hagen, 1967, Hubbs and Miller, 1969). IV. Applicability of phenotypic assessment Analysis for phenotypes G, H, and I, instead of the number of plates, reduces or eliminates error in assessment due to ontogenetic expression of plates. The plates used to assess phenotypes are among the earliest to develop. Individuals with plate I are thought to be genetically distinct from those without plate I (even though such individuals may have the same plate number). The presence of plate H i s characteristic of leiurus individuals with higher plate numbers than those lacking plate H. The relationship between plate number and plate pattern is shown in Fig. 5. Few individuals with plate H have a total plate number (both sides of body) less than 8. Similarly, few individuals without plate H (i.e,, phenotype G) have plate numbers greater than 10. 19 0 . 5 OA 0 . 3 0 . 2 V 0 .1 1 Z 3 a 0 . 4 . 0 . 3 -0 . 2 0 . 1 PHENOTYPE G (n=363) 6 7 8 9 1 0 1 1 1 2 PHENOTYPE H (n= 270) 8 9 1 0 1 1 1 2 1 3 1 4 T O T A L PLATE NUMBER F i g . 5. The number o f l a t e r a l p l a t e s of the G and H phenotypes. P l a t e s on both s ides of the body a re i n c l u d e d . The mean p l a t e number of phenotype G i s 9.01 (- 0.08) and the mean p l a t e number of phenotype H i s 10.98 (± 1 .11) . The d i f f e r e n c e between the means i s h i g h l y s i g n i f i c a n t ( t = 20 .14 , 631 d f , p < 0 . 0 1 ) . 20 The difference in mean plate number between G and H phenotypes is highly significant (p<0.01). This observation is particularly pertinent to a comparison of results of this study with other recent contributions (Moodie, 1972, Hagen and Gilbertson, 1972), that consider leiurus with seven plates to be qualitatively different, in terms of ecology and behaviour, from individuals with less than seven plates. By the phenotypic criterion proposed here, almost a l l 7-plated fish, in leiurus populations are H phenotypes. V. Phenotypic composition of other populations I examined stickleback collections, made from various l o c a l i t i e s outside of my study areas, to determine i f the polymorphism for plate pattern is seen in other populations. The analysis is shown in Table I I . Of 17 lakes examined from Vancouver Island, 10 have populations that I define as pure leiurus. (Chemainus Lake i s unique because plate H appears to be lacking entirely.) The remaining seven lakes show varying proportions of the I phenotype; four of them maintain the I phenotype in low frequencies (about 2%) and have I as the most posterior plate. Analysis of collections from rivers and creeks, in the v i c i n i t y of Vancouver, shows similar trends. Plate H is the most posterior plate in two collections; plate I in another. Two other collections contain high-plated semiarmatus (!+)• In addition to this survey, I have superficially examined 21 Table I I . Phenotypic composition of other populations. Collections from % % % %** Lakes bh Vancouver Island, B.C. G H I* 1+ number Blackjack Lake 7 91 2 100 Brannan Lake 57 43 100 leiurus only Brewster Lake 27 67 5 1 100 Campbell Lake 21 79 100 leiurus only Chemainus Lake 100 100 leiurus only Cowichan Lake 1 88 8 3 100 Crystal Lake 41 59 100 leiurus only Diver Lake 56 44 100 leiurus only Dougan's Lake 91 9 100 leiurus only Harwood Lake 88 12 100 leiurus only Hoiden Lake 59 41 100 leiurus only McCoy Lake 32 68 100 leiurus only Mesachie Lake 4 80 6 10 100 Mo him Lake 14 84 1 100 Nanaimo Lakes 26 72 2 100 Patterson Lake 17 82 1 100 Northy Lake 68 32 50 leiurus only Collections from [vers arid Creeks Fishtrap Creek (B.C.) 65 35 40 leiurus only Skagit River Slough (Wash., U.S.A.) 57 41 2 49 Bonsai Creek (Vancouver Island, B.C.) 20 68 12 50 Sumas River (Fraser System) 27 73 40 leiurus only Brunette Creek (Fraser System) 24 36 40 50 * I represents f i s h whose most posterior plate is plate I. **I+ represents, fish which have plate I plus additional posterior plates - but not f u l l y plated. 22 Gasterosteus collections from the U.K., Nova Scotia (including G. wheat-landi), and Alaska. A l l show similar plate patterns to those in the British Columbia populations. From this I conclude that the plate poly-morphism, with respect to phenotypes G, H and I, is universal among low-plated Gasterosteus populations. 23 THE COLLECTIONS I. Analysis of age and growth Following the precedent of Greenbank and Nelson (1959), and van Mullem and van der Vlugt (1964), I determined age by length fre-quency analysis. The total length of individuals was measured to the nearest millimeter. Figs. 6 and 7 (and Appendix Table I ) show the length frequencies for collections taken from the Waterfowl Refuge and Hybrid Pond study areas. Most fish appear to li v e only one year, but some survive into their second year, and very occasional individuals l i v e three years. New recruits (fry, or young-of-the-year) f i r s t appear in the summer months immediately following the spring breeding season. In both study areas fish grow rapidly from March to August and slowly during the rest of the year. II . Analysis of phenotypic frequencies - Waterfowl Refuge The frequencies of phenotypes were compared both within and between generations by chi-square analysis (Steele and Torrie, 1960). A. Changes within generations Tables III a-eyy and Fig. 8 show comparisions of phenotypic frequencies among collections made within generations. Changes in pheno-typic frequencies within generations are not significant (p > 0.05), and differences between collections may be due to normal sampling error. 24 Fig. 6. Age and growth of five generations (some incomplete) in the Water-fowl Refuge. Each collection, or age class within a collection, i s represented by a vertical line. The short horizontal line shows the mean size of the age class; the thin vertical line shows the standard deviation (on either side of the mean); the shorter, thicker, vertical line shows the standard error (on either side of the mean). The mean sizes of very small samples are shown by crosses. 24a Total length in mm. T 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1—1 1 1 1—I 1 1 -, I I I I I — I I T of O z" CO > >o o —4 «. o z o J > + ^ N O o z o o zf >l -> -?l - • — < J rO J 25 Fig. 7. Age and growth of six generations (some incomplete) in the Hybrid Pond. Each collection, or age class within a collection, i s represented by a vertical line. The short horizontal line shows the mean size of the age class; the thin vertical line shows the standard deviation (on either side of the mean); the shorter, thicker, vertical line shows the standard error (on either side of the mean). The mean sizes of very small samples are shown by crosses. 25a Total length in m m . > • o w * » a o M « . ( • s & t & f c s s s s ' i i s -i 1 1 1 1 1 1 1 — T — i 1 1 1 r-00 o • o 26 Fig. 8. Frequencies of the G and H phenotypes in the Waterfowl Refuge. Each line represents a distinct generation. Some overlapping points (older fish) are not included. 26a F r e q u e n c y Q * O 0 0 + o "0 3" 3" f\ 3 3 S5 s x I O o ^ - • M - 0''° ° ro * o--"tow.'' \ ^ I * 6 + d o +.. .0 / » — o''D "° +• ! + .+ o o V I rO / o Table I I I . Phenotypic frequencies in the Waterfowl Refuge., i - -Analysis of changes within generations. Table III a. 1968-1969 Generation  G H  Date No. Freq. No. Freq. TOTAL March 15, 1969 45 0.480 49 0.520 92 August 20, 1969 34 0.531 30 0.469 64 TOTAL 79 0.500 79 0.500 158 heterogeneity x2 = 0.001 with 1 df(p > 0.90) Table III b. 1969-1970 Generation  G H Date No. Freq. No. Freq. TOTAL August 20, 1969 62 0.466 71 0.534 133 June 1, 1970 47 0.603 31 0.397 78 June 8, 1970 19 0.633 11 0.366 30 TOTAL 128 0.531 113 0.469 241 heterogeneity x2 = 5.11 with 2 df (0.10 >p>0.05) 28 Table III c. 1970-1971 Generation Date G No. Freq. No. H Freq. TOTAL July 16, 1970 194 0.576 143 0.424 337 October 9, 1970 181 0.622 110 0.378 291 October 24, 1970 95 0.586 67 0.414 162 November 11, 1970 96 0.658 50 0.342 146 January 29, 1971 94 0.577 69 0.423 163 February 5, 1971 17 0.548 14 0.452 31 March 13, 1971 19 0.576 14 0.424 33 April 14, 1971 43 0.531 38 0.469 81 April 26, 19 71 98 0.587 69 0.413 167 May 4, 1971 132 0.592 91 0.408 223 May 17, 1971 129 0.586 91 0.414 220 June 2, 1971 66 0.595 45 0.405 111 June 9, 1971 83 0.659 43 0.341 126 July 30,1971 37 0.474 41 0.526 78 September 21, 1971 37 0.481 40 0.519 77 TOTAL 1321 0.588 925 0.412 2246 heterogeneity x2 = 16.38 with 14 df (0.5 >p> 0.25) 29 Table III d. 1971-1972 Generation Date No. Freq. No; Freq H TOTAL July 30, 1971 September 21, 1971 October 7, 1971 November 4, 1971 December 1, 1971 February 24, 1972 March 30, 1972 April 27, 1972 May 16, 1972 May 24, 1972 August 29, 1972 93 0.721 36 0.279 129 214 0.679 101 0.321 315 117 0.765 36 0.235 153 72 0.735 26 0.265 98 239 0.740 84 0.260 323 78 0.645 43 0.355 121 65 0.684 30 0.316 95 86 0.705 36 0.295 122 92 0.667 43 0.333 138 60 0.723 23 0.277 83 14 0.667 7 0.333 21 TOTAL 1130 0.707 468 0.783 1598 heterogeneity x2 = 9.66 with 10 df (0.75 •> p > 0.50) Table III e. 1971-(1973) Generation  G H  Date No. Freq. No. Freq. TOTAL May 24, 1972 69 0.527 62 0.473 131 July 13, 1972 146 0.582 105 0.418 251 August 29, 1972 184 0.613 116 0.387 300 TOTAL 399 0.585 283 0.485 682 heterogeneity x2 = 2.84 with 2 df (0.25>p>0.10) 30 B. Changes between generations The data for each generation (from 1969 to 1972) were pooled and frequencies compared between generations (Table IV). This procedure is appropriate when the data within generations are homogeneous. The phenotypic frequencies are significantly different (p < 0.01) which suggests that sampling error cannot account for the observed differences. The frequency of the G phenotype increased from a mean of 0.50 in 1969 to 0.71 in 1971, and then decreased to 0.58 in 1972. (This lower frequency for the 1972-73 generation was confirmed by a collection of 172 adult fish, made in July, 1973; the frequency of phenotype G was 0.56. This collection was not included in the previous analyses.) II I . Analysis of phenotypic frequency - Hybrid Pond A. Changes within generations Significant changes in phenotypic frequency occur i n the 1969, 1970, and 1971 generations (Tables Va-e and Fig. 9). A trend occurs in these years: phenotypes G and H tend to decline with time but phenotype I increases. The data for each generation are also analysed by a chi-square test that disregards phenotype I and considers only the proportions of phenotype G to phenotype H. This analysis i s similar to that performed on the Waterfowl Refuge data. The changes in the proportions of the G and H phenotypes, within generations, i s insignificant for the 1968, 1970, and 1971 generations. The probability of significant change in the 1969 generation i s approximately 0.05. Therefore, the relative 31 Table IV. Phenotypic frequencies in the Waterfowl Refuge. Analysis of changes between generations. G H TOTAL Generation No. Freq. No. Freq. (1968)-19 69 79 0.500 79 0.500 158 1969-1970 128 0.531 113 0.469 241 1970-1971 1321 0.588 925 0.412 2246 1971-1972 1130 0.707 468 0.293 1598 1972-(1973) 399 0.585 283 0.415 682 TOTAL 3057 1868 4925 heterogeneity x2 = 82.49 with 4 df (p « 0.01) 32 Table V. Phenotypic frequencies in the Hybrid Pond. Analysis,of changes within generations. Table V a. 1968-1969 Generation G H I Date No. freq. No. freq. No. freq. TOTAL July 15, 1968 02 0.218 31 0.564 12 0.218 55 Sept. 17, 1968 42 0.182 149 0.645 40 0.173 231 March 17, 1969 36 0.221 95 0.583 32 0.196 163 May 7, 1969 57 0.199 180 0.629 49 0.171 286 Aug. 20, 1969 4 0.235 9 0.529 4 0.235 17 TOTAL 151 0.201 464 0.617 137 0.182 752 heterogeneity X 2 = 2 .60 with 8 df (p > 0.95) heterogeneity X 2 (on G:H ratios) = 0.28 with 4 df (P > 0.99) Table V b. 1969-1970 Generation G H I Date No. freq. No. freq. No. freq. TOTAL Aug. 20, 1969 113 0.355 170 0.534 35 0.110 318 Jan. 29, 1970 35 0.227 87 0.565 32 0.208 154 May 12, 1970 42 0.237 103 0.582 32 0.181 177 June 10, 1970 28 0.237 63 0.534 27 0.229 118 Aug. 5, 1970 2 0.111 13 0.722 3 0.167 18 Oct. 1, 1970 4 0.250 11 0.687 1 0.063 16 TOTAL 224 0.280 447 0.558 130 0.162 801 heterogeneity x2 = 25.66 with 10 df (p < 0.01) heterogeneity x2 (on G:H ratios) = 11.23 with 5 df (p< 0.05) 33 Table V c. 1970-1971 Generation G H I TOTAL Date No. freq. No. • freq. No. freq. Aug. 5, 1970 41 0.224 122 0.667 20 0.109 183 Sept. 8, 1970 51 0.225 135 0.595 41 0.181 227 Oct. 1, 1970 147 0.254 355 0.614 76 0.131 578 Nov. 6, 1970 105 0.216 343 0.707 37 0.Q76 485 Feb. 2, 1971 59 0.190 204 0.656 48 0.154 311 Mar. 17, 1971 51 0.222 159 0.691 20 0.087 230 April 14, 1971 59 0.201 189 0.643 46 0.156 294 April 26, 1971 76 0.183 257 0.618 83 0.199 416 May 18, 1971 47 0.138 207 0.607 87 0.255 341 June 9, 1971 43 0.184 132 0.564 59 0.252 234 June 22, 1971 113 0.170 333 0.502 217 0.327 663 July 30, 1971 9 0.085 63 0.594 34 0.321 106 Sept. 23, 1971 14 0.136 58 0.563 31 0.301 103 Oct. 7, 1971 7 0.127 35 0.636 13 0.236 55 TOTAL 822 0.195 2592 0.613 812 0.192 4226 heterogeneity x2 = 218.53 with 26 df (p < < 0.01) heterogeneity x2 (on G:H ratios) = 21.58 with 13 df (0.10 > p > 0.05) 34 Table V d. 1971-1972 Generation G H I Date No. freq. No. freq. No. freq. TOTAL July 30, 1971 57 0.302 108 0.571 24 0.127 189 Sept. 23, 1971 120 0.294 236 0.578 52 0.127 408 Oct. 7, 1971 60 0.300 106 0.530 34 0.170 200 Nov. 4, 1971 111 0.273 208 0.512 87 0.214 406 Feb. 22, 1972 82 0.265 161 0.521 66 0.214 309 April 15, 1972 105 0.245 240 0.559 84 0.196 429 May 16, 1972 51 0.338 78 0.517 22 0.146 151 June 22, 1972 24 0.222 54 0.500 30 0.278 108 July 13, 1972 35 0.173 94 0.465 73 0.361 202 TOTAL 645 0.269 1285 0.535 472 0.196 2402 heterogeneity x2 = 68.848. 18 df (p < < 0.01) heterogeneity x2 (°n G:H ratios) = 7.033. 8 df (0.75 > p > 0.5) 35 Fig. 9. Frequencies of the G, H, and I phenotypes i n the Hybrid Pond. Each set of lines represents a different generation. Some overlapping points (older fish) are not included. 3 5a 36 proportions of G to H are constant within generations and do not exceed those expected through normal sampling error. This suggests that changes in the proportion of phenotype I (the hybrid or semiarmatus phenotype) are responsible for much of the observed heterogeneity within generations. B. Changes between generations The data within each generation were pooled and are shown in Table VI. Four different analyses were conducted: 1. A test for heterogeneity between generations when a l l three pheno-types are included. The chi-square i s highly significant (p < 0.01) but the interpretation of this is d i f f i c u l t because heterogeneous data (from within generations) were pooled. 2. A test for heterogeneity between generations when phenotype I is excluded. This compares the ratio of phenotype G to phenotype H between generations. The chi-square i s highly significant (p < 0.01), but again the interpretation i s not clear because some heterogeneous data were pooled (the G:H ratio changed within the 1969-70 generation). 3. A test for heterogeneity between generations, excluding phenotype I, and excluding the 1969-70 data. This compares the ratios of pheno-type G to phenotype H, but only homogeneous data were pooled. The chi-square i s highly significant (p < 0.01). This indicates real differences in the proportions of phenotype G to phenotype H between generations. 4. The proportions of phenotype I (the semiarmatus phenotype) was compared with the combined proportions of phenotypes G and H (the leiurus phenotypes). The chi-square i s not significant (p ? 0.10). This suggests that the relative proportions of the leiurus and semiar-37 Table V I . Phenotypic frequencies i n the Hybrid Pond. Analysis of changes between generations. Generation G H I freq. TOTAL No. freq. No. freq. No. 1968-1969 151 0.201 464 0.617 137 0.182 752 1969-1970 224 0.280 447 0.558 130 0.162 801 1970-1971 822 0.195 2592 0.6133 812 0.192 4226 1971-1972 645 0.268 1285 0.535 472 0.196 2402 TOTAL 1842 4788 1551 8181 heterogeneity x2 = 72.740 6 df (p < < 0.01) heterogeneity x2 (1969 excluded* ) = 56. 749 4 df (p -c< 0.01) *1969 heterogeneous within generations. heterogeneity x2 (ratios I:G+H) = 5.168 3 df (0.25 > p > 0.10) heterogeneity x2 (ratios G:H) = 67.614 3 df (p < < 6.01). 38 matus do not change between generations. However, some heterogeneous data were pooled for this comparison. Comparisions of phenotypic frequencies between generations in the Hybrid Pond are complicated because some heterogeneous data, within generations, must be pooled. However, some trends are clear. The ratios of phenotype G to phenotype H change significantly between gen-erations (as in the Waterfowl' Refuge). The frequency of phenotype I, although showing marked changes within generations, is relatively con-stant between generations. These trends are seen in Fig. 9. Phenotype I increased in frequency during 1969, 1970 and 1971. The i n i t i a l frequency during the late summer and f a l l months i s roughly 0.10 each year. This rises to 0.20 or 0.30 during the following spring and summer months. This trend in phenotype I seems to be independent of changes in the relative proportions of phenotypes G and H between generations. IV. Changes in phenotypic ratios - summary and conclusions 1. The ratios of phenotype G to phenotype H did not change significantly within generations in either study area. 2. The ratio of phenotype G to phenotype H changed significantly between generations in both study areas. In the Waterfowl Refuge the frequency of G increased from approximately 0.50 in 1969 to 0.71 in 1972. In the Hybrid Pond i t s frequency varied between 0.24 and 0.34. These changes in phenotypic frequency, by generation, are seen in Fig. 10. 39 3. The frequency of the I phenotype, found only in the Hybrid Pond, showed significant changes within generations. The data indicate a trend to increments in the frequency of I throughout each generation. 0.1 « 1 I I I 1968-69 1969-70 1970-71 1971-72 1972-73 G E N E R A T I O N Fig. 10. Frequencies of phenotypes between generations. G-W.R. refers to the frequency of phenotype G in the Waterfowl Refuge; G-H.P., and I-H.P. refers to the frequencies in the G and I phenotypes in the Hybrid Pond. 41 EXPLANATION OF THE CHANGES Two kinds of change in phenotypic frequency occur: changes within generations and changes between generations. The changes within generations occur in the relative frequency of the I phenotype in the Hybrid Pond. The changes between generations occur in the relative proportions of the G and H phenotypes in both study areas. To explain these changes I proposed a number of different hypotheses and attempted to test them in both the f i e l d and laboratory. There are three kinds of hypotheses: those explaining (1) both kinds of change (within and between generations); (2) changes within generations only; (3) changes between generations only. These hypotheses, and the reasons for suggesting them, are discussed below. I. Hypotheses explaining both kinds of change It is possible that the changes i n phenotypic frequency reflect mere errors in the methods used to collect the specimens, or errors in the assessment of the phenotypes. (The previous sections indicate that the changes are s t a t i s t i c a l l y significant, so normal, sampling error cannot explain them.) Therefore two hypotheses are relevant here. (!) The changes in phenotypic frequency occur as the result of non-random sampling; (2) error in assessment of the phenotypes of small fis h , whose plates were not f u l l y expressed, could explain the changes in phenotypic frequency. Consistent error in phenotypic assessment of large fish i s most unlikely. 