Behavioral Genetic Characterization of Hunting in Domestic Dogs, Canis Familiaris

BEHAVIORAL GENETIC CHARACTERIZATION OF HUNTING IN DOMESTIC DOGS, CANIS FAMILIARIS

Budhaditya Chowdhury

A Dissertation Submitted to the Graduate College of Bowling Green State University in partial fulfillment of The requirements for the degree of

DOCTOR OF PHILOSOPHY

December 2011

Committee:

Robert Huber, Advisor

Howard Casey Cromwell

Graduate Faculty Representative

Moira van Staaden

Paul F. Morris

Verner P. Bingman

ABSTRACT

Robert Huber and Moira van Staaden, co-Advisors

Humans have exerted strong selection pressures for the behavioral attributes which make dogs prized hunting companions, viz., superior sensory capabilities, social cognition, and tenacious pursuit of prey. With behavioral traits maintained in stable breed lines, and a unique genome organization, dogs are ideally suited for a genetic dissection of complex behavioral phenotypes. Despite studies employing substantial sampling regimes the promise of the canine model has yet to be realized, largely because of the dearth of robust quantitative behavioral metrics. Here we show that refined dissection of a complex behavior into component parts is instrumental in identifying genomic signatures associated with these elements. Using 11 spatially explicit measures we demonstrate the existence of four principal dispositions describing

Search Eagerness, Linear Running, Handler Reliance, and Zigzagged Searching. A

Single Nucleodide Polymorphic marker - based association study reveals that Search

Eagerness significantly associates with genomic regions on chromosomes 1 and 27

(Chromosome wide significance, p [100,000 permutations] < 0.05). Adjacent are sites for brain specific Demethylase- (MBD2) and GABA transporter genes (BGT1, GAT2,

GAT3) coding for transcriptional repressor proteins and regulating levels of extracellular neurotransmitters respectively. We anticipate that the wider application of such objective quantitative frameworks will improve the phenotypic characterization that remains the bottleneck in neurobehavioral genetic research.

iii

“Tell me this, Glaucon. I see that you have

in your house hunting dogs and a number of

pedigreed cocks. Have you ever considered

something about their unions and procreations?

What? he said.

In the first place, I said, among these themselves,

although they are a select breed, do not

some prove better than the rest?”

Plato. Republic, 5:459a

ACKNOWLEDGEMENTS

I am indebted to my advisors Robert Huber and Moira van Staaden for their advice and review of the various stages of this project, and their patience through the years. I am also thankful to Verner P. Bingman, Paul F. Morris, and Howard Casey

Cromwell, my committee members for their invaluable comments and suggestions. I also thank Danika bannasch, Noa Safra, and Carrie Finno for helping with the molecular aspect of the project at University of California, Davis.

I thank Jim Hamer, president of the Buckeye Brittany Club, for his support at the early stages of this project, Al Andrews for showing me the first hunting trials, and subsequent support through the years, Steve Thompson for the wonderful suggestions and free use of his hunting field, Dee Hempfield for her patient backing for the late stages of this project. Al Andrews, Dee Hempfield, Dave Hagemeyer, Steve Thompson,

Jessica Sewald, and Maggie Lauderdale allowing us to observe their dogs – this study would have been impossible otherwise.

This study would not have been possible without the devoted support of my mom, Kanta Chowdhury, and my aunt, Keya Maitra.

Support for this research was provided by the J. P. Scott Center for

Neuroscience, Mind, and behavior, Center for Biomolecular sciences, BGSU, Moira van

Staaden, and Paul F. Morris.

I also thank all the Huber lab and van Staaden lab members for their continuous support and wonderful suggestions.

TABLE OF CONTENTS

Page

CHAPTER I: INTRODUCTION ...... 1

CHAPTER II. QUANTITATIVE BEHAVIORAL ASSESSMENT AIDS

GENETIC DISSECTION OF HUNTING IN DOMESTIC DOGS ...... 6

Methods summary………………………………………………………………… 11

References………………………………………………………………………… 17

CHAPTER III. QUANTITATIVE CHARACTERIZATION OF SPATIALLY-

EXPLICIT HUNTING BEHAVIORS IN DOMESTIC DOGS, Canis familiaris...... 20

Introduction ...... 20

Methods…………………………………………………………………………… 24

Data Analysis ...... 26

Results…………………………………………………………………………….. 29

Discussion………………………………………………………………………… 30

References…………………………………………………………………………. 40

CHAPTER IV. QUANTITATIVE ETHO-GENETIC DISSECTION OF HUNTING

BEHAVIOR IN DOMESTIC DOGS, Canis familiaris...... 44

Introduction ...... 44

Materials and Methods...... 48

Data collection and processing………………………………………………….. 49

Results……………………………………………………………………………… 51

Discussion………………………………………………………………………….. 52

References…………………………………………………………………………. 57

APPENDIX I: SUPPLEMENTARY MATERIAL FOR CHAPTER IV...... 61

APPENDIX II: IACUC APPROVAL ...... 65

vii

LIST OF FIGURES

Page

Chapter II

1 Representative hunt tracks (10 mins) for 3 different dogs ...... 13

2 Individual variation characterized over three hunt trials...... 14

3 SNP association based on the first principal component...... 15

Chapter III

1a Google Earth image showing experimental field...... 34

1b Hunt trials illustrating markedly different spatial patterns...... 34

2 Characterization of objective behavioral metrics ...... 35

3 Canonical centroid plot describing separation of behavioral attributes...... 36

Chapter IV

1a Genome wide association on CFA 1...... 54

1b Genome wide association on CFA 2...... 54

viii

LIST OF TABLES

Page

Chapter II

1 Factor loadings on first four rotated principal components ...... 16

Chapter III

1 Standardized partial regression coefficients from multivariate regression analysis

on training and sex...... 37

2 Summary statistics for three types of dogs examined ...... 38

3 ANOVA table summarizing differences between different types of dogs...... 39

Chapter IV

1 Summary of dogs used in experimental trials...... 55

2 Subgroups of affected and control dogs created on the basis of the four principal

components ...... 56

CHAPTER I: INTRODUCTION

A long history of selection on rarified traits has endowed dogs with the extremes of form and function1,2. It comes as no surprise that the minds shaping the field of evolution, ethology and behaviorism, has always been intrigued by the uniqueness of this system.

Darwin’s argument for evolution by means of natural selection3, and the continuum of emotions that connected all vertebrates included significant mentions of this canid taxon4. Konrad Lorenz pondered over the inherent diversity in morphology and behavior and proposed some interesting ideas regarding domestication5. However, dogs suffered a setback as experimental animals when it was found out that the numerical competence displayed by “clever dogs” was nothing more than a keen eye to follow handler’s gestures. Banished from laboratories as unreliable subjects, dogs were relegated as “stimulus response automatons”6 under the increasing influence of behaviorism. Roughly around the same time canine breed clubs were beginning to flourish which forwarded very specific “standards” for both form and function7. This dual crisis overwhelmed the dog system as fixed standards only worthy of experimentation in behaviorism. Ironically, Pavlov, the classical figure connecting dogs and conditioned reflex paradigms, suggested a level of individual variation and complexity which went unnoticed. Even the cognitive revolution in ethology failed to vindicate dogs from their inglorious past.

On this side of the Atlantic, however, J. P. Scott and John L. Fuller set themselves to the daunting task of deciphering the genetics of highly penetrant social traits in dogs. Offering the first insight into breed specific behavioral variation, their work still remains one of the best examples in canine behavioral genetics8. In 2005 the high 2

quality draft genome sequence of the dog was decoded along with a dense map of

Single Nucleotide Polymorphisms (SNPs) across breeds9. The access to finer molecular tools saw a radical rise in canine genomic studies elucidating molecular mechanisms of morphologies and diseases. With widely available molecular tools, post J. P. Scott era in canine genetics saw an uprise10–15. Although the promise to better understand behavioral modalities with such molecular tools was evident, it is yet to be realized. The reason for this delay is a lack of rigorous quantitative tools, aimed at elucidating innate behaviors.