42 If these hypotheses do not explain the changes, then other hypotheses must assume that the changes are real and that they have a biological explanation (i.e., they are determined by interaction(s) between the animal and i t s environment). One biological explanation relevant to both kinds of changes concerns differential dispersal. A greater tendency by one phenotype to move into, or out of either study area could explain the changes. Alternative explanations are best directed at explaining only one kind of change (i.e., changes within generations or changes between generations). These are considered separately below. II . Hypothesis explaining changes within generations If changes in phenotypic frequency within generations are not due to dispersion, or error either in the assessment or sampling procedures, then differential mortality i s the l i k e l y explanation. Predation i s a l i k e l y mechanism for differential mortality. Stickle-backs have been described as prey species elsewhere (Hoogland, Morris, and Tinbergen, 1956; Semler, 1971). Further differential predation on plate morphs (determined by the number of plates) is convincingly shown by Moodie (1972). Therefore, predation experiments, using two common predators from both study areas, were conducted to test the hypothesis that selective elimination by predation causes the changes. III . Hypothesis explaining changes between generations The collection data indicate that changes of phenotypic 43 frequency between generations are most pronounced i n the G and H pheno-types i n both study areas. However, phenotype I also changes from one generation to the next. The frequency of phenotype I was generally higher among the adult population (one-year-old-fish) than among the young f i s h (offspring) i n the subsequent generations. To explain the changes between generations I proposed that phenotypes d i f f e r i n t h e i r reproductive p o t e n t i a l , and to test t h i s hypothesis I considered and examined a number of factors i n c l u d i n g : (1) inheritance of the phenotypes; (2) v i a b i l i t i e s of crosses (frequencies of successful crosses; s u r v i v a l of eggs to hatching; post-hatching s u r v i v a l of f r y ) ; (3) phenotypic composition of breeding f i s h ; (4) f e c u n d i t i e s of females; (5) a s s o r t i v e mating. Each of these factors i s examined, separately. Following t h i s , the r e l a t i v e importance of each f a c t o r , as a mechanism determining f r e -quency changes i s compared. From the considerations described above, I concluded that the changes i n phenotypic frequency could be explained by: (1) non-random sampling; (2) i n c o r r e c t assessment of phenotype i n small f i s h ; C3) d i f f e r e n t i a l d i s p e r s a l ; (4) d i f f e r e n t i a l predation within generations; (5) d i f f e r e n t i a l reproductive or developmental biology among phenotypes (leading to changes between generations). 44 The rest of this study is concerned with the tests of these hypotheses, which involved a combination of laboratory experiments and fi e l d work. The particular methods and materials, results, and relevant discussion are given separately for each test. A general discussion follows. IV. Non-random sampling Non-random sampling i s always possible in a heterogeneous en-vironment. If sticklebacks assort themselves into micro-habitats, such as patches of vegetation or open, non-vegetated areas, then distorted estimates of phenotypic ratios may be caused by sampling being confined to specific areas. However, the collecting gear (described earlier) was sufficiently large to cover extensive areas and sampling was not confined to particular l o c a l i t i e s within either study area. Field tests were made in both study areas to determine i f the phenotypic composition varied over short distances. A. Waterfowl Refuge Small collections, each consisting of several seine hauls, were made at intervals of 200 to 300 m for about 1.5 km upstream from the main study area. The phenotypic ratios of the collections, shown in Table VII, are not significantly different (p > 0.25). Because most of the fish were collected outside the main sampling area, these collections are not included i n the analysis of the Waterfowl Refuge collections referred to earlier. 45 Table VII. Sampling tests in the Waterfowl Refuge. Number of Number of Collection G H TOTAL 1 32 20 52 2 30 23 53 3 37 24 61 4 50 43 93 5 24 27 51 6 19 22 41 TOTAL 192 159 351 heterogeneity x2 = 4.32 with 5 df (0.5 > p > 0.25) 46 B.. Hybrid Pond Usually Hybrid Pond collections consisted of many seine hauls made over a wide area. However, two experimental collections were made to test for non-random distribution of the phenotypes in the pond. Experimental collection 1. On September 23, 1971, two samples were taken from the pond, one on the east side and one on the west side. The phenotypic composition of each collection is shown in Table VIII a. Two age groups were present: young-of-the-year, and adults (fish begin-ning their second year of l i f e ) . Chi-square analysis indicates that the phenotypic ratios, for both age groups, are not significantly d i f -ferent between the two sides of the pond (p > 0.25). Experimental collection 2. On October 7, 1971, seine hauls were made at seven different places i n the pond. Sampling sites were about 40 m apart so that small collections were made around the entire circumference of the pond. Chi-square analysis indicates that the phenotypic fre-quencies in the collections (Table VIII;-b)are not significantly different. Only young-of-the-year fish are considered in this analysis. The collecting experiments indicate that phenotypic composition does not vary among the sampling sites in the study areas. Therefore, i t i s unlikely that non-random sampling would produce the observed changes in phenotypic frequency. V. Developmental changes i n phenotype The inclusion of small fish could bias the estimate of the phenotypic composition of a population i f the plates (plates H and I) 47 Table VIII. Sampling tests in the Hybrid Pond Table VIII a.. Experimental collection 1. 23 Sept., 1971 Collection number H TOTAL Age 0+ (Young of Year) West East 63 55 118 118 33 19 214 192 Age 1+ (Adults) TOTAL 118 236 52 406 heterogeneity x2 = 3.13 with 2 df (0.25 > p > 0.10), West East 6 8 25 33 10 21 41 62 14 58 31 103 heterogeneity x2 = 1-05 with 2 df (0.75 > p > 0.5). Table VIII h. Experimental collection 2, Oct. 7, 1971 Collection number G H I TOTAL 1 6 12 4 22 2 14 16 5 35 3 9 26 8 43 4 9 17 8 34 5 6 15 3 24 6 11 9 1 21 7 5 11 5 21 TOTAL 60 106 34 200 heterogeneity x2 = 11.35 with 12 df (0.5 > p > 0.25). 48 were not yet fu l l y expressed In the small fi s h . In the Waterfowl Refuge phenotypic ratios do not change within generations. This suggests that the assessment of small fi s h (taken in the summer and early f a l l ) from this area was not affected by developmental changes in phenotype. In the Hybrid Pond,: phenotype I, the form with the most plates, increases in frequency within generations. This suggests that plates may be added with time and growth. However, as the frequency of I increases, the frequency of G and H decreases ( a necessary consequence), but the proportions of G to H remain nearly constant and do not gen-erally change within generations. Small f i s h , that were really pheno-type I, but incorrectly assessed because of incomplete plate development, would probably be erroneously classified as phenotype H (the intermed-iate phenotype). Completion of plate development in phenotype I would then have the effect of changing the frequency of H relative to G. This does not occur in the collections. From the beginning of this study, I attempted to eliminate error in phenotypic assessment by staining a l l specimens with ali z a r i n dye and carefully examining them with a dissecting microscope. However, even with these measures, accurate counts of a l l the plates of small f i s h were not always possible. Therefore, I chose plate patterns, instead of plate numbers for phenotypic identification. Further, I emphasized the large plates (plates G, H, and I) that develop early in the l i f e of sticklebacks. This method of phenotype identification, plus the observations and considerations described earlier, indicate 49 that error in phenotypic assessment cannot explain the changes in phenotypic composition. VI. Dispersal If the tendency to disperse differs among the phenotypes, then changes in phenotypic frequency might be due to unequal immigration or emigration. Such factors are unlikely to affect the Hybrid Pond, since fish passage into, or. out of, the pond is impossible for most of the year. Dispersal, as a mechanism producing changes in phenotypic frequencies is more feasible in the Waterfowl Refuge study area. How-ever, there are three reasons to suggest that i t does not occur: (1) Beaver dams are common in this study area. In 1971 there were 6 dams within one mile upstream of the main collecting site and several more downstream. These dams do not seriously hinder the passage of anadromous salmonids, but do represent a formidable obstacle to the upstream passage of small fish l i k e sticklebacks. (2) Hagen (1967) carried out extensive dispersal tests in the Waterfowl Refuge study area (Hagen's station D). On March 19, 1965, he introduced 2000 leiurus, captured from an adjacent habitat, into the Waterfowl Refuge. Hagen had previously marked the fish by clipping their f i r s t dorsal spines. Several days later, he began to systematically search for marked fish. He found, 30 days after the introduction, that " v i r -tually no leiurus (marked) were recaptured beyond 600 ft - from the point 50 of r e l e a s e . Most were recovered w i t h i n 100 f t " . Even a f t e r 6 months, Hagen cont inued to recapture l a r g e numbers o f marked f i s h . (3) Hagen (1967) a l so observed tha t l e i u r u s i n o ther p a r t s of the r i v e r were sedentary . Approx imate ly one m i l e below the Waterfowl Refuge a l o c a l p a r a s i t e i n f e c t i o n (Neascus) produces b l a c k spots on about 50% of the s t i c k l e b a c k s . Hagen s t a t e s : "The h a b i t a t i s i d e a l and l a r g e numbers of l e i u r u s are present i n densely vegetated s h a l l o w s , ye t w i t h i n 300 f t e i t h e r upstream or downstream there are dense popu l a t i ons of l e i u r u s , p r a c t i c a l l y none of which are p a r a s i t i z e d . " The re fo re , from my o b s e r v a t i o n s , and from the r e s u l t s of Hagen (1967), d i f f e r e n t i a l d i s p e r s a l , as an exp l ana t i on of the changes i n pheno-t y p i c r a t i o s , i s r e j e c t e d . V I I . S e l e c t i v e p reda t i on and changes w i t h i n generat ions In the L i t t l e Campbell R i v e r , p r eda t i on i s a l i k e l y cause o f death f o r many s t i c k l e b a c k s . There are a number of known s t i c k l e b a c k predators i n both study a reas . These i n c l u d e : b i r d s (ducks and he rons ) , snakes, sa lmonid f i s h e s , and c e r t a i n aqua t i c i n s e c t s . Other s t u d i e s have demonstrated tha t p r e d a t i o n i s d i s p r o p o r t i o n a t e l y g rea te r on some s t i c k l e b a c k phenotypes than o t h e r s . For i n s t a n c e , Semler (1971) showed tha t the i n t e n s i t y of p r e d a t i o n by cu t th roa t t r o u t d i f f e r s acco rd ing to the polymorphic n u p t i a l c o l o u r a t i o n of males i n Lake Wapato, Washington S t a t e . A l s o , McPha i l (1969) found tha t f r y , produced from males w i t h b l a c k n u p t i a l c o l o u r a t i o n , are l e s s v u l n e r a b l e to p r e d a t i o n by the 51 mudminnow (Novurobra) than are fry produced from males with red nuptial colouration. Moodie (1972) found that intensity of predation by cutthroat trout varies with the plate number of sticklebacks. He suggested that differences in plate numbers among populations i n lakes on the Queen Charlotte Islands are determined by ithe abundance of predators. Hagen and Gilbertson (1972) propose that 7-plated sticklebacks are more re-sistant to predation than are those with other plate numbers, and they correlated trends i n phenotypic variation (of plate number) along the Pacific Coast of North American with the presence of predator species. In this study I tried to investigate the possibility that selective elminati6n:'by predation was responsible for the changes in phenotypic ratios. I conducted predation tests in the laboratory using two of the most common predators from both study areas: the water scor-pion (Ranatra fusca P. de B.), and the cutthroat trout (Salmo clarki). Other predators, present in both study areas, may be important in selectively eliminating stickleback phenotypes, but I had no time or space to include them in this study. A. Water scorpion predation Water scorpions are common but cryptic inhabitants of many parts of the L i t t l e Campbell River. I never observed Ranatra in the study areas by looking into the water, but I caught them frequently as they were crawling in the net after a seine haul. However, Ranatra were1 easily seen in Deek's Pond, a water f i l l e d gravel pit adjacent to 52 the river. In this pond, Ranatra appears to feed almost exclusively oh Gasterosteus. It i s d i f f i c u l t to estimate the abundance of this predator in the study areas because of i t s cryptic characteristics. Even when they are present in the seine net, they are frequently mistaken for debris and l i t t e r from the bottom. However, a rough estimate of the relative proportions of Ranatra to sticklebacks can be determined as follows: during the late winter and spring of 1971 about 50 Ranatra were captured in the Hybrid Pond and during the same period about 10 were captured in the Waterfowl Refuge. Comparing the number of Ranatra to the number of sticklebacks captured in the same interval gives a predator-prey ratio of approximately 1:38 in the Hybrid Pond and 1:99 in the Waterfowl Refuge. This suggests that Ranatra predation i s more intense in the Hybrid Pond. As a predator, Ranatra i s physically and behaviourally similar to the terrestrial praying mantid. It is a voracious "ambush" predator and in spite of i t s small size (6 to 7 mm in length), i t can capture and feed on adult and young sticklebacks as well as other aquatic organisms. Ranatra does not devour i t s prey, but merely sucks their juices. This process k i l l s sticklebacks, but does not affect their external morphology. When finished with i t s prey, Ranatra discards the intact carcass, and resumes i t s characteristic predatory hunting position. From observations i n the laboratory, i t seems that stickle-backs can escape only during the f i r s t few seconds after capture. Once 53 Ranatra has a firm grip, escape, even by large adults, is rare. 1. Predation experiments I conducted nine laboratory tests to determine whether pre-dation intensity by Ranatra differed among phenotypes. In each test, a predetermined number of sticklebacks, taken from random samples in the Hybrid Pond were exposed to Ranatra predators. Most of the tests were done in 40 l i t e r aquaria. The number of predators and prey di f -fered in each test (according to their a v a i l a b i l i t y ) , but usually 3 or 4 Ranatra were present. Vegetation was placed in each aquarium and the Ranatra were invariably positioned in i t . During each day of the test dead sticklebacks were removed from the bottom of the tank and preserved in 10% formalin. When about half the fish were dead, I stopped the test, and k i l l e d and preserved the survivors. Survivors and dead fish (those k i l l e d by Ranatra) were stained in alizeri n dye, assayed for phenotype, and their lengths were measured to the nearest mm. A single test never continued for more than a week, and most lasted only 2-3 days. The fish were not fed during a test so that negligible growth would be expected during this period. 2. Results a. Analysis by phenotype Table IX shows the relative proportions of the phenotypes in each test. Chi-square analysis indicates that the phenotypic ratios 54 Table IX. Phenotypic analysis of Ranatra predation experiments. The heterogeneity chi-square (corrected for continuity) i s not signi f i c a n t for individual or pooled results. In six of the nine tests, the phenotype with the largest i n i t i a l size increases in frequency. The l e t t e r L refers to leiurus phenotypes (G and H); I refers to semiarmatus phenotypes. Phenotype Phenotype With Whose Hetero- Largest Frequency geneity I n i t i a l Increases Test No. Survivors Freq . Dead Total Freq. X 2 Size 1. L 42 (.19) 43 85 (.17) I 10 7 17 Total 52 50 102 0.20 I I 2. L 40 (.80) 34 74 (.77) I 10 12 22 0.22 L L Total 50 46 96 3. L 20 (.77) 13 33 (.70) I 6 8 14 0.64 L L Total 26 21 47 4. L 19 (.73) 17 36 (.73) I 7 6 13 0.07 L -Total 26 23 49 5. L 21 (,78) 21 42 (.75) I 6 8 14 0.02 L -Total 27 29 56 6. L 36 43 79 I 15 (,29) 6 21 C 2 1 ) 3.47 I I Total 51 49 100 7. L 59 (.87) 43 102 (.84) I 9 (.13) 10 19 (.16) 0.32 I L Total 68 53 121 8. L 29 45 74 I 7 (.19) 7 14 (.16) 0.21 I I Total 36 52 88 9. L 34 29 63 I 10 (.23) 9 19 (.23) 0.01 I -Total 44 38 82 POOLED L 300 288 588 I 80 73 153 0.04 Total 380 361 741 55 of the dead (preyl fish, are not significantly different from the ratios among the survivors. This i s the case for each test (although test no. 6 approaches significance at the 0.05 probability level). Further, the pooled results for a l l the tests show no significant deviation of phenotypic ratios between the survivors and the prey. These tests indicate that Ranatra do not distinguish between prey phenotypes. However, they do appear to select their prey accord-ing to size. Table IX shows that in six of the nine tests, the pheno-type with the largest i n i t i a l size is the phenotype with the highest survival; two of the tests show no change; only one test (no. 7) shows a change in the opposite direction. These results, indicating that prey size might be related to prey vulnerability, prompted a different analy-sis of the data and some further tests. b. Predation intensity and prey size Three additional predation tests were conducted i n the same way as those described above. The data from a l l tests (a total of 12) were used to compare sizes between k i l l e d sticklebacks and the survivors (Table X). In 11 of the 12 tests the mean size of the survivors was greater than the mean size of the k i l l e d sticklebacks. An analysis of co-variance was conducted on the data from a l l the tests. Here the mean size of the sample, before any predation occurred, was compared with the mean sizes of both the surviving and k i l l e d fish. The two regressions resulting from this analysis (mean sample size vs. mean prey size, and mean sample size vs. mean survivor size) have similar slopes but significantly different elevations (Fig. 11). This suggests 56 Table X. Size analysis of Ranatra predation experiments Test Mean Size (mm) n Mean Size n Mean Size n'-of Sample of Killed of Survivors (Survivors & Killed) 1 22.77 102 23.00 50 22.56 52 2 30.11 96 29.54 46 30.64 50 3 31.23 47 29.57 21 32.57 26 4 29.39 49 29.22 23 29.54 26 5 32.39 56 31.33 29 33.00 27 6 26.31 100 25.78 49 27.80 51 7 24.25 121 23.79 53 24.60 68 8 24.33 88 23.56 52 25.44 36 9 25.60 82 25.32 38 25.84 44 10 26.38 84 24.02 45 29.10 39 11 35.02 59 34.06 32 36.15 27 12 21.19 43 20.70 27 22.00 16 Analysis Source of Covariance Regression df coefficient df %2-(Sxy)/£x2 Mean Square Killed 11 0.9558 10 4.9024 0.4902 Survivors 11 1.0291 10 6.1204 0.6120 Within 20 11.0228 0.5511 Regression 1 0.5191 0.5191 Common 22 0.9940 21 11.5419 0.5496 Adj. means 1 14.8051 14.8051 Total 23 22 26.3470 Test of difference in slopes: F = 0.5191/0.5511 = 0.9419, 1, 20 df (n. s. ) . Test of difference i n elevation: F = 14.8051/0.5496 = 26.9379, 1, 21 df (p«0.01) 57 13 14 15 3* 37 mean length of sample before predation Fig. 11. Size selective predation by Ranatra. The mean size of the dead (prey) and survivors are plotted against the mean size of the sample (before predation). 