Most canine behavioral studies have assessed and compared breeds on emotional traits16. However, the subjective categorical characterization on such traits not only raises the question of judgmental validity and truthfulness, the adequacy of such methods to fully represent the variability in dog breeds has also been called into question. It has been debated to what extent dogs relate to anthropocentric emotions and whether these traits carry enough meaning to undertake breed comparison.

This thesis was aimed at solving this Gordian knot by providing objective behavioral assessment tools, which are robust enough to track behaviors to a molecular level. We approach the problem from an ethological standpoint focusing on performance traits. Hunting, in the course of domestication, played a pivotal role in bringing humans and dogs into close association. Thousands of years of artificial selection that followed, endowed hunting dogs with unique behavioral repertoires with high heritability. Components of the original predatory sequence in the wolves are found across hunting dog breeds today, and a strong genetic signature makes these dogs ideally suited for behavioral genetic characterization and breed comparison. Overall,

this dissertation comprises three interrelated studies investigating the efficiency of objective characterization in elucidating molecular understanding of innate behaviors.

The first part is in the form of a short communication submitted to the Journal

Nature which bridges behavioral characterization and molecular dissection in order to better understand the underlying genetic mechanism regulating behavioral function.

The second part focuses on a rigorous characterization of instinctive traits with the goal of inter- as well as intra-breed assessment of behavioral modalities. Hunting behaviors in domestic dogs were assessed using 15 different objective metrics describing spatial search patterns, aerobic capabilities, and social effect of the human handler. Towards that goal we performed principal components analysis, where correlations between dependent variables are converted into a set of underlying concepts – called principal components. Taking a multivariate analysis approach, hunting dogs from two different breeds was compared to non hunting breeds to determine breed defining characteristics. Within breed comparisons were also performed to estimate within breed variation. This manuscript will be submitted to

Animal Behaviour.

The third section aims to determine whether the complex behavior of hunting can be dissected into component parts describing latent dispositions, and to what extent such components could be traced to a molecular level. Using multivariate behavioral tools dogs were assigned to different subsets based on performance in different underlying categories. We then characterized the full genome of 10 dogs using Single

Nucleotide Polymorphism chips (SNP-chips) and searched for sections of the genome

which may provide significant genetic signatures to explain behavioral performance.

This manuscript will be submitted to Behavior Genetics.

References:

1. Sutter, N.B. & Ostrander, E.A. Dog star rising: the canine genetic system. Nat Rev

Genet 5, 900-910 (2004).

2. Karlsson, E.K. & Lindblad-Toh, K. Leader of the pack: gene mapping in dogs and

other model organisms. Nat Rev Genet 9, 713-725 (2008).

3. Darwin, C. The Origin of Species. (Gramercy: 1995).

4. Darwin, C. The Expression of the Emotions in Man and Animals. (Penguin Classics:

2009).

5. Lorenz, K.Z. MAN MEETS DOG. (Pan Books: 1959).

6. Miklosi, A. Dog Behaviour, Evolution, and Cognition. (Oxford University Press, USA:

2008).

7. Club, A.K. The Complete Dog Book: 20th Edition. (Ballantine Books: 2006).

8. Scott, J.P. & Fuller, J.L. Genetics and the Social Behavior of the Dog. (University Of

Chicago Press: 1998).

9. Lindblad-Toh, K. et al. Genome sequence, comparative analysis and haplotype

structure of the domestic dog. Nature 438, 803-819 (2005).