58 tha t Ranatra s e l e c t i v e l y chose sma l l e r i n d i v i d u a l s r ega rd l e s s of the s i z e ranges a v a i l a b l e . c. S i z e - s e l e c t i v e p reda t i on and prey a v a i l a b i l i t y I f the tendency of Ranatra to choose s m a l l e r prey i s r e a l , then predators should take p r o g r e s s i v e l y l a r g e r specimens, as the sma l l e r ones become l e s s a v a i l a b l e . This was t e s t ed by p l a c i n g 100 s t i c k l e b a c k s , o f v a r y i n g s i z e s i n an 80 l i t e r aquarium w i t h 6 Ranat ra . The s i z e s of the dead f i s h were measured accord ing to the t ime of death (at i n t e r v a l s of one or two days ) . A f t e r 10 days , 73 of the f i s h were dead and the experiment was te rmina ted . The r e s u l t s are shown i n Table XI and F i g . 12 . Gene ra l l y the sma l l e s t s t i c k l e b a c k s were among the f i r s t k i l l e d ; l a r g e r specimens su r v i v ed l o n g e r . The r e g r e s s i o n c o e f f i c i e n t (prey s i z e v s . day of e x p e r i -ment) i s p o s i t i v e (0.601) and s i g n i f i c a n t (p < 0 .05 ) . Th is conf i rms the t rend from the e a r l i e r t e s t s : Ranatra f i r s t take s m a l l p rey , and as these become l e s s a v a i l a b l e , they take p r o g r e s s i v e l y l a r g e r p rey . The re levance of s i z e - s e l e c t i o n p r e d a t i o n , as an exp l ana t i on of the changes i n phenotypic f requenc ies w i t h i n gene ra t i ons , i s d i s cussed l a t e r . B. Cut throa t t r o u t p r e d a t i o n The cu t t h roa t t r o u t (Salmo c l a r k i c l a r k i ) i s a common predatory f i s h i n the L i t t l e Campbell R i ve r and i s the ob j e c t o f l o c a l a n g l i n g a c t i v i t y . Cut throa t enter the upper reaches du r ing the f a l l and w i n t e r months to spawn; a t o ther t imes of the year they are probab ly con f ined to the lower areas of the r i v e r (around the Hybr id Pond), and some may 59 Table XI. Size selective predation and prey availability. Dav of Experiment 3 5 7 8 9 10 Survivors Total 18 3 2 5 19 2 2 20 4 1 1 2 8 21 4 1 1 6 22 3 1 1 5 23 1 1 1 3 24 1 4 2 7 25 3 3 1 1 3 11 26 5 2 2 2 1 6 18 C •H 27 1 2 1 2 2 3 11 CO 28 1 2 2 1 6 12 N •rl 29 1 1 1 4 7 30 2 2 31 1 1 32 33 number 28 14 13 25 98 (2 specimens were not recovered) Analysis of Regression (Weighted) - Survivors Excluded Source df S.S. .M.S.. TOTAL 72 733.1233 Among Days 5 170.1751 34.0350 4.7623** Linear Reg. 1 155.8819 155.8819 21.8117* Deviation 2 14.2934 7.1467 < 1 - n.s Within Days 67 562.9482 8.4022 Regression coefficient = 0.601. **(p < 0.01) *(p < 0.05) 32 r i i i i i i • • 3 4 S * 7 S 9 10 surv ivors Day of exper iment Fig. 12. Size-selective predation and prey a v a i l a b i l i t y . Prey size taken by Ranatra predators i s plotted against the day of the experiment. 61 enter marine waters. Small trout (15 to 20 cm), were occasionally taken from the Hybrid Pond. These fish probably entered the pond during high water and became trapped when water levels declined. I have observed these trout feeding on sticklebacks i n the lab-oratory, but have no direct evidence of their feeding on sticklebacks in the L i t t l e Campbell River. Other studies, however, provide evidence of natural Cutthroat trout predation on sticklebacks (Semler, 1971; Moodie, 1972). 1. Predation experiments Cutthroat trout, similar in size to those found in the Hybrid Pond (15 to 20 cm), were obtained from the Provincial Government Fish Hatchery in Abbotsford, B.C. Collections of sticklebacks, to be used as prey, were made in the irrigation ditches i n Ladner; their sizes were generally similar to those of the Hybrid Pond sticklebacks. Two predation tests were conducted: (1) Six trout were placed in a 800 l i t e r tank. About 1200 sticklebacks were randomly divided into three groups. One group, the control, was ki l l e d and preserved; the other two groups were placed in the tank with the cutthroat trout. After three days, when over half of the sticklebacks had been consumed, I ended the test, removed the survivors, and compared their length frequency with that of the control group. (2) Procedures for the second test were similar to those described above, except that 10 trout were used in a 4000 l i t e r tank. Roughly 4000 stickle-backs were used, 1348 as a control, the remainder exposed to the trout. 62 2. Results (Table XII and Fig. 13) In both tests the mean size of the survivors was significantly greater than that of the control group (p < 0.01). This indicates that the trout selectively preyed on the smaller sticklebacks. Many of the sticklebacks used in these tests were too small for phenotypic analysis, but as discussed below, the results of these tests, and the predation tests using Ranatra, are important in explaining changes in phenotypic frequencies between generations. C. Relevance of the predation experiments In Table XIII, the mean size of the G and H phenotypes (or leiurus phenotypes ) i s compared with the mean size of phenotype I (or semiarmatus phenotype) for a l l the Hybrid Pond collections. The larger phenotype (mean length in mm) is underlined. Each sample was compared for differences in variances and means. If the variances were signif-icantly different, the means were compared by a modified " t " test (Steele and Torrie, 1960). In 23 of 33 collections, semiarmatus was larger than leiurus. Semiarmatus was generally larger in the f a l l and winter, but during the early spring and summer the size differences between the two groups diminished. The fact that relative sizes of the phenotypes vary, and that predators selectively prey on smaller f i s h , i s consistent with the hypothesis that selective elimination i s responsible for the changes in phenotypic frequency. If, at the beginning of each generation, semiar- matus is generally larger than leiurus, then predation which selectively 63 Table XII. Results of cutthroat trout predation experiments. Test No. 1. Test No. 2.  Total length no. of no. of no. of no. of in mm. control survivors control survivors 13 1 1 14 7 12 5 15 11 1 47 20 16 14 3 100 40 17 25 2 153 84 18 44 6 149 104 19 43 13 162 121 20 46 14 151 1146 21 38 15 129 120 22 38 17 97 133 23 27 17 71 107 24 31 16 65 81 25 24 14 54 95 26 17 15 49 84 27 14 14 19 55 28 15 12 25 55 29 1 12 13 41 30 10 15 17 35 31 3 6 15 24 32 2 2 9 12 33 9 6 ' 5 13 34 3 4 2 10 35 2 3 3 5 36 1 3 4 37 3 4 38 1 3 39 1 number 430 218 1348 1394 mean lengths 21.94 25.44 20.56 2: comparison of means t =8.78 p « 0.01 t = 12.56 p « 64 Test No. 1 Test No. 2 40 3» 36 34 32 30 '.It 26 24 22 _* w O J) 20 18 16 14 U 12 10 Survivors Control Control Survivors Fig. 13. Cutthroat trout predation tests. Mean sizes of sticklebacks are shown by the thin horizontal line; the thick, dark area represents the standard error (on either side of the mean); the light bar represents the standard deviation (on either side of the mean); the thin vertical line represents the ranges. 65 Table XIII. R e l a t i v e s i z e s of l e i u r u s and semiarmatus phenotypes i n the Hybrid Pond. The lar g e r s i z e i s underlined. If variances d i f f e r e d s i g n i f i c a n t l y , the means were compared by a modified " t " test (Steele and T o r r i e , 1960). A s i n g l e a s t e r i s k i n d i c a t e s s i g n i f i c a n c e at the 0.05 p r o b a b i l i t y l e v e l ; two a s t e r i s k s at the 0.01 p r o b a b i l i t y l e v e l . The d i f f e r e n c e (d) between means i s designated as p o s i t i v e when semiarmatus i s l a r g e s t , and negative when l e i u r u s i s l a r g e s t . l e i u r u s semiarmatus  Mean Mean Date Age No. s i z e No. s i z e Freq. F t d July 15,1968 0+ 43 30 .05 12 30. .75 0.218 1.11 0. 67 +0. 70 1+ 184 44 .36 65 44. ,51 0.261 1.00 0. 24 +0. .15 Sept.17,1968 0+ 123 30 .30 40 30. .70 0.173 1.39 0. 50 +0. .40 1+ 79 46. .11 16 48. .69 0.173 1.20 2. 05* +2. .58 March 17,1969 0+ 131 31, .05 32 33. .37 0.196 1.07 1. 57 +2. .32 May 7,1969 0+ 123 38 .14 49 36. ,27 0.171 1.04 1. 44 -1. ,87 Aug.20,1969 0+ 283 30 .47 35 30. 74 0.110 1.33 0. 34 +0. ,27 Jan.29, 1970 0+ 122 33 .62 32 32. .28 0.208 1.41 9. 95 -1. .34 May 12,1970 0+ 145 41, .96 32 38. .84 0.181 1.05 1. 95 -3. .84 June 10,1970 0+ 91 47 .87 27 46. .26 0.229 1.21 1. 33 -1. ,61 Aug. 5,1970 0+ 163 24 .50 20 25. .70 0.109 1.79* 1. 57 +1. .02 Sept.8,1970 0+ 186 27, .38 Al 30. .61 0.181 1.04 3. 74** +3, .23 O c t . l , 1970 0+ 502 24, .51 76 29. .49 0.131 1.20 6. 42** +4. .98 Nov. 6,1970 0+ 448 22, .27 37 25. .65 0.076 1.55 6. 21** +3. .38 Feb. 2,1971 0+ 263 27 .46 48 32. ,85 0.145 1.11 4. 01** +5. .39 March 17,1971 0+ 210 23 .66 20 25. .75 0.087 1.01 2. 59** +2. ,09 A p r i l 14,1971 0+ 248 24 .70 46 26. .37 0.156 1.13 1. 91 +1, .67 A p r i l 26,1971 0+ 333 24, .74 83 26. .10 0.200 1.91* 2. ,34* +1, .36 May 18,1971 0+ 254 33 .66 87 32. .30 0.255 1.59* 0. ,99 -1, .36 June 9,1971 0+ 175 40 .87 59 39. .88 0.252 1.16 0. 97 -0, .99 June 22,1971 0+ 446 41 .79 217 39. .95 0.327 1.74* 3. ,77 -1, .84 July 30,1971 0+ 165 27 .47 24 29, .08 0.127 1.02 1. ,82 +0. .83 1+ 72 44 .76 34 45.59 0.321 1.24 0. ,46 +0, .63 Sept.23,1971 0+ 356 28 .13 52 29. .58 0.127 1.09 1. ,83 +1. .45 1+ 72 44 .79 31 45, .42 0.301 1.05 0. ,58 +0, .63 Oct.7,1971 0+ 166 30, .74 34 31, .58 0.170 1.24 0. 75 +0, .84 Nov.4,1971 0+ 319 27 .81 87 28. .55 0.236 1.07 1. 19 +0, .74 Feb. 22,1972 0+ 243 29.45 66 28, .75 0.214 1.34 0. .86 -0, .70 Apr.15,1972 0+ 345 32 .46 84 30, .61 0.196 1.20 2. ,46* -1, .85 May 16,1972 0+ 129 37 .60 22 35, .54 0.146 1.49 1. ,34 -2, .06 June 22,1972 0+ 78 40.87 30 41, .23 0.278 1.10 0. 83 +0, .36 July 13,1972 0+ 30 24 .07 8 25, .13 0.210 1.89 0. ,78 +1, .06 1+ 129 38, .79 73 39. .34 0.361 1.19 0. .91 +0. .55 66 eliminates, smaller f i s h w i l l have two consequences: CI), the size difference between leiurus and semiarmatus w i l l decrease because only the largest leiurus w i l l survive; (2) the frequency of semiarmatus w i l l increase because predation on this group, i s less intense. Both effects are seen in the data, especially during the 1970 and 1971 generations, and are shown in Fig. 14. Here, the difference in mean size between leiurus and semiarmatus i s arrayed against the frequency of semiarmatus (larger mean sizes for semiarmatus are considered positive). In the summer and f a l l of 1970 semiarmatus was larger than leiurus. In February, 1971, the size difference started to diminish, and concurrently, the frequency of semiarmatus increased. Early in the 1971-72 generation, semiarmatus was again larger than leiurus, but the size difference started to decrease during the late summer, and at this time the frequency of semiarmatus started to increase. Generally, the data support the hypo-thesis that size-selective predation causes changes in phenotypic fre-quencies within generations (i.e., increase in frequency of semiarmatus). Other changes in frequency however, such as those occurring during the breeding seasons, may be due to other factors such as phenotypic di f f e r -ences in post-spawning mortality. D. Explanation of the changes within generations - summary and conclusions The observed changes in phenotypic frequency within generations are not explained by: error i n sampling; error in phenotypic assessment; or differential dispersal between the phenotypes. Alternative explanations 67 Fig. 14. Relative size and frequency of semiarmatus in the Hybrid Pond. Each line represents a different generation. The semiarmatus phenotype is generally larger than the leiurus phenotypes at the beginning of the generation. Later, the size difference decreases, and this is accompanied by a relative increase in the relative frequency of the semiarmatus phenotype. 68 must i n v o l v e some k i n d of s e l e c t i v e e l i m i n a t i o n of some phenotypes from the p o p u l a t i o n . P reda t ion by water sco rp ions and c u t t h r o a t t r ou t was i n v e s t i g a t e d as an exp l ana t i on of the changes. Both these predators tend to prey s e l e c t i v e l y on the s m a l l e s t s t i c k l e b a c k s , but appear not to d i s t i n g u i s h between phenotypes. The r e s u l t s of the p r eda t i on exper iments , however, are c o n s i s t e n t w i t h the hypothes i s that s e l e c t i v e p reda t i on causes the i nc rease i n the frequency of semiarmatus i n the Hybr id Pond. Semiarmatus i s g e n e r a l l y l a r g e r than l e i u r u s du r i ng the summer. L a t e r , the f requency of semiarmatus i n c r e a s e s , but the s i z e d i f f e r e n c e between the two groups decreases . There fo re , p r eda t i on may be more i n t ense on l e i u r u s du r ing the summer and f a l l . The e f f e c t of t h i s p r eda t i on on l e i u r u s lowers i t s f requency, but i nc reases i t s mean s i z e r e l a t i v e to tha t of semiarmatus. V I I I D i f f e r e n t i a l r ep roduc t i on and changes between generat ions A. The crosses Genet ic e f f e c t s on the de te rmina t ion of p l a t e number have been e s t a b l i s h e d by s e ve r a l workers . Munzing (1959) was the f i r s t to r epor t on the i n h e r i t a n c e of p l a t e number i n Gas te ros teus ; he proposed a s i n g l e l o c u s , two a l l e l e system to e x p l a i n the d i f f e r e n c e s among t rachurus (TT), semiarmatus (T t ) , and l e i u r u s ( t t ) , but d i d not check h i s conc lus ions by doing F2~generat ion c r o s s e s . However, s i m i l a r r e s u l t s i n F]_ c rosses are repor ted by Hagen (1967) and McPha i l (pers . comm.). L indsey (1962) i n v e s t i g a t e d s a l i n i t y and temperature as 69 env i ronmenta l f a c t o r s tha t might a f f e c t p l a t e number. He found that these f a c t o r s had l i t t l e i n f l u e n c e on p l a t e s , a l though they a f f e c t e d numbers of ve r tebrae and g i l l r a k e r s . U n l i k e the prev ious work, the c rosses performed i n t h i s study were designed to i n v e s t i g a t e the gene t i c c o n t r o l of the p l a t e p a t t e r n -s p e c i f i c a l l y f a c t o r s a f f e c t i n g the presence or absence of p l a t e s G, H, and I. Because these charac te rs ( i n d i v i d u a l p l a t e s ) are not con t inuous l y d i s t r i b u t e d , most conven t iona l es t imates of h e r i t a b i l i t y are i n a p p r o p r i a t e ( Fa l coner , 1960) . S i m i l a r i l y , there are no f i r m a p r i o r i reasons to assume s imple Mendel ian s eg r ega t i on , a l though segregat ion was observed i n p r e l i m i n a r y c rosses . I n i t i a l l y there were no reasons e i t h e r to con f i rm or deny any assumptions rega rd ing the i n h e r i t a n c e of p l a t e s , except perhaps to accept tha t gene t i c f a c t o r s are c l e a r l y i n vo l ved and that the environmental f a c t o r s desc r ibed by L indsey (1962) are r e l a t i v e l y unim-p o r t a n t . 1. Methods and M a t e r i a l s F i s h used i n the c rosses were e i t h e r i n r ep roduc t i ve c o n d i t i o n when taken from the f i e l d or were brought i n t o breed ing c o n d i t i o n i n the l a b o r a t o r y . I assessed the phenotypes o f f i s h used f o r c rosses a f t e r they had been k i l l e d , p rese r ved , and s t a i n e d acco rd ing to the procedures desc r i bed e a r l i e r . Crosses were made throughout the b reed ing season of 1971 and pa r t o f 1972. I made c rosses by s t r i p p i n g females of t h e i r eggs and f e r t i l -i z i n g them w i t h a sperm suspens ion , p r e v i o u s l y prepared by e x c i s i n g and 70 s l i c i n g the testes in a few jnls of water. After a drop of the sperm suspension was placed on the eggs, the eggs and sperm were thoroughly-mixed with blunt forceps to ensure f e r t i l i z a t i o n . Then the eggs were immersed in water and incubated at 18°C. After one day of incubation the eggs were water-hardened, and could be examined and counted without damage. Dead eggs, which were soft or opaque, were removed and recorded. Eggs hatched about 7 days after f e r t i l i z a t i o n . Two or three days later, when the fry began to feed, I transferred them to aquaria and gave them daily feedings of brine shrimp nauplii (Artemia)until they were large enough to eat Tubifex worms. Generally, fry were reared to as large a size as possible, then k i l l e d and fixed in 10% formalin. After fixation they were stained with alizeri n dye and their phenotypes recorded. a. Waterfowl Refuge crosses The parents used for the crosses were not selected for pheno-type. Rather, ripe females were used either as they were captured or as they came into breeding condition in the laboratory. In this way crosses made throughout the breeding season are representative of the types of matings occurring in nature. b. Hybrid Pond and other crosses Very few crosses were made using parents originating from the Hybrid Pond as i t was d i f f i c u l t to transport l i v e fish between the pond and the collecting vehicle that had to be parked about one-half mile away. Instead, crosses were made using parents collected from the 71 Ladner Ditches. These f i s h had similar phenotypic yariation to those in the pond (i.e., phenotypes G, H, and I were present). 2. Results of crosses Results from the crossing experiments are presented in Appendix Table II; the tables in this section are only summaries. For the purposes of analysis the crosses are grouped as follows: (1) leiurus crosses (within and between G and H phenotypes) where both parents originated from the Waterfowl Refuge - a pure leiurus area; (2) leiurus crosses where one, or both, parents originated from a mixed population (i.e. the Hybrid Pond or Ladner ditches) where both leiurus and semiarmatus are found; (3) F£ crosses where both parents were brother and sister of G phenotype; (4) other crosses where one or both parents was a semiarmatus (or I phenotype) or trachurus. a. Frequency of successful crosses Pre-hatching mortality varied among the crosses: some failed to produce any fry; others produced varying numbers. In Table XIV the fre-quency of successful crosses, which yielded one or more fry, i s compared to the frequency of a l l the crosses that were attempted. The frequency of successful crosses i s similar, regardless of parental origin, except, perhaps, for the leiurus crosses in which one parent comes from a mixed population. Here H x H crosses were more suc-cesful (7/9) than the G x H crosses (9/17); G x G crosses were unsuccesful (0/3). The small numbers preclude tests of significance. The reason for the high frequency of unsuccessful crosses Table XIV. Frequence of successful crosses. Type of cross-G x G Go*x H 9 G?x Ho* H x H G, or H x I I x I G, or H x trachurus Waterfowl Refuge crosses:  No. crosses 52 30 31 36 attempted No. successful (produced some fry) 28 16 16 20 Success rate 0.54 0.53 0.52 0.56  Success rate Ladner and Hybrid Pond crosses:* No. crosses attempted 3 10 7 9 31 4 5 No. successful 0 5 4 7 16 1 2 0 0.50 0.57 0.78 0.52 0.25 0-40 F2 (brother x sister) crosses: No. crosses attempted 13 1 No. successful 8 1 Success rate 0.62 (l-Pl) *one or both parents from Ladner or Hybrid Pond. 73 (approximately 50%) is unknown. The use of premature eggs, or testes, may be partly responsible, but I doubt i t because (1) obviously premature egg clutches, when extracted from a female, were discarded; (2) even in unsuccessful crosses eggs developed normally during the f i r s t four or five days and substantial mortality did not occur u n t i l the hatching period. b. V i a b i l i t i e s among successful crosses The survival of the progeny from the successful crosses was com-pared over two periods: from f e r t i l i z a t i o n to hatching, and from hatching un t i l the fry stage (Table XV and Appendix table II)-. The crosses are grouped according to.the origin and phenotype of the parents. Tests of significance on the differences in survival among the groups are inappropriate because of heterogeneity within each group. Analysis is further complicated by differences in egg number in the crosses. For these reasons survival estimates (both pre-hatching and post-hatching) were determined in two ways: (1) For each group the i n i t i a l egg number, the number of eggs surviving to hatching, and the number of fry were determined. Pre-hatching survival was estimated as the fraction: . *. . Post-hatching survival i n i t i a l egg no. was estimated by the fraction: no. of fry  i n i t i a l egg no. (2) For each cross the pre-hatching and post-hatching survival were _ , no. eggs hatched , no. of fry „ . -estimated as: -—. "g, and -—:———* . Survival i n i t i a l egg no. i n i t i a l egg no. estimates of the group were determined by calculating the average survival Table XV. Vi a b i l i t i e s of successful crosses. Estimates of survival to hatching Freq. of I n i t i a l No. Freq. of No. of survival to No. of egg that hatching Average fry fry stage Average crosses no.(a) hatch(b) (a)/(b) freq. (c) (c)/(d) freq. Waterfowl Refuge crosses: G x G 24 1602 1377 0.859 0.844 729 0.455 0.470 G^xH? 15 1339 1137 0.849 0.876 479 0.385 0.333 G?xHtf 12 963 855 0.888 0.887 251 0.293 0.279 H x H 18 1860 1385 0.745 0.759 620 0.333 0.366 F2 (brother x sister) crosses: G x G 8 639 458 0.717 0.710 108 0.169 0.188 Ladner and Hybrid Pond crosses: GCxH? 5 551 461 0.837 0.828 174 0.316 0.337 G&Ho* 4 448 326 0.727 0.750 87 0.194 0.193 H x H 7 903 735 0.814 0.825 178 0.197 0.210 G x I 8 629 493 0.783 0.802 175 0.278 0.320 •H x I 8 895 522 0.583 0.580 243 0.272 0.273 75 of the crosses w i t h i n the group. The f i r s t es t imate of s u r v i v a l cons iders d i f f e r e n c e s i n egg number and may be cons idered to be we igh ted ; the second es t imate i s unweighted. The estimates of ha t ch ing success are s i m i l a r among the groups o f c r o s s e s ; most es t imates vary between 75 and 85%. Values f o r the F2 crosses are s l i g h t l y lower (71%) and va lues f o r the H x I c rosses are l ow -es t of a l l (58%). However the range of ha t ch ing success va lues i n H x I c rosses v a r i e s between 32 and 96%, and i t seems l i k e l y tha t the low average va lue i s a t t r i b u t a b l e to sampl ing e r r o r r a the r than to b i o l o g i c a l p rocesses . The es t imates of s u r v i v a l to the f r y do d i f f e r among the groups. The h ighes t success i s seen i n the G x G crosses (parents from the Waterfowl Refuge) w i t h s u r v i v a l es t imates of 45 to 47%. This i s approx imate ly 25% h igher than that f o r other phenotypic c rosses from the same a rea , which have s u r v i v a l es t imates between 29 and 36%. The F2 c rosses have the lowest s u r v i v a l to the f r y s tage (17 to 19%) , wh ich suggests tha t consanguineous crosses r e s u l t i n i nb reed ing dep re s s i on . S u r v i v a l est imates of l e i u r u s c rosses from mixed p o p u l a t i o n s . These s u r v i v a l es t imates i n d i c a t e tha t G x G c ro s ses , i n pure l e i u r u s p o p u l a t i o n s , may y i e l d more progeny than the G x H, o r H x H c ro s ses . Higher s u r v i v a l among the G x G c rosses may be a s i g n i f i c a n t f a c t o r determin ing changes i n phenotypic frequency between genera t ions . T h i s , however, would be dependent on the phenotypic compos i t ion of the progeny of these and other c r o s s e s . Th is sub jec t i s cons idered below. 76 c. Phenotyp ic a n a l y s i s of the progeny In F i g . 15a-c, the number of f i s h (progeny) w i t h G phenotype are ar rayed a long the a b s c i s s a and the number of H phenotypes a long the o r -d i n a t e . Each symbol represents a separate c r o s s ; symbols c l o s e to the o r i g i n i n d i c a t e c rosses w i t h s m a l l numbers of f r y , and those f a r from the o r i g i n i n d i c a t e crosses w i t h many f r y . The crosses are d i v i d e d i n t o 4 c a t e g o r i e s : (1) G x G; (2) G x H; (3) H x H; and (4) c rosses us ing phenotype I. A l though there i s much v a r i a t i o n w i t h i n each type of c r o s s , there are a l so some d i s t i n c t d i f -ferences between them. (1) G x G Crosses ( F i g . 15a) o f t e n produced on l y G o f f s p r i n g . Phenotype H was produced i n some crosses but u s u a l l y a t a low f requency . Severa l F£ c rosses produced a ma jo r i t y of phenotype H. No I phenotypes were produced from any G x G c r o s s . (2) G x H Crosses ( F i g . 15b) produced the g rea tes t v a r i a t i o n i n phenotypic r a t i o s ; s e ve r a l produced on ly G o f f s p r i n g and one produced on ly H o f f s p r i n g . Genera l l y the r a t i o s are between these two extremes, and no t rend towards g rea te r p ropo r t i ons of e i t h e r phenotype i s obv ious . Fu r t he r , no obvious d i f f e r e n c e s between r e c i p r o c a l c rosses are seen. No I phenotypes were produced from any G x H c r o s s . (3) H x H Crosses ( F i g . 15c) always produced some H o f f s p r i n g , and i n most cases H was more f requent than G. Only one cross produced e n t i r e l y H phenotypes. No I phenotypes were produced from any H x H c r o s s . (4) Crosses us ing phenotype I as one parent and e i t h e r phenotype G o r H as the other parent are shown i n F i g . 15d. Here the number of 77 Fig. 15. Phenotypic composition of the progeny. In each group the number of G progeny is plotted along the abscissa and the number of H progeny along the ordinate. (a) G x G crosses; (b) G x H crosses; Cc) H x H crosses; (d) leiurus x semiarmatus crosses, and results of two leiurus x trachurus crosses. 78 I (semiarmatus) are arrayed along the abscissa and the. number of the leiurus phenotypes (G + H) along the ordinate. Phenotype I was produced in every cross and generally accounted for about 50% of the progeny in most crosses. Fig. 15d also shows the results of two crosses between leiurus and trachurus. Both produced high frequencies of I. One I x I cross (actually an F2 - brother x sister cross) yielded six leiurus and two I. d. Inheritance of lateral plates The observation that crosses made among phenotype G can produce phenotype H suggests that some individuals with the G phenotype carry the genetic capacity (or part of the capacity) to produce plate H, but this capacity i s not always expressed. This i s consistent with the ob-servation that crosses made among phenotype H usually produce substantial numbers of G. The inheritance of plate I, however, is different. Plate I always occurs in some offspring of crosses where one parent was pheno-type I, but i t never occurs in the offspring of leiurus x leiurus crosses. This is true both for crosses made from pure leiurus populations (Waterfowl Refuge), and leiurus x leiurus crosses made from mixed pop-ulations, (Hybrid Pond and Ladner ditches). Sixteen leiurus x leiurus crosses, where one or both parents came from a mixed population, produced 410 progeny - none was an I phenotype. Although I cannot explain the phenotypic ratios of the crosses by simple Mendelian segregation, I can offer some speculative conclusions concerning the inheritance of the plates H and I. 79 Genetic control of plate H probably involves two or more inde-pendently segregating l o c i . The number of segregation ratios that can be generated from a single locus, regardless of the number of alleles, are too few to explain the observed ratios of the crosses, and consistent segregating classes with frequencies less than 0.25 would not be expected. Further, phenotypic ratios observed from the crosses do not support an assumption of many l o c i controlling plate H. If many l o c i were involved (an assumption usually made for her i t a b i l i t y estimates) then the crosses should exhibit average effects of the two parents. Continuously d i s t r i -buted t r a i t s , such as length, weight, or fecundity are l i k e l y to be con-trolled by interactions among many l o c i (Falconer, 1960). If many l o c i were controlling the presence or absence of a plate, then the frequency distribution of phenotypic ratios in crosses would l i k e l y be unimodal, and probably normal; most H x H crosses would be expected to yield high frequencies of H offspring. In the absence of epistatic or dominance effects, one would not expect some H x H crosses to yield both high fre-quencies of G and others high frequencies of H, as shown in Fig. 15c'. The wide range in phenotypic ratios, which do not appear clustered around a mean, suggest a limited number of segregating l o c i . The ratios from the leiurus x semiarmatus crosses suggest that one major locus i s associated with plate I. Fifteen of 16 crosses seg-regate closely to a 1:1 ratio (one leiurus to one semiarmatus). Most semiarmatus are l i k e l y heterozygotes (Ii) with leiurus being homozygous ( i i ) . Capital letters need not imply dominance of al l e l e ' I'; rather co-dominance 80 i s suggested by the phenotyp ic r a t i o s o f the c r o s s e s . Crosses between l e i u r u s and t rachurus would be expected to produce on l y he te rozygotes or semiarmatus ( I i ) . Th is i s c l o s e to the r e s u l t s of two c rosses i n t h i s study ( F i g . 15d) and r e s u l t s of Hagen (1967) and Munzing (1959). Th is exp l ana t i on may be o v e r s i m p l i f i e d because recombinat ion between semiarmatus c rosses ( I i ) should produce t rachurus ( I I ) , but f u l l y -p l a t e d f i s h , presumably t r a c h u r u s , are very r a re i n both of the mixed popu l a t i ons i n v e s t i g a t e d i n t h i s s tudy . F u r t h e r , one G x I c ross produced 33 l e i u r u s to 3 semiarmatus, which i s i n c o n s i s t e n t w i t h the assumption of a s i n g l e l o c u s , 2 a l l e l e h ypo thes i s . This suggests tha t the l ocus may break down i n t o sma l l e r segrega t ing u n i t s i n some he te rozygo tes . However, f o r the most p a r t , p l a t e I seems to be c o n t r o l l e d by one major l o c u s , and i t seems reasonable tha t the gene t i c c o n t r o l o f p l a t e H would l i k e w i s e be determined by a l i m i t e d number of l o c i . On t h i s b a s i s , a hypothes is e x p l a i n i n g the seg rega t i on r a t i o s was formulated and i s d i s cussed be low, e. A hypothes i s on the i n h e r i t a n c e of p l a t e H The range of phenotypic r a t i o s among the l e i u r u s c rosses can be exp la ined by an assumption of two segrega t ing l o c i ; one w i t h two a l l e l e s and the o ther w i t h t h r e e . Dominance w i t h i n l o c i , and e p i s t a s i s between l o c i , are r e q u i r e d . L e t h a l or i n v i a b l e genotyp ic combinat ions a s s i s t i n f i t t i n g the hypo thes i s to the d a t a . The model i s shown below i n the form of a Punnett square : l ocus 2 ( a l l e l e s X , Y, and Z) locus 1 XX YY YY XY YZ ZZ ( a l l e l e s V and W) w G G 1 G G 1 VW G H H G G H WW H H H G H H 81 There are two l o c i : one has two a l l e l e s (V and W), and the other has. three a l l e l e s C;X, vY, and Z). The s p e c i f i c genotypes which correspond to each phenotype are not presented f o r the sake of s i m p l i c i t y . Each a l l e l e i s designated with a d i f f e r e n t c a p i t a l l e t t e r . The l e t t e r ' i ' represents a genotypic combination which i s l e t h a l or r e l a t i v e l y i n v i a b l e . The expected phenotypic r a t i o s generated by the model correspond w e l l to those observed. This i s seen i n F i g . 16, where the l i n e s represent a l l p o ssible expected frequencies and the points represent the observed values (as i n Figs. 15a-c) from the crosses. The close approximation of the hypothetical r a t i o s to the ob-served data does not constitute support for the model because the model was conceived on the basis of the data, and designed to explain i t . As such, the model i s an a p o s t e r i o r i hypothesis and requires t e s t i n g . For the purpose of explaining the data i n Mendelian terms I assume that the phenotypic r a t i o s of the progeny are unaltered by s e l e c t i v e m o r t a l i t y e i t h e r at the pre-hatching or post-hatching stages. This assump-t i o n may not be correct, e s p e c i a l l y i n view of the higher s u r v i v a l of the G x G progeny seen e a r l i e r . I f s e l e c t i v e m o r t a l i t y does a l t e r pheno-typic r a t i o s of the progeny, then the genetic explanation may be simpler than that proposed here. f. A quantitative estimate of inheritance I determined the average r e s u l t s of each type of cross by pooling the data i n a manner s i m i l a r to that described e a r l i e r f o r the estimates of v i a b i l i t y . Two estimates for the average phenotypic r a t i o s were t o -CO cr LU , OO Z : i o -ta) G X G / / ' ' s / / s / s s ' ' ' s " - ^ ^  1* 2 0 SO 4 0 SO NUMBER OF G ~r ~i 4 0 7 0 7 0 -• o -3 0 -U. 40—1 O a o -(b) G X H ' / ' ' / / / / / / / / / / / / / / / / i / ' 0 i z *>-; 1 4 * * --~ -----0 io 20 30 40 50 60 70 NUMBER OF G 7 0 - ' 6 0 -5 0 -U- 4 0 -O 5 z> z to-rn H X H ' / // i i i i 10 1 0 3 0 4 0 SO 4 0 70 NUMBER OF G Fig. 16. Predicted phenotypic ra t i o s . The dashed lines represent a l l phenotypic ratios predicted by the hypothesized model; the ci r c l e s represent the observed values. In each graph, the number of G progeny i s plotted along the abscissa and the number of H progeny along the ordinate. 83 calculated as follows: (1) For each type of cross, the total number of fry of one phenotype was divided by the total number of fry produced. (2) The phenotypic ratio was calculated for each cross. For each group of crosses (i.e. G x G, H „ H etc.) the frequencies were summed and divided by the number of crosses. The two estimates, shown in Table XVI, are similar. Crosses made with fish from the Waterfowl Refuge show the follow-ing patterns: G x G crosses produce mostly G offspring (app. 90%); H x H crosses produce the fewest G offspring (app. 50%); G x H crosses produce intermediate frequencies (app. 67%). The G x H crosses from mixed pop-ulations yield similar phenotypic frequencies (60-80%) as G x H crosses from the pure leiurus population, but the H x H crosses produce fewer G offspring (27-31%). This suggests that the genotypic composition of H phenotypes from Ladner and the Hybrid Pond (mixed populations), i s different than the genotypic composition from the Waterfowl Refuge (pure population). The G x G, F2 crosses (brother x sister) generally produced fewer G offspring than the non-consanguineous G x G crosses. Crosses using one parent of phenotype I produced the least number of G offspring (G x I crosses produced 24-31% G; H x I produced 12-14% G). The propor-tion of I phenotypes in the progeny was highest among the H x I crosses (47-48%) and lowest in the G x I crosses (38-40%). B. Fecundity I estimated fecundity for each phenotype in both study areas by Table XVI. Phenotypic composition of the progeny. No. of No. of Freq. of Average No. of fry G progeny No. of G progeny freq. of crosses (a) (b) H progeny (b)7(a) G. progeny Waterfowl Refuge crosses: G x G 28 757 681 76 0.900 0.905 G3CH9 16 490 331 159 0.675 0.682 G$xHo" 16 309 205 104 0.663 0.673 GxH (pooled) 32 799 536 263 0.671 0.678 H x H 20 636 328 308 0.516 0.492 F2 (brother x sister) crosses: G x G 8 108 71 37 0.675 0.545 Ladner and Hybrid Pond crosses: GdkH? 5 145 116 29 0.800 0.742 G?xHo* 4 87 52 35 0.598 0.674 • H x H 7 178 49 129 0.275 0.309 No. of No. of No. of Freq. of Average freq. crosses fry leiurus (G+H) No. of leiurus progeny of leiurus (a) progeny (b) I progeny (b)/ (a) progeny Semiarmatus crosses: I x G 8 175 109 66 0.623 0.603 (freq. of G=0.31) (freq. of G=0..24) I x H 8 253 116 117 0.519 0.527 (freq. of G=0.14) (freq. of G=0.12) 85 counting eggs (egg clutches from single females) from perserved f i s h taken i n the collections, and from females used in the crosses. Egg counts were made throughout the breeding season. In the Waterfowl Refuge, H females have significantly higher fecundities than G females (Fig. 17 and Table XVII). Differences in fecundity among the phenotypes in the Hybrid Pond are not significant, but lik e the Waterfowl Refuge, H females have higher fecundities than G females. Hybrid Pond females have much lower fecundities than Waterfowl Refuge females. The reason for the differences in fecundity between phenotypes, and between study areas i s unknown. Fecundity i s often associated with size (positive co-variance) in many fishes (Norman and Greenwood, 1963) and this may be the case with these populations of sticklebacks, but size-fecundity relationships are not considered here. The significance of differences - in fecundity are discussed later. C. Analysis of breeding populations I analysed the collections made during the breeding seasons of 1971 and 1972 to determine the relative numbers of breeding fi s h . The cri t e r i a used to distinguish breeding from non-breeding fish was based on the degree of gonad development and several other factors. A female was considered to be in breeding condition i f the ovaries were enlarged and contained developing (or developed) eggs. Malescwere considered to be in breeding condition i f the testes were enlarged and darkly pigmented. Testes in this condition would invariably produce viable sperm. Also, during previous laboratory work, I noticed that males which were rearing 86 200 ISO E Z 100L Oft 0) so W a t e r f o w l Refuge H y b r i d Pond H O O ( 4 9 ) r H (»•) (m) Fig. 17. Fecundities of the phenotypes. The thin horizontal line shows the mean egg number for each group; the dark bar shows one standard error (on either side of the mean); the light bar shows one standard deviation (on either side of the mean); the thin vertical line represents the range. 