10. Clark, L.A., Wahl, J.M., Rees, C.A. & Murphy, K.E. Retrotransposon insertion in

SILV is responsible for merle patterning of the domestic dog. Proceedings of the

National Academy of Sciences of the United States of America 103, 1376 -1381

(2006).

11. Jones, P. et al. Single-Nucleotide-Polymorphism-Based Association Mapping of Dog

Stereotypes. Genetics 179, 1033 -1044 (2008).

12. Mignot, E. et al. Genetic linkage of autosomal recessive canine narcolepsy with a

mu immunoglobulin heavy-chain switch-like segment. Proc Natl Acad Sci U S A 88,

3475-3478 (1991).

13. Mosher, D.S. et al. A mutation in the myostatin gene increases muscle mass and

enhances racing performance in heterozygote dogs. PLoS Genet. 3, e79 (2007).

14. Parker, H.G. et al. An Expressed Fgf4 Retrogene Is Associated with Breed-Defining

Chondrodysplasia in Domestic Dogs. Science 325, 995 -998 (2009).

15. Pelé, M., Tiret, L., Kessler, J.-L., Blot, S. & Panthier, J.-J. SINE exonic insertion in

the PTPLA gene leads to multiple splicing defects and segregates with the autosomal

recessive centronuclear myopathy in dogs. Hum. Mol. Genet. 14, 1417-1427 (2005).

16. Diederich, C. & Giffroy, J.-M. Behavioural testing in dogs: A review of methodology

in search for standardisation. 97, 51-72 (2006).

CHAPTER II: QUANTITATIVE BEHAVIORAL ASSESSMENT AIDS GENETIC

DISSECTION OF HUNTING IN DOMESTIC DOGS

Budhaditya Chowdhury, Danika Bannasch, Paul Morris, Moira van Staaden, and Robert

Huber

Humans have exerted strong selection pressures for the behavioral attributes which make dogs prized hunting companions1, viz., superior sensory capabilities, social cognition, and tenacious pursuit of prey. With behavioral traits maintained in stable breed lines2,3, and a unique genome organization4, dogs are ideally suited for a genetic dissection of complex behavioral phenotypes5,6. Despite studies employing substantial sampling regimes7 the promise of the canine model has yet to be realized, largely because of the dearth of robust quantitative behavioral metrics8. Here we show that refined dissection of a complex behavior into component parts is instrumental in identifying genomic signatures associated with these elements. Using 11 spatially explicit measures we demonstrate the existence of four principal dispositions describing

Search Eagerness, Linear Running, Handler Reliance, and Zigzagged Searching. SNP marker - based association study reveals that Search Eagerness significantly associates with genomic regions on chromosomes 1 and 27 (Chromosome wide significance, p [100,000 permutations] < 0.05). Adjacent are sites for brain specific Demethylase- (MBD2) and GABA transporter genes (BGT1, GAT2,

GAT3) coding for transcriptional repressor proteins9 and regulating levels of extracellular neurotransmitters10 respectively. We anticipate that the wider application of

such objective quantitative frameworks will improve the phenotypic characterization that remains the bottleneck in neurobehavioral genetic research11.

Exhibiting the greatest extremes of form and function, domestic dogs are rapidly becoming the scientists’ pet model to explore genetic – phenotypic interactions12. To the delight of the molecular biologist, thousands of years of selection on rarified traits are represented today by over 400 closed breeding populations2, with unique patterns of morphology and behavior13. Making gene hunting even more feasible are two bottleneck events that shaped the dog genome, one at domestication and one at breed creation.

However, since the sequencing of the complete dog genome in 20054, the commitment to decipher “man’s best friend” has been largely met at the “form” level14–19, with less success in behavioral functionality20.