87 Table XVII. Analysis of fecundities. G females Waterfowl Refuge Number of fish examined 49 H females 54 Mean egg number 87.96 106, .07 Standard error 4.28 4. .32 Comparison of means t = 2.96, df = 101 (p < 0.01) Analysis of variance Source • . df Mean Square F Between phenotypes 1 8425. .393 8.74**(p<0 Within phenotypes 101 964. 709 Total 102 Hybrid Pond Number of fish examined Mean egg number Standard error Comparison of (extreme) means (G and H) Analysis of variance G females 38 51.92 3.01 H females 112 56.59 2.15 I females 44 61.75 4.55 t = 0.764, df = 80 (0.5 > p > 0.4) Source df Mean Square F Among phenotypes 2 993.621 1.734 Within phenotypes 191 572.985 Total 193 88 eggs, or guarding young, sometimes had enlarged, but lig h t l y pigmented testes. This condition was usually accompanied by especially bright nuptial colour around the throat. Males with these features, when captured from the f i e l d , were classified as breeding males. 1. Age of breeding f i s h Some fish, classified as in breeding condition, were probably two years old. The frequency of these fish i s generally low in the pop-ulation as a whole, but this age group appears to make a disproportionate contribution to the breeding population. The frequencies of breeding fi s h , analysed by sex, phenotype, and age, are shown i n Tables XVIII and XIX. Two-year-old fish made up only 0.6% of the Waterfowl Refuge population, and 0.5% of the Hybrid Pond population during the 1971 breed-ing season. However, nearly a l l of the two-year-old fish were in breeding condition, and they constituted about 2.4% of the breeding populations in both study areas. The frequency of two-year-old fish was higher during the 1972 breeding season: 6.8% of the Waterfowl Refuge population and 6.9% of the Hybrid Pond population. Nearly a l l of these fish were in breeding condition and made up 32% of the breeding population in the Waterfowl Refuge and 15% of the breeding population in the Hybrid Pond. The cause and significance of these differences in age structure of the breeding populations is unknown. There are no striking differences in the phenotypic proportions of breeding fish between the age groups, although, during Table XVIII. Age distribution during the breeding season in the Waterfowl Refuge. Table XVIII a. 1971 breeding season. Population A l l breeding at large Go* Go H ? fish Age 1 928 51 69 70 58 248 Age 2 6(0.6%) 3 _1 _0 _2 6(2.4%) Total 934 54 70 70 60 254 Table XVIII b. 1972 breeding season. Population A l l breeding at large Go* H-o* G? H ? fish Age 1 438 26 13 17 11 67 Age 2 32 (6.8%) _8 _5 _7 12 32(32.4%) Total 470 34 18 24 23 99 90 Table X I X . Age distribution during the breeding season in the Hybrid Pond. Table X I X a. 1971 breeding season. Population at large Her Id* G? H? I ? A l l breeding fish Age 1 Age 2 1238 7 33 4 89 0 57 1 24 1 65 0 20 1 288 7 (2.4%) Total 1245 37 89 58 25 65 21 295 Table X I X b. 1972 breeding season. Population at large Go" Hd* Id* G$ H? I ? A l l breeding fish Age 1 Age 2 880 65 27 6 62 19 33 7 31 2 75 17 23 4 251 55 (17.9%) Total 945 33 81 40 33 92 27 306 91 1972, the frequency of phenotype H among the two-year-old breeders was slightly higher than in the one-year-old breeders. In further analyses, I have attempted to determine i f pheno-typic frequencies of breeding fish change during the breeding season, and whether the phenotypic composition of the breeding population differs from that of the population as a whole. For the purposes of these analyses, I ignored the ages of breeding fish. However, the possible significance of changes in age composition of breeding f i s h , between years, i s discussed later. 2. Phenotypic frequencies during breeding seasons (Tables XX and XXI) The phenotypic frequencies of breeding fish do not change throughout the breeding season in either study area, except for a sig-nificant increase in the frequency of semiarmatus females in the Hybrid Pond, during the 1972 breeding season. With this exception, the differ-ences in phenotypic frequency among breeding fish can be explained wholly by sampling error. 3. Comparison of breeding and non-breeding fish I pooled the data for a l l the collections made during the breed-ing seasons and calculated a single estimate of the phenotypic frequency of breeding fish during that period. The phenotypic frequency of the breeding fish is compared to that of the non-breeding fish by chi-square analysis (Tables XXII and XXIII). a. Waterfowl Refuge During 19.71 the frequency of phenotype G among breeding fish was lower than the frequency of G in the non-breeding fish. The frequency 92 Table XX. Analysis of breeding populations in the Waterfowl Refuge. The numbers include both one-, and two-year-old fish. The numbers of two-year-old fish are shown in brackets. Table XX a. 1971 Breeding season Total sample Breeding males Breeding females Frequency Date G H Total G H Total G H Total of breeders 14 April 43 38 81 3 11 14 8 10 18 0.395 26 April 100(2) 70(1) 170(3) 11(2) 20 31(2) 21 18(1) 39(1) 0.411 4 May 133(1) 92(1) 225(2) 12(1) 14(1) 26(2) 14 16 30 0.248 17 May 129 91 220 13 12 25 16 6 22 0.214 2 June 66 45 111 6 4 10 6 6 12 0.198 9 June 83 44(1) 127(1) 9 9 18 5 4(1) 9(1) 0.212 Total 554 380 934 54 70 124 70 60 130 Male heterogeneity x 2 = 5.81 with 5 df (0.5 > p > 0.25) Female heterogeneity x 2 = 4.49 with 5 df (0.5 > p > 0.25) Table XX b. 1972 Breeding season Total sample Breeding males Breeding females Frequency Date G H Total G H Total G H Total of breeders 30 March 68(3) 34(4) 102(7) 7(2) 2 9(2) 6(1) 6(4) 12(5) 0 198 27 April 91(5) 40(4) 131(9) 11(2) 1 12(2) 9(3) 5(4) 14(7) 0 191 16 May 98(6) 52(6) 150(12) 9(3) 8(4) 17(7) 5(3) 6(2) 11(5) 0 187 24 May 62(2) 25(2) 87(4) 7(1) 7(1) 14(2) 4 6(2) 10(2) 0 275 Total 319 151 470 34 18 52 24 23 47 Male heterogeneity x 2 = 6.89 with 3 df (0.10 > p > 0.05) Female heterogeneity x 2 = 1-61 with 3 df (0.9 > p > 0.75) Table XXI. Analysis of breeding populations in the Hybrid Pond. The numbers include both one-, and two-year-old f i s h . The numbers of two-year-old f i s h are shown in brackets. Table XXI a. 1971 Breeding season. Total sample Breeding males Breeding females Frequency Date G H I Total G H I Total G H I Total of breeders 18 May 50(3) 207 89(2) 346(5) 6(3) 15 6(1) 27(4) 1 14 3(1) 18(1) 0 .130 9 June 44(1) 132 59 235(1) 8(1) 25 13 46(1) 6 10 4 20 0 .280 22 June 114(1) 333 217 664(1) 23 49 39 111 18(1) 41 14 73(1) 0 .277 Total 208 672 365 1245 37 89 58 184 25 65 21 111 Male heterogeneity x 2 = 2.64 with 4 df (0.75 > p > 0.5) Female heterogeneity x 2 = A.46 with 4 df (0.25 > p >0.10) Table XXI b. 1972 Breeding season. Total sample Breeding males Breeding females Frequency Date G H I Total G H I Total G H I Total of breeders 15 A p r i l 113(8) 263(23) 93(9) 469(40) 11(6) 22(9) 9(5) 42(20) 10(2) 40(14) 6(4) 56(20) 0. .209 16 May 51 91(13) 24(2) 166(15) 7 19(10) 4(2) 30(12) 14 24(3) 2 40(3) 0. ,422 22 June 24 54 30 108 8 16 13 37 4 11 7 22 0. .546 13 July 35 94 73 202 7 24 14 45 5 17 12 34 0. .391 Total 223 502 220 945 33 81 40 154 33 92 27 152 Male heterogeneity x = 6.33 with 6 df (0.57 > p > 0.25) Female heterogeneity x 2 = 20.22 with 6 df (p < 0.01) vO 94 Table XXII. Comparison of phenotypic frequencies, between breeding and non-breeding f i s h in the Waterfowl Refuge. Table XXII a. 1971 Breeding season. Phenotype G Phenotype H Total No. Freq. No. Freq.  Breeding males 54 0.435 70 0.565 124 Breeding females 70 0.538 60 0.462 130 Non-breeding f i s h 4 3 0 0.632 250 0.368 680 Total 554 0.593 380 0.407. 934 Male heterogeneity x 2 = 16.15 with 1 df (p < 0.01) corrected for continuity Female heterogeneity x 2 = 3.68 with 1 df (0.1 > p > 0.05) corrected for continuity Table XXII b. 1972 Breeding season. Phenotype G Phenotype H Total No. Freq. No. Freq. Breeding males 34 0.654 18 0.346 52 Breeding females 24 0.511 23 0.489 47 Non-breeding fish 261 0.704 110 0.296 371 Total 319 0.679 151 0.321 470 Male heterogeneity X 2 = 0.32 with 1 df (0.75 > p > 0.50) corrected for continuity Female heterogeneity x 2 = 6.22 with 1 df (p < 0.05) corrected for continuity 95 Table XXIII. Comparison of phenotypic frequencies between breeding and non-breeding fi s h in the Hybrid Pond. Table XXIII a. 1971 Breeding season Phenotype G Phenotype H Phenotype I Total No. Freq. No. Freq. No. Freq.  Breeding males 37 0.201 89 0.484 58 0.315 184 Breeding females 25 0.225 65 0.586 21 0.189 111 N o n - b r e e d i n g 1 4 6 0.154 518 0.545 286 0.301 950 Total. 208 0.167 672 0.540 365 0.293 1245 Male heterogeneity X 2 = 3.34 with 2 df (0.25 > p > 0.10) Female heterogeneity x 2 = 7.74 with 2 df (0.025 > p > 0.01)* Table XXIII b. 1972 Breeding season Phenotype I Total No. Freq. No. Freq. No. Freq. Breeding males 33 0. 214 81 0. 526 40 0. 260 154 Breeding females 33 0. 217 92 0. 605 27 0. 178 152 Non-breeding fish 157 0. 246 329 0. 515 153 0. 239 639 Total 223 0. 236 .502 0. 531 220 0. 233 945 Male heterogeneity x = 0.74 with 2 df (0.75 > p > 0.50) Female heterogeneity x 2 = 4-36 with 2 df (0.25 > p > 0.10) 96 of breeding G males (43.5%) is. significantly lower (p < 0.01)_ than that of the non-^breeding individuals (63.2%_, The difference _ between the frequency of G females (53.8%) and the non-breeding population approaches significance at the 0.05 probability level. During 1972, as in 1971, the frequency of G phenotype was lower in breeders than the non-breeders. The difference between the males (65.4%) and the non-breeders (70.3%) was not significant ( p > 0.50) but the difference between the females (51.1%) and the non-breeders was quite significant (p > 0.02). During both breeding seasons phenotypic proportions of breeding fish were different from those expected on the basis of the phenotypic composition of the non-breeding fish. In 1971 the greatest discrepancy was seen in the phenotypic proportions of breeding males; in 1972 i t was seen in the breeding females, b. Hybrid Pond The phenotypic proportions of breeding fish did not differ sig-nificantly from the non-breeding individuals. The phenotypic ratios of each sex are representative of the population as a whole. ',4. Relative frequency of breeding fish*. The proportion of fish in breeding condition changes throughout the breeding season. For instance, during A p r i l , 1971, about 40% of the fish captured in the Waterfowl Refuge were classified as breeders. Later in the same season breeders had declined to about 21% (Table XX). The cause of the decline is not clear, but I do not think that i t i s due to fish changing their breeding condition (i.e., changing from breeders 97 to non-breeders), for the following reason: during the breeding season, and immediately afterwards, dead fi s h are frequently found around the edge of the water and are often scooped from the bottom in the seine. Many appear to be recently-dead males, and show the nuptial colouration around the throat, characteristic of breeding condition. Therefore i t seems l i k e l y that the changes in abundance of breeding fish are due to mortality among the breeding individuals, and recruitment of non-breeding fish into the breeding group as ;the season progresses. A consequence of mortality among spent individuals in the breeding population in the Waterfowl Refuge would be an increase in the proportion of phenotype G with time. This i s expected because the proportion of H phenotypes is higher among the breeders than among the non-breeders. Higher mortality among the breeders would have the effect of decreasing the frequency of phenotype H, or increasing the relative frequency of G as the breeding season progressed. This trend occurred during both years (Figs. 8 and 9) but was most pronounced in 1971. D. Habitats of breeding f i s h The sampling tests, described earlier, indicated that the pheno-types were randomly distributed over the two study areas. Both study areas, however, are characterized by intermittent patches of vegetation although vegetation i s thicker in the Hybrid Pond. Attempts to sample fish exclusively i n vegetated areas, and alternatively, from open,' non-vegetated areas were unsuccessful. A clear demarcation between open and vegetated areas does not exist in 98 much of the study areas. Rather, there were a few open areas, areas with light or moderate vegetation, and areas with dense vegetation. A single seine haul would usually cover a l l extremes of the habitat in terms of the degree of vegetation. Smaller sampling gear, such as a small hand net, was largely ineffectual because i t was slow and often frightened the fish as i t approached. Under these circumstances, fish in open areas usually seek shelter in vegetated areas. I observed this even for breeding males with a nest. Deek's Pond (Fig. 1), provided an opportunity to test for habitat preference among the phenotypes. Several discrete patches of vegetation, about 10 m i n diameter, are surrounded by areas with no vegetation. The substrate of these areas i s a mixture of rocks (gravel) and sand. Mud i s the predominant substrate of the vegetated areas. In June, 1971, I sampled the pond with a 15-meter beach seine, and made separate collections from open and vegetated habitats. I analysed the fish according to their habitat, phenotype, and breeding condit ion (Table XXIV). Most were probably one year old but some may have been older. No fry or young-of-the-year were included in this analysis. Both males and females were classified as either breeders or non-breeders (immature). Most breeding males in vegetated areas were phenotype H, but most in open areas were phenotype G. The difference between their distributions approaches significance at the 0.05 probability level (Table XXIV-a). Females show a similar, but less pronounced trend 99 Table XXIV. Habitat analysis leiurus phenotypes.. heterogeneity No . in No. in yl (corrected for vegetation open Ca)Breeding males:G 32 42 74 x2=3.84 with 2 df H 46 30 76 (0.10 > p > 0.05) Total 78 72 150 (b)Breeding females: G 15 23 38 X2=0.39 with 2 df H 15 15 30 (0.75 > p > 0.50) Total 38 30 68 (c) Non-breeding males: G 7 25 32 x2=0.0^6 with 2 df H 4 11 15 (p > 0.99) Total 11 36 47 (d) Non-breeding females: G 32 24 56 X2=3.48 with 2 df H 13 24 37 (0.10 > p > 0.05) Total 45 48 93 (e) Breeding males and breeding females: G 47 65 112 x2=4.68 with 2 df H 61 45 106 (p < 0.05) Total 108 110 218 (f) Breeding males compared to non-breeding males: Breeding males 78 72 150 X2=10.69 with 2 df Non-breeding (p < < 0.01) males 11 36 47 Total 89 108 197 100 (Table XXIV b). When the data for both sexes are combined, the difference is habitat distribution between phenotypes i s highly significant (Table XXIV e). The phenotypic frequencies of the non-breeding males do not differ between the habitats (Table XXIV c), but hardly any non-breeding males are found in vegetated areas. The difference between the habitats of breeding males, compared to non-breeding males, i s highly significant (Table XXIV f) . Non-breeding female, r-phenotypes tend to assort in an opposite direction to the breeding fish (Table XXIV d) but the difference is not significant. Positive assortment of the phenotypes into different habitats, especially among the breeding fi s h , suggests the possibility of assort i v e - , mating among the phenotypes. The significance of this i s discussed later. IX Explanation of the changes between generations: a synthesis of previous sections A number of differences in reproductive biology were demonstrated. The most important of these are summarized below: (1) Genetic differences between the phenotypes as determined by the pheno-typic composition of the progeny from different crosses. A genetic mechanism i s suggested for leiurus phenotypes (G and H) that results in an abundance of G offspring in crosses from the Waterfowl Refuge. Similar but less pronounced trends are noted among the leiurus crosses from mixed populations. The inheritance of phenotype I (or plate I) seems clearer, and i s probably controlled by a single locus with two co-dominant alleles. Approximately 50% of the offspring of leiurus x semiarmatus crosses are phenotype I. 101 (2) Differences in v i a b i l i t i e s among the crosses. Survival of fry was highest in the G x G crosses. (3) Significantly higher fecundities in H females than in G females from the Waterfowl Refuge. The fecundity data from the Hybrid Pond shows the same trend, but the differences are not significant. (4) An apparent reproductive advantage associated with the H phenotype in the Waterfowl Refuge (the frequency of H is significantly higher in the breeding population than in the population at large). The relative fre-quency of the three phenotypes (G, H, and I) among breeding fish in the Hybrid Pond is similar to that in the population at large. However, the frequency of phenotype I increases throughout the year. This results in a higher frequency of I among breeding fish than is present earlier in the year. The increase in the frequency of phenotype I within gener-ations is probably due to a selective advantage associated with predation, but the net result is a reproductive advantage. (5) Differences in micro-habitat between G and H breeders. In the one area (Deek's Pond) where this could be tested, both G males and females were more common in open, non-vegetated areas, and H males and females were more common in vegetated areas. These differences appear to substantiate the hypothesis that changes in phenotypic frequency between generations are the result of differences in reproductive biology between the phenotypes. However, I think that examination of other aspects of the reproductive biology might show more differences between the phenotypes. For instance, the ab i l i t i e s of the males to attract females to their nests, or to rear 102 young in their nests, may vary between the phenotypes. Similarly, the number of egg clutches produced by individual females during the breeding season may vary between the phenotypes. Compiling a l i s t of differences between the phenotypes, although useful, does not indicate the relative importance of each difference in explaining the changes in phenotypic ratios between generations. One way of assessing the relative significance of the differences i n re-productive biology is to use the data presented in the previous sections to calculate, a posteriori, the expected phenotypic ratios for each gen-eration, and then compare this with the observed ratios. The simplest predication of expected phenotypic ratios in the fry of the next generation, requires only two kinds of information: (1) the relative proportions of each phenotype in the breeding population; (2) an estimate of the phenotypic composition of fry expected from a l l possible crosses. A prediction based on this information assumes that random mating occurs among the phenotypes, and that differences in v i a b i l i t y and fecundity are not important. (It also assumes that the estimates of inheritance, and proportions of breeding phenotypes are accurate.) This simple prediction can then be adjusted to incorporate differences in v i a b i l i t y or fecundity between the phenotypes and can be altered to consider the effects of assortive or non-random mating between the phenotypes. 103 A. Predicting phenotypic ratios The estimates of the inheritance of plates and the data on the frequency of phenotypes breeding, can be combined to predict the fre-quency of each phenotype in the fry of the next generations. This i s done separately for each study area. Waterfowl Refuge The phenotypic frequencies of breeding fish are as follows: 1971 1972 G males 0.435 0.650 H males 0.565 0.350 G females 0.538 0.510 H females 0.462 0.490 The expected frequency of each kind of cross, assuming random mating i s : 1971 expected frequency G x G: CO-435)CO.538) = 0.234 Gx.H: (0.435).CO.462) + (0.565)(0.538) = 0.505 . H x H: (0.565) (0.462) = 0.251 1.000 1972 expected frequency G x G: CO.650)CO.510) = 0.332 G x H: (0.650)CP-490) + (0.350)CO.510) = 0.497 H x H; (0-350)CO.490) = 0.171 1.000 The expected phenotypic ratios in the fry of these crosses i s : G x G - 90% G; 10% H G x H - 67% G; 33% H • _ _ , 1 (Estimates from Table XVI) H x H - 50% G; 50% H 104 The expected phenotypic composition for the fry for each breeding season is then determined as follows: relative frequency frequency 1971 1972 proportion Cross Cross frequency G progeny H progeny G H G X G: 0.234 0.90 0.10 0.21 0.02 G x H: 0.505 0.67 0.33 0.34 0.17 H x H: 0.261 0.50 0.50 0.13 0.13 0.68 0.32 G x G: 0.332 as as 0.30 0.03 G x H: 0.497 above above 0.335 0.16 H x H: 0.171 0.085 0.08 0.72 0.28 On the basis of two factors: the breeding frequency of the adults, and the phenotypic frequency of progeny in the crosses, the calculated fre-quency of G offspring in the 1971 breeding season i s 68%. The estimate for G progeny from the 1972 breeding season is 72%. Hybrid Pond The observed phenotypic frequency in breeding f i s h from this area i s : 1971 1972 G males 0.201 0.214 H males 0.484 0.526 I males 0.315 0.260 1.000 1.000 G females 0.225 0.217 H females 0.586 0.605 I females 0.189 0.178 1.000 1.000 Random mating among these phenotypes would produce the following combina-tions of phenotypic crosses: 105 1971 1972 G x G 0.045 0.046 G x H 0.228 0.244 G x I 0.109 0.095 H x H 0.283 0.318 H x I 0.276 0.251 I x I 0.059 0.046 1.000 1.000 Estimates of the phenotypic composition of the progeny for each type of cross are based on the results of Ladner crosses using semiarmatus and H parents; expectations from the G x G crosses are based on Waterfowl Refuge crosses. Expected frequency % contribution to 1971 frequency of cross of offspring next generation G H I G H I G X G 0.045 0.9 0.1 4.1 0.5 _ G X H 0.228 0.67 0.33 - 16.0 6.8 -G X I 0.109 0.31 0.31 0.38 3.4 3.4 4.1 H X H 0.283 0.31 0.69 - 8.8 19.5 -H X I 0.276 0.14 0.38 0.48 3.9 10.4 13.2 I X I 0.059 (0.10) (0.15) 0.75 0.6 0.9 4.4 Total 1971 - 36.8% + 41.5% + 21.7% 1972 = 100% G X G 0.046 0.9 0.1 - 4.1 0.5 -G X H 0.244 0.67 0.33 - 16.3 8.0 -G X H 0.095 0.31 0.31 0.38 2.95 2.95 3.6 H X H 0.318 0.31 0.69 - 9.9 21.9 -H X I 0.251 0.14 0.38 0.48 3.5 9.5 12.0 I X I 0.046 (0.10) (0.15) 0.75 0.5 0.7 3.5 37.25 43.55 Total 1972 - = 99.9% 10 6 The calculated and observed estimates of the phenotypic frequency for each study area and each year are shown below. Waterfowl Refuge: 1971 1972 expected frequency of G 68% 72% observed frequency of G 70.7% 58.5% expected frequency of H 32% 28% observed frequency of H 29.3% 41.5% Hybrid Pond expected frequency of G 36.8% 37.3% observed frequency of G 30.2% no estimal expected frequency of H 41.5% 43.6% observed frequency of H 57.1% -expected frequency of I 21.7% 19.1% observed frequency of I 12.7% -1972. The observed estimates from the Waterfowl Refuge represent the average of a l l the collections taken during each generation. The observed estimate from- the Hybrid Pond i s based on my f i r s t collection in 1971 (July 30), because significant changes in frequency occur in later collections. The prediction for the 1971 generation in the Waterfowl Refuge i s close to the observed values, but the prediction for the 1972 gen-eration i s not, and overestimates the frequency of phenotype G. The prediction for the 1971 generation in the Hybrid Pond overestimates the frequency of phenotype H, but underestimates the frequency of phenotype I. This discrepancy between predicted and observed estimates may be due to incorrect assumption(s). The assumptions made for the 1D7 prediction are: (11 that mating i s random; (2) that v i a b i l i t i e s are similar among different types of crosses; C3) that inheritance of pheno-type i s similar between years (and to a limited extent, between study areas); (4) that the reproductive potential of the phenotypes are iden-tical. In the following sections, I adjust the prediction to account for error in these assumptions. 1. Effects of assortive mating If positive assortive mating between the phenotypes were complete (i.e. only G x G, H x H and I x I crosses occurred) the expected pheno-typic composition of the progeny can be calculated for each study area. With complete, positive assortive mating, the most important parameter is the relative proportion of females of each phenotype. If a l l other factors are equal, the number of males of each phenotype is not important. Complete positive assortive mating in the Waterfowl Refuge (based on the frequency of breeding females), generates 71.5% G progeny during 1971 breeding season, and 70.4% G progeny during the 1972 breeding season. In the Hybrid Pond the estimate for 1971 is 51% G, 35% H, and 14% I progeny. If positive assortive mating does occur in nature, i t is unlikely to be complete, but rather somewhere between the two extremes of complete assortive and complete random mating. A non-random distribution between breeding phenotypes was observed in Deek's Pond (discussed earlier). If i t i s assumed that matings occurred only between fish within each habitat then, based on the frequencies of breeding individuals within each habitat, the phenotypic composition of the progeny in each habitat i s : 108 v e g e t a t i o n : 66% G, 34% H open a r eas : 71% G, 29% H. The important p o i n t emerging from t h i s c a l c u l a t i o n i s tha t the assumption of a s s o r t i v e mating does not apprec i ab l y a l t e r the phenotyp ic compos i t ion expected i n the f r y from tha t c a l c u l a t e d on the b a s i s of random mat ing . Th is i s not t rue f o r the Hybr id Pond,but here the assumption of a s s o r t i v e mat ing i nc reases the d i sc repancy between the c a l c u l a t e d and observed v a l u e s . Incomplete* or on ly p a r t i a l l y e f f e c t i v e a s s o r t i v e ma t ing , as assumed f o r Deek 's Pond, changes the phenotyp ic r a t i o s expected i n the f r y by on l y 5%. Th is suggests that i f a s s o r t i v e mating does occu r , i t i s r e l a t i v e l y i n e f f e c t i v e i n a l t e r i n g phenotypic r a t i o s between gene r a t i ons . 2. E f f e c t s of d i f f e r e n c e s i n f e cund i t y The f e c u n d i t y of H females i n the Waterfowl Refuge i s s i g n i f -i c a n t l y h ighe r than G females (by a f a c t o r of approx imate ly 1 .2 ) . Th is was determined by d i v i d i n g the mean egg numbers o f H females (106.07) by the egg number of G females (87 .96 ) . A d j u s t i n g the c a l c u l a t e d pheno-t y p i c f requency of f r y , by assuming tha t H females produce about 1.2 t imes the number of eggs as G fema les , changes the expec ta t i ons to the f o l l o w i n g : 1971 f r y - expect 67% G; 33% H 1972 f r y - expect 70.8% G; 29.2% H. The d i f f e r e n c e s between these f i g u r e s , and the p r e d i c t i o n s when f e cund i t y d i f f e r e n c e s are i g n o r e d , are l e s s than 2%. Phenotyp ic d i f f e r e n c e s i n f e cund i t y o f Hybr id Pond f i s h are 109 insignificant and these small differences do not markedly change the estimates of phenotypic frequencies among the fry. Therefore, fecundity differences, l i k e differences in assortive mating, seem unimportant as mechanisms responsible for changes in phenotypic frequency. 3. Effects of differences in v i a b i l i t y Three estimates of v i a b i l i t y were determined: CI) The frequency of successful crosses; (2) the frequency of eggs surviving to hatching; (3) the survival of eggs to fry. (1) The frequency of successful crosses was similar among the crosses from the Waterfowl Refuge, but i t seemed different among leiurus crosses from mixed populations. Table XIV shows that no G x G crosses (0/3) were successful; 0.53 or 9/17 of the G x H crosses were successful; 0.78 or 7/9 of the H x H crosses were successful; 0.51 or 16/31 of the leiurus x semiarmatus crosses (G or H x I) were successful; and 0.25 or 1/4 of the I x i crosses were successful. These estimates of the prob-a b i l i t y of success change the predicted phenotypic frequency in the Hybrid Pond to: 35% G; 47% H; 18% I. This i s closer to the observed ratios (30% G; 57% H; 13% I ) , but some discrepancy remains. (2) The estimates of the v i a b i l i t i e s of the eggs ( f e r t i l i z a t i o n to hatching) show no consistent differences between the types of crosses. The predicted estimate, of the phenotypic composition of fry is not altered by consideration of this data. (3) The estimates of survival to the fry stage do differ among the crosses from the Waterfowl Refuge. The survival to the fry stage 110 in the G x G crosses (about 46%) is- about 1.4 times greater than the survival in the other crosses (about 33% in G x H and H x H crosses). Consideration of this factor changes the predicted estimate to 70% G in 1971 and 74% G in 1972. The resulting changes are not consistent with the observed differences. B. Effects of differences in reproductive biology - summary and conclusions Computing a predication of the phenotypic composition of the fry for a particular generation, based on the estimates of inheritance and the relative proportions of phenotypes among the parents, seems a reasonable way to test certain assumptions about phenotypic va r i a b i l i t y in natural populations. However, in many ways, such a prediction i s most useful when i t f a i l s , (i.e., A correct prediction may be right for the wrong reasons.) The prediction of phenotypic ratios for the fry in the Waterfowl Refuge during 1971 was the only one that came close to the ob-served values. The 1972 prediction and the prediction for the 1971 gen-eration in the Hybrid Pond are inconsistent with observed values. The predictions were adjusted to account for the effects of assortive mating, differences in fecundity, and differences in v i a b i l i t y , but these did not reduce the difference between the predicted and observed phenotypic fre-quencies . Therefore, other differences in the reproductive biology between the phenotypes may be important in determining changes in phenotypic ratios I l l between generations. Females might differ in the number of clutches they produce. Males may differ in their a b i l i t i e s to rear young or attract females to their nests. The observation that the age structure of the breeding populations changed between 1971 and 1972 may also be relevant to explaining the changes. In 1971, the number of 2 year old fish in the breeding population was very low in both study areas. In 1972, a considerable number of 2-year-old fish were found in breeding condition in both populations. The frequency of phenotype H was slightly higher among these breeders than in the one-year-old breeders. If the 2-year-old fish produced proportionately more H offspring than the average one-year-old breeder (perhaps some genetic difference permits survival for a second year) then the predicted estimates of phenotypic composition would underestimate the frequency of the H phenotype. This may explain the difference between the predicted and observed estimates of phenotypic composition for the 1972 generation in the Waterfowl Refuge, where the predicted estimate of phenotype H was lower than the observed value. Of the factors examined in this study, the phenotypes of the parents (specifically their genotypes, because i t is certain that one phenotype consists of more than one genotype) and the relative abundance of each kind seem to be the most important factors determining the changes between generations. Other factors, such as difference in v i a b i l i t y , fecundity, and assortive mating seem less important. Under different environmental circumstances, and different selective pressures, these components may assume much more importance. 112 DISCUSSION Distribution, zoogeography, and evolution of Gasterosteus Gasterosteus i s a coastal animal occurring in marine and fresh-water habitats adjacent to most major land masses in northern temperate areas. It is. circumpolar, but in spite of i t s widespread occurrence along coastlines i t seldom penetrates far into freshwater. Most freshwater populations are found within 200 km of the sea. Trachurus, the fully-plated and generally marine or anadromous form has a more northerly distribution than does the lower-plated leiurus form. The Gasterosteus populations of northern Scandinavia, the Baltic area, and Alaska are mostly trachurus (Munzing, 1963; McPKail, 1969). More southerly populations are often mixed (trachurus, semiarmatus, and leiurus). Populations in the extreme south are generally of the leiurus form, and are usually confined to freshwater habitats (Munzing, 1963; Miller and Hubbs, 1969). Munzing (1963) suggested two alternative explanations for this distribution. The f i r s t is that leiurus represents a widespread 'rsouthern glacial r e l i c t that was cut off from the northern trachurus form during the last glaciation. Following glacial recession, trachurus extended i t s range to the south and leiurus dispersed to the north. Therefore, areas of sympatry, in intermediate latitudes, were colonized twice, and interbreeding between leiurus and trachurus has produced mixed populations 113 in the intermediate latitudes. An alternative explanation i s that the two forms, are different morphs (of a polymorph) of a widespread species. The leiurus morph, when exposed to certain environmental circumstances, such as fresh water (and perhaps higher temperatures) has a selective advantage that enables i t to repeatedly colonize freshwater lakes and streams in the absence of trachurus. The local distribution of Gasterosteus supports the latter interpretation. On the Pacific coast Gasterosteus i s found in almost a l l streams and rivers that drain into the ocean, as well as many coastal lakes of low elevation. The fish are widespread on the coastal islands including Vancouver Island and the Queen Charlotte Islands. Table II shows only leiurus; others contain mixed populations with varying pro-portions of semiarmatus. The fi s h fauna of the British Columbia coastal islands i s characterized by a complete absence of so-called primary freshwater fishes (Miller, 1958). A l l species in these areas had to colonize freshwater habitats via the sea. Pure leiurus populations are unknown from marine habitats. On the other hand, some marine populations are characterized by a range of phenotypes that include forms which can be categorized as leiurus on the basis of plate number and pattern. This strongly suggests that the isolated freshwater leiurus populations are descendents of an-cestral mixed populations. Further, i t follows that different leiurus populations, must represent convergent evolution - a higher plated, mixed population repeatedly giving rise to low-plated leiurus populations. 