Similar to the issues imposed by the discipline of behavioral analysis21, the challenges in studying canine behavior genetics are twofold - the quantification of dog behavior can look deceptively easy, and a pervasive subjective bias often accompanies quantitative behavioral assessments. Therefore, studies have suggested that a trait such as pointing ability is reported from being non - heritable to being highly heritable22. Currently, the most high-throughput approaches for dog behavioral phenotyping are based on owner directed surveys1, as it offers the most flexible method of data collection. As a result, finer evaluation of behaviors becomes a formidable task. Though useful in assessing binary traits like presence or absence of diseases, breed specific phenotypes of large differences (e.g. herding or pointing), a drawback of this approach is the complete

inadequacy in estimating individual variation. Marker based association techniques, which are becoming increasingly common in detecting molecular signatures underlying morphology in dogs, however, depend on this within breed variation.

We took a decidedly quantitative approach focusing on innate tendencies related to hunting. We used a dual strategy whereby we concentrated on traits under strong selective pressure, and used a series of objective metrics based on fine resolution satellite telemetry to assess relevant behavioral phenotypes. Hunting dogs in this study were selected from field lines of English Pointer and Brittany breeds, and control dogs were members of non performance line pet dogs from Labradoodle and Labrador

Retriever group.

We extracted an array of quantitative, behavioral attributes including aerobic capabilities, search strategies, spatial cognition, and orientation relative to the handler from spatial tracks of focus dogs during hunting situations. We selected a 400 meter X

400 meters field with an overall homogeneity of natural vegetation, and unobstructed, uniform wind flow. We also predefined the path of the human handlers and kept it constant for all the trials to reduce variation in handler-relative metrics. To ensure that the experimental trials mimic actual hunt expeditions, single game birds were placed at one of several predefined locations in the field. All the hunting dogs located the birds and showed the characteristic pointing instance to indicate that the trials were successful in simulating actual hunting excursions. All the trials were carried out during fall (Aug – Oct) between 7.00 to 11.00 in the morning during a defined range of temperature or humidity conditions.

The movements of the dogs (n = 16) in each trial (fig1) were characterized by linear and circular metrics. We assessed speed, angular preference, relative orientation consistency, fractal dimension and straightness index (scale-invariant vs scale-variant measure), and total area covered. Handler effects were assessed by quartering ability

(competence of running between +45 to -45 degrees in front of the handler), distance from handler, and the consistency of movement relative to the handler. The ability to assess behavior at such a refined level enabled us to extract subtle and not so subtle differences in behavioral variation among individuals. The speed of dogs, as a main predictor for locating the bird quickly, exhibits condsiderable variable even within breeds of hunting dogs (Tukey’s post hoc analysis, p = 0.02). Similar significant within breed variation was observed in orientation consistency, and straightness of running paths which reflect search eagerness and circuitous searching of the dogs (fig 2).

To tackle multivariate data, principal components analysis was used to classify observed variation into linear combinations of genetically accessible phenotypes23.

Following in the same line, four independent correlated groups of behavior (explaining

75% variance) emerged from individual measures (table 1). The four latent dispositions described Search Eagerness, Linear Running, Handler Reliance, and Zigzagged

Searching. Energetic searching behavior was characterized by running at high speeds, covering a wide search area, and ranging far from the human handler. This principal component explained 34.5% of the variance between the dog breeds. Linear Running was majorly characterized by the straightness of the dog’s search path, whereas

Handler Reliance was explained by a strong orientation consistency relative to the handler. Sinuous Searching loaded high fractal dimension values and a relatively

reduced active search area. Each dog was then assigned a binary value of “affected” or

“control” based on high or low values on the principal component axis. Subsequent marker associations were performed on such assessment. Using genomic array data

(from the Illumina HD 27K array) for 5 energetic searchers and 5 control dogs which failed to attain a state of invigorated searching, genome wide association mapping found a 175 kb haplotype block on chromosome 1 (CFA1) (Genome wide p value 100,000

2 permutations = 0.07; Chromosome wide significance, X p – value 100,000 permutations = 0.016*) and another SNP on chromosome 27 (CFA27) (Genome wide p value 100,000 permutations =

2 0.07; Chromosome wide significance, X p – value 100,000 permutations = 0.014*). The associated SNPs on CFA 1 lies within 200 kb of a brain specific demethylase gene

(MBD2 – methyl cpGbinding domain protein 2), whereas the SNP on CFA 27 is adjacent to sodium and chloride dependent GABA transporter genes expressed in the brain (BGT1, GAT2, and GAT3).