114 However, with geographical isolation between leiurus populations in d i f f e r -ent freshwater habitats, divergence is also a common phenomenon. Unique leiurus populations, some of which may well be distinct species, have been described by a number of authors. McPhail (1969) des-cribed a stickleback in the Chehalis River system, Washington State, that develops black nuptial colouration and appears reproductively isolated from a sympatric form which displays the more usual (red) nuptial colours. Semler (1971) described a polymorphism inmptial colouration of stickle-backs in Lake Wapato, Washington State, that seems unlike that of the ad-jacent but geographically isolated populations. Moodie (1972) describes an unusual stickleback in the Queen Charlotte Islands that is apparently reproductively isolated from a sympatric, and more typical leiurus form. McPhail (pers. comm.) has found stickleback populations in lakes on Texada Island, B.C. which show a marked reduction in pelvic girdle form-ation. Miller and Hubbs (1969), Hagen and Gilbertson (1972), and Moodie (1972) have surveyed Gasterosteus populations from various regions of the Pacific Coast. Variation within and among leiurus populations is striking. In view of the variation exhibited by Gasterosteus i t i s not surprising that different interpretations arise regarding the taxonomic significance of the forms. Hagen (1967), and Hagen and McPhail (1970) argue strongly that leiurus and trachurus are different, reproductively isolated species. Miller and Hubbs (1969) maintain that the subspecific category i s adequate taxonomic recognition and state that introgression of characters, particularly plates, precludes any interpretation of 115 reproductive isolation between leiurus and trachurus (which they refer to as G.a. microcephalus and G.a. aculeatus respectively), A third, intermediate position is expressed here. Some popul-ations of Gasterosteus represent distinct species according to morpho-logical and other biological c r i t e r i a . Other populations represent various degrees of genetic and morphological specialization but do not warrant recognition as species. I believe that the various leiurus populations are not identical. They are either distinct species, or they are potential incipient species, although not a l l have evolved specific status with distinct isolating mechanisms. Therefore, a simple morphological criterion of species in Gasterosteus is insufficient to explain a complex array of sibling species, each with considerable intra-specific variation. Evidence for speciation in the L i t t l e Campbell River In an analysis of the sticklebacks of the L i t t l e Campbell River, using the c r i t e r i a of reproductive isolating mechanism established by Mayr (1963) and Dobzhansky (1951), Hagen (1967) concluded that two distinct species are present. The trachurus and leiurus forms (Hagen advocates G. trachurus and G. aculeatus) have different distributions. Trachurus is found close to the sea; leiurus occupies the upper reaches and headwaters. Hagen (1967) attributed reproductive isolation to differences in spawning times (although considerable overlap occurs) and to habitat differences. Hay (1969), in a re-examination of ethological isolating mechanisms found 116 significant differences between the courtship of leiurus and trachurus, and a preference of females for conspecifics. No one, howeyer, has been able to demonstrate any selective disadvantage to the intermediate, semiarmatus phenotype. Some of the results of the present study reinforce the hypothesis of reproductive isolation between leiurus and trachurus in the L i t t l e Campbell River. Semiarmatus, or phenotype I, defined by the presence of plate I, i s restricted to the lower reaches of the river. It does not penetrate beyond the hybrid zone as defined by Hagen. Every cross made using one or both parents with plate I produced some progeny with plate I. If the I phenotype were moving upstream, and i f introgression of plates i s occurring, as Miller and Hubbs (1969) suggest, then some of the upstream residents should exhibit plate I. Consequently, the absence of plate I from the upper reaches strongly indicates a barrier to gene flow from the hybrid zone (region of a mixed population) to the leiurus habitat. Conversly, there i s strong evidence to suggest that gene flow does not occur in the opposite direction. R. Jones (pers. comm.) has examined the muscle myogen patterns of leiurus and trachurus through electrophoretic techniques. He confirms Hagen's (1967) finding that trachurus seems to be a homogeneous population producing only a slow -migrating pattern. Leiurus produces two patterns: a slow-migrating pattern as in trachurus, and a fast-migrating pattern. Analysis of crosses suggests that these patterns are simply inherited: a single locus, 2 a l l e l e system. If gene flow between leiurus and trachurus occurs, then 117 some trachurus that exhibit the fast pattern should be found. Analysis of trachurus patterns indicate that this is not the case. The evidence suggests that gene exchange does not occur between the leiurus and trachurus populations. This, combined with the evidence of reproductive isolating mechanisms of Hagen (1967), and Hay (1969), is sufficient to regard the two types of sticklebacks in the L i t t l e Campbell River as distinct species. Morphological analysis of other populations indicates that some appear to be purely leiurus (on the basis of the presence or absence of plate I) whereas others can only be regarded as mixed populations contain-ing leiurus, and semiarmatus but no trachurus. (i.e., the Hybrid Pond, Ladner Ditches, and lakes on Vancouver Island). Here phenotypic integrity of semiarmatus i s preserved i n the absence of trachurus. Under these circumstances semiarmatus, the so-called hybrid form, does not result from leiurus-trachurus crosses. It i s not produced by crosses within the leiurus forms - as established in this and other studies. Therefore, i t must replicate and preserve i t s e l f or become extinct. Elimination of the semiarmatus form has probably occurred in many circumstances but i t s persistence in numerous populations precludes the interpretation of semiarmatus in these circumstances as a hybrid. Rather, i t is best re-garded as a distinct morph - although i t is most l i k e l y a heterozygous morph within a Gasterosteus complex consisting of several morphs: a f u l l y plated (trachurus), a partially plated (semiarmatus), and a low plated (leiurus) form. Such a population could easily give rise to homogeneous, populations of leiurus by gradual elimination, through selection, of ful l y 118 plated and partially plated forms. Gradual change in a phenotypic frequency can best be regarded as a transient polymorphism * a phenomenon known in a number of other species (Ford, 1965). If selection that leads to the elimination of the higher-plated morphs continued, then a possible consequence i s the establishment of reproductive isolation between the morphs. Therefore, I interpret this as a polymorphism that in many instances is the basis of incipient species. If speciation has occurred and continues to occur among Gaster- osteus populations, then i t i s l i k e l y that future investigations w i l l find many different species. Ultimately, every stream and lake where sticklebacks reside could produce a distinct species. This i s not incon-ceivable. It has happened elsewhere. For instance, Darlington (1957) estimates the cypriniform fauna of the Amazon contains over 2000 species, a l l derived from a limited number of ancestors. Species flocks in the Great African lakes (Lowe-McConnell, 1969) and Lake Lanao (Myers, 1960) demonstrate that explosive radiation has occurred i n different circum-stances and in different groups of fish. However, the reasons for the establishment and maintenance of diversity (as seen in the tropics) are not firmly established. Among the more credible arguments i s the theory that tropical situations are more stable in time than temperate environ-ments and diversification of the habitat (niche) allows the development of a richer fauna (Hutchinson, 1959). Temperate areas, intermittently affected by glaciation are thought to be less stable. Consequently, diversity i s less pronounced 119 and temperate organisms exhibit fewer specializations than tropical forms. If s t a b i l i t y were Imposed on temperate forms, i t would seem reasonable that they would undergo specialization and speciation. For instance, temperate marine habitats, and even temperate freshwater habitats such as Lake Baikal, that have not been subject to the same effects of gla- , ciation as most freshwater and terrestrial habitats, show considerable d i -versity of fish fauna. Gasterosteus with i t s short generation time and apparent genetic variability (as expressed in the phenotypic variation) may be an example of an organism undergoing rapid evolution and special-ization under comparatively stable temperate conditions since the last glaciation. The many isolated river systems, a l l directly emptying into the sea, provide many different habitats permitting regional specializa-^ tion and independent evolutionary pathways. Adaptive significance of lateral plates Discussions of geographical variation and zoogeography often tend to ignore the adaptive value of the characters under consideration. Although plate patterns and plate numbers in sticklebacks may be i n -dicative of gross evolutionary trends, they must have adaptive value in themselves, or presumably they would be lost. Loss of plates i s , in fact, known in several populations, and reduction of plates i s known for many (Moodie, 1972). No one has yet offered a satisfactory explana-tion of the function of plates although they are often found associated 120 with, other characteristics that have reasonably clear adaptive value. For instance, Heuts (1947) demonstrated differences between plate morphs and salinity tolerance. Moodie (1972) demonstrated that fish with seven plates are at a selective advantage under cutthroat trout predation. Plate numbers and plate patterns are correlated with other morphological characters. High plate numbers are often associated with high g i l l raker numbers (a trophic adaptation), high vertebrate numbers, a terete body shape, and particular electrophoretic patterns (Hagen, 1967). The present study presents evidence that the plate pattern i s also associated with different fecundities and different breeding habitats. This association between plates and other characters need not suggest that plates are of l i t t l e consequence by themselves. Rather, i t may be that selection has operated to link the l o c i responsible for other characters with those responsible for plates. There is some association between the plate pattern and the electrophoretic patterns of muscle myogens described earlier. Recent evidence indicates that the muscle myogens are really different forms of the enzyme creatine kinase (R. Jones, pers. comm). Within leiurus, the G and the H phenotypes have three patterns (i.e., homozygous fast, homozygous slow, and heterozygous fast-slow patterns). However, the G phenotype is predominately the fast pattern (unique to leiurus) and the H phenotype i s associated with the slow patterns. The heterozygous bands occur in high frequency in both the G and H phenotypes. 121 Genetic homeostasis, ecological strategies, and phenotypic variation Whatever the function of the plates, i t i s clear that their inheritance is not simple and that i t demonstrates, as Lerner (1958) calls i t , genetic homeostasis. Crosses indicate that both the G and H phenotypes have the potential to generate each other. Consequently, the genetic system controlling plates i s probably resistant to changes through selection. For instance, i f selection i n the Waterfowl Refuge allowed only phenotype G to survive and breed during one generation, they would s t i l l produce 10 percent phenotype H among their progeny. Alternatively, i f selection allowed only phenotype H to breed, they would result in 50 percent of G offspring. The genetic structure of other populations is obviously di f -ferent. In some pure leiurus populations, phenotype H makes up a majority of the individuals. This circumstance could not arise on the basis of the estimates of inheritance established for the Waterfowl Refuge. The genotypes of G and H must be different in other populations. This does not necessarily imply that the mechanism of inheritance is different; rather, that the frequency of alleles that determine plates varies among populations. Genetic homeostasis i s not a characteristic of the I phenotype in the L i t t l e Campbell River. This phenotype i s replicated in only f i f t y percent or less of i t s progeny when crossed to a leiurus phenotype. 122 Crosses among the I phenotype (semiarmatus) generate a higher frequency of phenotype I among the progeny but i t seems clear that the frequency of I phenotypes among young of mixed populations (leiurus and semiar- matus) must be lower than frequency of I among the parents. Main-tenance of this morph is dependent upon bhher factors such as a se-lective advantage under predation, or a reproductive advantage in terms of the fecundity of females, or a disproportionate participation of I phenotypes among the breeding population compared with that among the non-breeding population. In the Hybrid Pond I appears to be main-tained by some selective advantage during the non-reproductive season. This permits phenotype I to increase in frequency, so that the frequency is highest during the breeding season (even though the frequency of I is not dissimilar between the breeders and non-breeders). Therefore, In the absence of selective forces that maintain the I phenotype (by increasing i t s frequency within generations), i t should rapidly dis-appear. In fact, the rate of extinction under these circumstances can be calculated i f population size can be estimated. A population estimate of the Hybrid Pond was made during the summer of 1968. On the basis of mark and recapture experiments, I estimated that the pond contained about 100,000 fish. Marking involved clipping the f i r s t dorsal spine. This i s a relatively innocuous procedure for larger f i s h , but the hand-ling may be injurious to smaller ones. Therefore, the population estimate i s crude, but sufficient for the purposes of i l l u s t r a t i o n . Only about one-third of the fish appear to take part in reproductive activities 123 C24% breeding i n 19.71 and 32% i n 19.72).. This reduces the e f f e c t i v e population number to about 30,000. The proportion of I among the breeding population i s about one t h i r d (a high estimate), so that about 10,000 f i s h with phenotype I contribute to the subsequent generation. If the proportion of progeny r e s u l t i n g from crosses where one or both parents i s phenotype I i s 60% (again a high estimate because some crosses w i l l be I x I crosses and probably y i e l d more than 50% I as described f o r I-leiurus crosses) then the proportion of I among the subsequent generation w i l l be (10,000/30,000)•(0.6) = 0.20 or 20 percent. I f the proportions of phenotype I breeding i n subsequent generations were proportional to i t s representation i n the population, and i f no other reproductive advantages were associated with t h i s phenotype, i t would be completely eliminated i n less than 20 generations ( i . e . 10,000 x ( 0 . 6 ) 2 0 = 0.366). Although the estimate i s crude, i t i s f a i r l y conservative. As the frequency of I decreases with time, the p r o b a b i l i t y of I x I matings also decreases, and the expected contribution of I o f f s p r i n g to the next generation would probably be less than the 60% estimate here. This i l l u s t r a t i o n demonstrates that rapid elimination of the I phenotype i s possible i f s e l e c t i v e pressures changed. If the supposition i s extend-ed to suggest that the I phenotype (or something s i m i l a r ) was the most l i k e l y progenitor of pure l e i u r u s populations, then a rapid c o l o n i z a t i o n and subsequent elimination of I phenotype as l e i u r u s phenotypes became established i s not d i f f i c u l t to imagine. 124 The selective pressures maintaining the G and H phenotypes in the Waterfowl Refuge area must be different. The proportions of each are generally constant within generations. Therefore, selective pres-sures l i k e predation, operating between breeding seasons, seem relatively unimportant as mechanisms that change or maintain the polymorphism. However, the phenotypic frequencies do change between generations and the phenotypic frequenceis of breeding fish are significantly different from the non-breeding fish. Phenotype H i s more abundant among the breeders than among the non-breeders. The apparent reproductive advantage of this phenotype may be important in maintaining phenotype H i n the popu-lation. It is tempting to interpret the plate morphs as r and K strate-gists (Cody, 1966). The I phenotype, and to a lesser extent, the H phenotype, correspond to r strategists - forms with good powers of dis-persal, high reproductive capacities (high intrinsic rates of increase), and a tendency to exploit ephemeral habitats. The r strategist allocates more energy to reproduction than the K strategist. In this regard, both the H and I phenotypes have higher fecundities than the G pheno-type. K strategists, adapted to more stable environments, generally have low powers of dispersal. Hagen's (1967) dispersal experiments clearly show the sedentary nature of leiurus. Parallels between r and K strategists and the phenotypic morphs described here can be extended to general geographical distributions. The r strategist (I phenotype) is often found in ephemeral- habitats. 125 The Ladner Ditches did not exist before the turn of the century, when they were dug for irrigation purposes. The Hybrid Pond i s relatiyely new; i t was excavated about 20 years ago to water livestock. Coloniza--tLon of such recent habitats is best accomplished by r strategists and indeed the frequency of the I phenotype is high in such areas. Presuma-bly, as these environments stabilize with time, the K strategists in the form of phenotype G, w i l l prevail. This relationship between plate pattern (or plate number) and environmental strategies in terms of r and K strategists i s conjectural. It i s apparent, however, that the phenotypes are associated with different selective pressures, and that they, have evolved differences in their response to selection - especially to .-factors associated with reproduction. Future research, concerned with the function and adaptive significance of plates, would be most useful in resolving these problems. 126 CONCLUSIONS Recording the presence or absence of certain individual plates is a convenient and r e a l i s t i c method of phenotypic classification for Gasterosteus. When analysed this way, Gasterosteus appears to exhibit a complex polymorphism but one can unequivocally distinguish between semiarmatus and leiurus. Phenotypic frequencies change with time, and two kinds of change, can be distinguished: (1) changes within generations and (2) changes between generations. These changes are not due to error i n sampling, error in phenotypic assessment, or differential dispersal. Semiarmatus increases in frequency within generations. This i s probably because predation, by size-selective predators l i k e water scorpions and cutthroat trout, i s more intense on the smaller leiurus phenotypes. There are many potential causes of changes in frequency between generations but the most important are associated with (1) the mechanisms of plate (or phenotype) inheritance, and (2) an apparent reproductive advantage of one leiurus phenotype which has a proportionately higher frequency among breeders than among the population at large. The results of this study reinforce earlier suggestions that leiurus and trachurus are reproductively isolated in the L i t t l e Campbell River. Plate I, which is diagnostic of semiarmatus, is never found in 127 leiurus habitats. Every semiarmatus cross produced some semiarmatus progeny, but crosses among leiurus never produced semiarmatus progeny. Therefore, the genetic material which controls plate I is excluded from the leiurus phenotypes that live above the 'hybrid zone'. Previous suggestions that the semiarmatus phenotype represents introgression or hybridization between leiurus and trachurus are inadequate to explain the frequent abundance of semiarmatus in some populations -especially i n mixed populations of leiurus and semiarmatus where trachurus i s absent. In these circumstances, semiarmatus must replicate i t s e l f . However, many of the progeny of semiarmatus crosses are leiurus phenotypes, and unless semiarmatus increases i t s frequency between breeding seasons, i t w i l l go extinct. Recurrent extinction of semiarmatus, from ancestral mixed populations, is a reasonable explanation for the origin of pure leiurus population. 128 ACKNOWLEDGEMENTS I wish to thank my supervisor, Dr. J.D. McPhail, for his encouragement and criticisms. Drs. D. Chitty, D. Holm, and G. Scudder offered many useful suggestions for the preparation of this manuscript. I have benefitted from discussions with Drs. R. Harger and C. Wehrhahn, Institute of Animal Resource Ecology, University of B.C.; Dr. D. Hagen, Department of Biology, University of N.B.; Dr. T.Garside, Department of Biology, Dalhousie University. I am grateful to Dr. L. Licht, Messrs. R. Jones and J. Maclean for helping me i n the f i e l d . I am especially grateful to my wife, Elly, for her encourage-ment and patience during the study. 129 REFERENCES Cain, A.J., and Sheppard, P.M. 1954. Natural selection in Cepaea. Genetics 39:89-116. Cody, M.L. 1966. A general theory of clutch size. Evolution 20:174-184. Darlington, P.J. 1957. Zoogeography. Wiley. New York. 675 p. Dobzhansky, T. 1951. Genetics and the origin of species. 3rd ed. Columbia Univ. Press. New York. 364 p. Ford, E.B. 1965. Genetic polymorphism. Mass. Inst. Tech. Press. Cambridge. 101 p. Ford, E.B. 1971. Ecological genetics. Chapman and Hall. London 410 p. Falconer, D.S. 1960. Introduction to quantitative genetics. Ronald Press. New York. 365 p. Greenbank, A.J., and Nelson, P.R. 1959. Life history of the threespine stickleback Gasterosteus aculeatus Linnaeus in Karluk Lake and Bare Lake, Kodiak Island, Alaska. U.S. Fish Wildl. Serv. Fish. Bull. No. 153, 59:537-559. Hagen, D.W. 1967. Isolating mechanisms in threespine sticklebacks (Gasterosteus). J. Fish Res. Bd. Canada 24:1637-1692. Hagen, D.W., and Gilbertson, L.G. 1972. Geographic variation and environ-mental selection in Gasterosteus aculeatus L. in the Pacific Northwest, America. Evolution, 26:32-51. Hagen, D.W., and McPhail, J.D. 1970. The species problem within Gasterosteus  aculeatus on the Pacific Coast of North America. J. Fish Res. Bd. Canada 27:147-155. Hay, D.E. 1969. Mate selection in threespine sticklebacks (Gasterosteus). M. Sc. thesis. Univ. of British Columbia. 77 p. Heuts, M. J. 1947. Experimental studies on adaptive evolution in Gaster- osteus aculeatus L. Evolution 1:80-102. Hoogland, R., Morris, D., and Tinbergen, N. 1956. The spine of sticklebacks, (Gasterosteus and Pygosteus) as a means of defence against predators (Percan and Esox). Behaviour 10:200-230. Hutchinson, G.E. 1959. Homage to Santa Rosalia or why are there so many kinds of animals? Amer. Naturalist 93:145-159. 130 Kettlewell, H.B.D. 1956. A resume of investigations on the evolution of melanism in the Lepidoptera. Proc. Roy. Soc. (London) 145:297-303. Lerner, I.M. 1958. The genetic basis of selection. Wiley, New York. 298 p. Lindsey, C.C. 1962. Experimental study of meristic variation in a population of threespine sticklebacks, Gasterosteus aculeatus. Canad. J. Zool. 40:271-312. Lowe-McConnell, R.H. 1969. Speciation in tropical freshwater fishes. Bull. Linn. Soc. 1:51-79. Mayr, E. 1963. Animal species and evolution. Harvard Univ. Press. Cambridge. 797 p. McPhail, J.D. 1969. Predation and the evolution of a stickleback (Gasterosteus) J. Fish. Res. Bd. Canada 26:3183-3208. Miller, R.R. 1959. Origin and a f f i n i t i e s of the freshwater fish fauna of Western North America, p. 187-222. In C.L. Hubbs, Zoogeo-graphy. Amer. Assoc. Adv. Sci. Publ. 51. Miller, R.R., and Hubbs, C.L. 1969. Systematics of Gasterosteus aculeatus , with particular reference to intergradation and introgression along the Pacific Coast of North America: a commentary on a recent contribution. Copeia 1969:52-69. Moodie, G.E.E. 1972. Predation, natural selection, and adaptation in an unusual threespine stickleback. Heredity 28:155-167. Mullem, P. van, and Vlugt, J.C. van der. 1964. On the age, growth and migration of the anadromous stickleback Gasterosteus aculeatus L. investigated in mixed populations. Arch. Neerl. Zool. 16:111-139. Munzing, J. 1959. Biologie, Variabilitat and Genetik von Gasterosteus aculeatus L. (Pisces), im Elbegebiet. Int. Rev. Ges. Hydrobiol. 44:317-382. Munzing, J. 1963. The evolution of variation and distributional patterns in European populations of three-spined sticklebacks, Gasterosteus aculeatus. Evolution 17:320-332. Myers, G.S. 1960. The endemic fish fauna of Lake Lanao, and the evolution of higher taxonomic categories. Evolution 14:323-333. 131 Norman, J.R., and Greenwood, P.H. 1963. A history of fishes. Ernest Benn. London. 398 p. Sender, D.E. 1971. Some aspects of adaptation in a polymorphism for breeding colours in the threespine stickleback (Gasterosteus aculeatus). J. Zool. London. 165:291-302. Steel, R.G.D., and Torrie, J.H. 1960. Principles and procedures of st a t i s t i c s . McGraw-Hill, New York. 481 p. 132 Appendix Table I. Analysis of the collections. I a. Collections from the Waterfowl Refuge Generation Mean Collection (year Size Standard Standard Frequency date Age spawned) Number (mm) Deviation Error G H March 15,1969 0+ 1968 94 40.74 4.77 0.49 0.479 0.521 1+ 1967 1 58.00 Aug. 20,1969 0+ 1969 133 23.90 3.58 0.31 0.466 0.534 1+ 1968 64 42.91 3.81 0.48 0.531 0.469 June 1,1970 1+ 1969 78 48.90 2.68 0.30 0.602 0.398 2+ 1968 9 58.40 1.83 0.61 3+ 1967 1 78.00 June 8,1970 1+ 1969 30 45.13 3.61 0.66 0.633 0.367 2+ 1968 2 59.50 July 16,1970 0+ 1970 454 19.80 4.50 0.21 (0.576) (0.424)* Oct. 9,1970 0+ 1970 291 26.42 5.00 0.29 0.622 0.378 1+ 1969 18 50.44 4.10 0.96 Oct. 24,1970 0+ 1970 162 29.12 5.15 0.40 0.586 0.414 1+ 1969 8 49.63 3.78 1.33 Nov. 11,1970 0+ 1970 146 27.47 5.39 0.45 0.658 0.342 1+ 1969 8 52.00 4.04 1.43 Jan. 29,1971 0+ 1970 163 25.10 6.46 0.51 0.577 0.423 1+ 1969 5 54.60 4.22 1.89 Feb. 5,1971 0+ 1970 31 30.84 5.03 0.90 0.548 0.452 1+ 1969 1 55.00 Mar. 13,1971 0+ 1970 33 29.73 5.71 0.99 0.576 0.424 Apr. 14,1971 0+ 1970 81 37.94 6.93 0.77 0.531 0.469 Apr.26,1971 0+ 1970 167 38.92 7.01 0.55 0.587 0.413 1+ 1969 3 58.67 May 4,1971 0+ 1970 223 37.92 6.40 0.43 0.592 0.408 1+ 1969 2 61.00 May 17,1971 0+ 1970 220 39.40 5.46 0.37 0.586 0.414 June 2,1971 1+ 1970 111 40.18 4.62 0.44 0.595 0.405 June 9,1971 1+ 1970 126 40.97 5.00 0.45 0.659 0.341 2+ 1969 1 63.00 July 30,1971 0+ 1971 129 24.04 4.50 0.40 0.721 0.279 1+ 1970 78 46.03 4.25 0.48 0.474 0.526 Sept.21,1971 0+ 1971 315 24.56 4.24 0.24 0.697 0.321 1+ 1970 77 49.36 . 3.96 0.45 0.481 0.519 Nov. 4,1971 0+ 1971 98 29.50 4.50 0.45 0.735 0.265 1+ 1970 5 47.40 5.41 2.42 Dec. 1,1971 0+ 1971 323 29.47 4.30 0.24 0.740 0.260 1+ 1970 4 49.00 4.97 2.48 Feb.24, 1972 0+ 1971 121 32.39 5.42 0.49 0.645 0.355 1+ 1970 1 52.00 Mar. 30,1972 0+ 1971 95 38.17 5.00 0.41 0.684 0.316 1+ 1970 7 58.14 3.67 1.39 Apr. 27,1972 0+ 1971 122 40.97 4.71 0.43 0.705 0.295 1+ 1970 9 57.67 2.92 0.97 May 16,1972 0+ 1971 138 40.78 4.50 0.38 0.667 0.333 1+ 1970 12 57.50 3.68 1.06 May 24,1972 0+ 1972 131 20.15 4.43 0.39 0.526 0.473 1+ 1971 83 43.36 4.05 0.44 0.723 0.277 2+ 1970 4 57.25 2.50 1.25 July 13,1972 0+ 1972 251 22.51 3.29 0.21 0.582 0.418 Aug. 29, 1972 0+ 1972 341 24.99 5.00 0.27 0.613 0.387 1+ 1971 300 48.10 5.10 1.11 Aug. 8, 1973 1+ 1972 172 0.558 0.442 *frequency calculated on subsample of largest f i s h . 133 I b. Collections from the Hybrid Pond Generation Mean Collection date Age (year spawned) Number Size (mm) Standard Deviation Standard Error Frequency G H I July 15,1968 0+ 1968 55 30.20 3.08 0.41 0.218 0.564 0.218 1+ 1967 249 44.40 4.22 0.27 0.120 0.618 0.261 Sept.17,1968 0+ 1968 231 30.38 4.55 0.30 0.182 0.645 0.173 1+ 1967 95 46.55 4.63 0.47 0.221 0.611 0.168 Mar. 17,1969 0+ 1968 163 31.51 7.52 0.59 0.221 0.583 0.196 1+ 1967 3 54.00 May 7, 1969 0+ 1968 286 37.82 8.21 0.49 0.199 0.629 0.171 1+ 1967 4 61.00 1.15 0.58 Aug.20,1969 0+ 1969 318 30.50 4.14 0.23 0.355 0.535 0.110 1+ 1968 17 49.88 3.16 0.77 Jan. 29,1970 0+ 1969 154 33.34 6.98 0.56 0.227 0.565 0.208 1+ 1968 19 48.95 2.01 0.46 May 12,1970 0+ 1969 177 41.40 8.18 0.62 0.237 0.582 0.181 1+ 1968 4 59.75 1.71 0.85 June 10,1970 1+ 1969 118 47.50 5.42 0.50 0.237 0.534 0.229 2+ 1968 2 65.50 Aug. 5,1970 0+ 1970 183 24.63 4.06 0.30 0.224 0.667 0.109 1+ 1969 18 52.56 3.26 0.77 Sept. 8,1970 0+ 1970 227 27.96 5.10 0.34 0.225 0.595 0.181 1+ 1969 4 49.50 5.40 0.27 Oct. 1,1970 0+ 1970 578 25.16 6.48 0.27 0.254 0.614 0.131 1+ 1969 16 46.94 4.20 1.05 Nov.6,1970 0+ 1970 485 22.53 3.26 0.15 0.216 0.707 0.076 Mar. 17,1971 0+ 1970 230 23.80 4.26 0.28 0.222 0.691 0.087 1+ 1969 1 52.00 Apr.14,1971 0+ 1970 294 24.96 5.44 0.32 0.201 0.643 0.156 1+ 1969 5 56.71 3.51 1.57 Apr. 26,1971 0+ 1970 416 25.02 3.94 0.19 0.183 0.618 0.200 1+ 1969 5 52.80 2.95 1.32 May 18, 1971 0+ 1970 341 33.32 7.78 0.42 0.138 0.607 0.255 1+ 1969 5 65.20 8.70 3.89 June 9,1971 1+ 1970 234 40.62 7.15 0.47 0.184 0.564 0.252 2+ 1969 1 66.00 June 22,1971 1+ 1970 663 40.22 5.37 0.21 0.170 0.502 0.327 2+ 1969 1 75.00 July 30,1971 0+ 1971 189 27.67 3.91 0.28 0.302 0.571 0.127 1+ 1970 106 45.03 6.71 0.65 0.850 0.594 0.321 Sept.23,1971 0+ 1971 408 28.32 5.13 0.25 0.294 0.578 0.127 1+ 1970 103 45.03 4.38 0.43 0.136 0.563 0.301 Oct. 7,1971 0+ 1971 200 30.88 5.88 0.42 0.300 0.530 0.170 1+ 1970 55 45.78 3.71 0.50 0.127 0.636 0.236 Nov. 4,1971 0+ 1971 406 27.97 5.07 0.25 0.273 0.512 0.214 1+ 1970 25 45.48 4.30 0.86 Feb. 22,1972 0+ 1971 309 29.30 6.15 0.35 0.265 0.521 0.214 1+ 1970 26 49.62 3.45 0.68 Apr. 15,1972 0+ 1971 429 32.10 6.20 . 0.30 0.245 0.559 0.196 1+ 1970 40 54.00 3.80 0.60 May 16,1972 0+ 1971 151 37.30 6.54 0.53 0.338 0.517 0.146 1+ 1970 15 54.87 6.12 0.64 June 22,1972 0+ 1972 6 22.67 2.16 0.88 1+ 1971 108 40.97 3.78 0.12 0.222 0.500 0.278 July 13,1972 0+ 1972 38 24.29 3.20 0.52 1+ 1971 202 39.00 4.06 0.29 0.173 0.465 0.361 Appendix Table I I . Results of the crosses. II a. G x G crosses - Waterfowl Refuge number number number number frequency Cross of that number of G of H Total of G number eggs hatch frequency of fry frequency progeny progeny progeny 8 Q<-> 110 107 0.970 54 0.491 52 2 54 0.963 10A 62 49 0.790 34 0.548 29 5 34 0.853 16 A 55 48 0.870 20 0.364 16 0 16 1.000 23A - 34 0 34 1.000 24A , -• • 38 0 38 1.000 25A 40 19 0.475 12 0.300 10 2 12 0.833 38 73 69 0.945 22 0.301 22 0 22 1.000 40 97 97 1.000 41 0.423 38 3 41 0.927 47 118 117 0.001 52 0.441 33 19 52 0.635 50B 40 40 1.000 27 0.675 27 0 27 1.000 53A 54 35 0.648 16 0.296 13 3 16 0.812 53B 46 38 0.826 26 0.565 26 0 26 1.000 84 49 47 0.959 40 0.816 17 6 23 0.739 85 44 39 0.886 28 0.636 18 7 25 0.720 102 107 103 0.963 33 0.308 33 0 33 1.000 29A , • 17 5 22 0.773 69 - — 31 0 31 1.000 217 107 51 0.477 27 0.252 27 0 27 1.000 218 77 70 0.909 39 0.506 27 12 39 0.692 220 27 19 0.704 8 0.296 8 0 8 1.000 221 43 39 0.907 29 0.674 27 2 29 0.931 222 48 48 1.000 36 0.750 36 0 .36 1.000 225 35 35 1.000 31 0.886 26 5 31 0.839 228 63 55 0.873 11 0.175 8 3 11 0.727 230 40 38 0.950 19 0.475 17 2 19 0.895 239 44 19 0.432 8 0.182 8 0 8 1.000 241 90 63 0.700 12 0.133 12 0 12 1.000 285 133 132 0.993 104 0.782 31 0 31 1.000 TOTAL 1602 1377 729 681 76 757 Appendix Table I l a continued. Hatching success Frequency of survival to fry stage Frequency of phenotyp G in progeny estimated from estimated from totals average frequencies 0.859 0.844 0.455 0.470 0.900 0.905 II b. F2 crosses • -number number number Cross of that number of G number eegs hatch frequency of fry frequency progeny 208 & ° 50 24 0.480 12 0.240 5 264 112 56 0.500 4 0.036 1 280 69 55 0.797 5 0.072 3 286 81 44 0.543 5 0.062 2 287 78 69 0.885 50 0.641 44 296 68 48 0.705 21 0.309 8 305 57 49 0.860 6 0.105 5 263 124 113 0.911 5 0.040 3 number of H progeny Total frequency of G progeny TOTAL 639 458 108 71 7 3 2 3 6 13 1 2 37 12 4 5 5 50 21 6 5 108 0.416 0.250 0.600 0.400 0.880 0.381 0.833 0.600 estimated from estimated from totals average frequencies Hatching success 0.717 0.710 Frequency of survival to fry stage 0.169 0.188 Frequency of phenotype G in progeny 0.657 0.545 as II c. Go* x H? crosses - Waterfowl Refuge number number number number frequenc Cross of that number of G of H Total of G number eggs hatch frequency of fry frequency progeny progeny progeny 22 108 102 0.944 88 0.815 60 28 88 0.680 44 67 64 0.955 11 0.164 10 1 11 0.901 68 67 66 0.985 9 0.134 6 3 9 0.667 2 127 107 0.842 70 0.551 47 23 70 0.671 3 192 143 0.745 57 0.297 13 44 57 0.228 13B 55 51 0.927 32 0.582 20 12 32 0.625 14B 47 44 0.936 4 0.085 3 1 4 0.750 31A - 19 2 21 0.905 41 106 96 0.906 19 0.179 5 14 19 0.263 52B 46 45 0.978 9 0.196 6 3 9 0.667 224 57 57 1.000 37 0.649 37 0 37 1.000 240 89 43 0.483 2 0.022 1 1 2 0.500 223A 53 52 0.981 4 0.075 3 1 4 0.750 244 96 79 0.823 37 0.385 22 15 37 0.595 253 110 99 0.900 22 0.200 10 2 12 0.833 262 119 89 0.748 78 0.655 69 9 78 0.885 Totals 1339 1137 479 331 159 490 estimated from estimated from totals average frequencies Hatching success 0.849 0.876 Frequency of survival to fry stage 0.385 0.333 Frequency of phenotype G to progeny 0.675 0.682 ^ II d. Go" x H ? crosses - Ladner and Hybrid Pond numb er number number number frequency Cross of" that number of G of H of G number eggs hatch frequency of fry frequency progeny progeny Total progeny 89 103 33 0.320 15 0.146 14 1 15 0.933 97 121 112 0.926 51 0.421 0 22 22 0.000 99 118 115 0.975 37 0.313 37 0 37 1.000 70 77 73 0.948 49 0.636 47 2 49 0.959 71 132 128 0.970 22 0.167 18 4 22 0.818 Totals 551 461 174 116 29 145 estimated from estimated from totals average frequencies Hatching success 0.837 0.828 Frequency of survival 0.316 0.337 to hatching Frequency of phenotype G in progeny 0.800 0.742 II e. G? x Ho" crosses - Waterfowl Refuge number number number number frequency Gross of that number of G of H of G number eggs hatch frequency of fry frequency progeny progeny Total progeny 10B 57 48 0.842 2 0.035 1 1 2 0.500 9 140 135 0.964 12 0.086 11 1 12 0.917 24B 12 . 2 14 0.857 25B 31 26 0.839 4 0.129 4 0 4.. 1.000 23B 35 2 37 0.946 29B 12 19 31 0.387 202 90 90 1.000 6 0.067 3 3 6 0.500 34 105 90 0.857 9 0.086 8 1 9 0.888 210 51 48 0.941 27 0.523 14 13 27 0.518 211 70 62 0.886 52 0.743 49 3 52 0.942 212 93 80 0.860 25 0.269 17 8 25 0.680 215 48 43 0.896 10 0.208 4 6 10 0.400 232 9 3 12 0.750 236 110 75 0.682 10 0.090 6 4 10 0.600 269 84 78 0.918 43 0.506 4 3 7 .0.571 273 83 80 0.964 51 0.614 16 35 51 0.314 Totals 963 855 251 205 104 309 estimated from estimated from totals average frequencies Hatching success Frequency of survival to hatching Frequency of phenotype G to progeny 0.888 0.293 0.663 0.887 0.279 0.673 I l f . G? x Ho" crosses - Ladner and Hybrid Pond. number number '/ number number frequency Cross of that number of G of H of G number eggs hatch frequency of fry frequency progeny progeny Total progeny 95 158 79 0.500 8 0.051 3 5 8 0.375 256 133 122 0.917 53 0.398 40 13 53 0.930 257 81 75 0.926 23 0.284 9 14 23 0.391 259 76 50 0.658 3 0.041 0 3 3 1.000 Totals 448 326 87 • 52 35 87 estimated from estimated from totals average frequencies Hatching success 0.727 0.750 Frequency of survival to hatching 0.194 0.193 Frequency of phenotype G in progeny 0.598 0.674 II g. H x H crosses - Waterfowl Refuge number number number number f requen< Cross of that number of G of H of G number eggs hatch frequency of fry frequency progeny progeny Total progeny 11B 54 37 0.685 9 0.167 8 1 9 0.889 13A 69 60 0.870 4 0.058 1 3 4 0.250 14A 46 40 0.870 23 0.500 18 5 23 0.783 31B 13 3 16 0.812 1 189 151 0.800 74 0.397 69 6 75 0.920 4 95 85 0.895 49 0.516 28 21 49 0.571 5 88 87 0.989 78 0.886 0 78 78 0.000 26 110 101 0.918 60 0.545 36 24 60 0.600 15 70 56 0.800 36 0.514 29 7 36 0.806 18 137 102 0.744 7 0.051 2 5 7 0.286 19 106 106 1.000 22 0.207 11 11 22 0.500 53B 39 25 0.641 16 0.410 9 7 16 0.563 77 78 67 0.859 38 0.487 32 6 38 0.842 237 113 67 0.592 4 0.035 0 4 4 0.000 312 163 52 0.319 48 0.294 6 42 48 0.125 315 108 50 0.463 4 0.037 0 4 4 0.000 235 113 55 0.487 8 0.070 4 4 8 0.500 272 66 48 0.727 39 0.591 27 12 39 0.692 275 164 152 0.927 83 0.506 29 54 83 0.349 233 52 44 0.846 17 0.327 6 11 17 0.353 Totals 1860 1385 620 328 308 636 estimated from estimated from totals average frequencies Hatching success 0.745 0.759 Frequency of survival to fry stage 0.333 0.366 Frequency of phenotype G in progeny 0.516 0.492 II h. H x H crosses - Ladner and Hybrid Pond Cross number 92 104 94 96 105 270 271 numb er of _ _S£S 189 106 140 122 178 89 79 number that number number of G number of H frequency of G hatch frequency of fry frequency progeny pro geny Total progeny 179 0.947 22 0.116 11 11 22 0.500 102 0.962 13 0.123 2 11 13 0.154 68 0.486 36 0.257 1 35 36 0.028 116 0.951 61 0.500 20 41 61 0.328 128 0.719 ' 10 0.056 6 4 10 0.600 63 0.708 27 0.303 6 21 27 0.222 79 1.000 9 0.114 3 6 9 0.333 Totals 903 735 178 49 129 178 estimated from estimated from totals ' ' average frequencies Hatching success 0.814 0.825 Frequency of survival to hatching 0.197 0.210 Frequency of phenotype G in progeny 0.275 0.309 H i . G x I crosses. number number number number number frequency Cross of of number of \G of H of I of I number eggs hatch frequency of fry frequency progeny frequency progeny progeny progeny 81 132 105 0.795 24 0.182 3 0.124 10 11 0.458 90 83 82 0.988 21 0.253 ' 1 0.048 12 8 0.381 98 100 47 0.470 15 0.150 6 0.040 2 7. 0.467 100 99 98 0.989 27 0.273 9 0.333 5 13 0.481 323B 54 51 0.944 34 0.630 2 0.058 15 17 0.500 325A 60 55 0.917 37 0.617 29 0.780 5 3 0.081 306B 26 23 0.885 9 0.346 3 0.333 1 5 0.555 327B 75 32 0.427 8 0.107 2 0.250 4 2 0.250 Totals 629 493 175 55 54 66 estimated from estimated from totals average frequency 0.783 0.802 0.278 0.320 0.314 0.245 0.623 0.603 0.377 0.397 OJ Hatching success Frequency of survival to the fry stage Frequency of phenotype G in progeny Frequency of leiurus (G+H) in progeny Frequency of I in progeny II j . H x I crosses. number number number number number number Cross of that of of G of H of I number eggs hatch frequency fry frequency progeny frequency progeny progeny frequency 80 87 56 0.645 51 0.586 13 0.255 14 24 0.470 83A 110 56 0.509 33 0.300 11 0.333 5 17 0.515 107 169 128 0.753 15 0.084 0 - 7 8 0.533 110 129 41 0.318 18 0.140 4 0.222 5 9 0.500 112 76 35 0.461 9 0.118 0 - 4 5 0.555 113 122 75 0.615 23 0.188 1 0.043 16 6 0.261 318 94 90 0.957 60 0.638 5 0.083 19 36 0.600 319 108 41 0.380 34 0.315 0 - 22 12 0.353 Totals 895 522 243 34 92 117 estimated from estimated from totals average : Hatching success 0.583 0.580 Frequency of survival to fry stage 0.272 0.273 Frequency of phenotype G in progeny 0.140 0.117 Frequency of leiurus (G+H) in progeny 0.519 0.527 Frequency of I in progeny 0.481 0.473 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0099985/manifest

Comment

Related Items