We did not attain genome wide significance value of 0.05, but considering the modest sample size of our association study, a significance of 0.07 is nonetheless a compelling signal along with the significant chromosome wide association. Behavioral variation has been suggested in the past to arise from altered gene copies of regulatory1 and neurotransmitter related genes24. Coding for transcriptional repressor proteins (MBD2)9, and regulating levels of extracellular neurotransmitters in the brain (BGT1, GAT2,

GAT3)10, these genes can be strong candidates for such behavioral modulation.

Our results strongly suggest that a molecular understanding of complex behavior has to begin with a robust quantification at an individual level, and the best candidates for such assessments are innate traits under strong selection pressure. Dogs have been

suggested as ideal model for behavioral genetics for years25. An English pointer hunts, a Border collie herds, and a Greyhound chases even in the absence of any instructions.

The understanding of these complex phenotypes however requires careful estimation of the within breed variability. The logistical constraints of behavioral assessment from an ethological standpoint are often a small sample size, because all the animals need to be observed under similar natural environment. This is a significant issue with the potential to vitiate experimental power. However, here we show that a robust multidimensional characterization on carefully selected innate traits can potentially overcome such limitations.

Methods summary:

The dogs are brought to the field site (41°24'34. 55"N, 83°46'24. 35"W) by the owners on the day of behavioral experiments. The dogs’ collars are fitted with GPS receivers

(Garmin Forerunner 205, weight 90 grams) and the owners’ spatial movements are tracked with GPS receivers as well. As an experimental trial is in motion, the owners walk the square at a steady pace with folded arms, and refrain from issuing any commands to the dogs. The dogs are allowed to run freely. Spatial track points are recorded by the GPS receivers at an interval of one second and at an accuracy of ± 2 meters. For the re-projection and plotting of GPS tracks we used Google Earth pro 5.0.

After screening for artifacts and irregularities the raw GPS data was Haversine transformed to convert geographic coordinates into metric measurements. The tracks are then analyzed using Java DataGrinders http://caspar.bgsu.edu/~software/Java/).

Speed variables were calculated for successive moves captured by the GPS receivers.

The straightness index was calculated by an algorithm, where actual length of the path

(W) was fixed at 2 meters, evenly distributed around each GPS capture. The bee-line distance (D) was calculated between the two ends of this 2 meter path. The D/W ratio was measured to estimate straightness index. The fractal dimension was calculated for each path using the traditional dividers method17, but incorporated replications by remeasuring the path multiple times with a range of divider sizes (1 meter – 20 meters, with even increment of 1 meter in divider sizes) and taking a mean of all the measures.

Handler – dog distance was calculated for successive locations of handler and dog over the entire trial. Angular variables were calculated using circular statistics18.

Genomic DNA was isolated using the QiaAmp DNA blood extraction kit (Qiagen Inc).

DNA samples were purified according to Illumina sample preparation guidelines and eluted in reduced EDTA –TE buffer (10 mM Tris, pH 8.0, 0.1 mM EDTA, pH 8.0).

Illumina HD custom canine SNP arrays were used to obtain genotype calls (Illumina).

Only genotype calls with a p value of <0.01 were used for analysis. The program PLINK was used for the genome wide association analysis using the SNP set with minor allele frequencies >0.05, 75% genotype calls and no more than 10% missing genotypes per individual. Permutation testing for whole genome association with 100,000 permutations was performed using PLINK. Haplotype and association analysis for single chromosomes was performed using the program Haploview.

Figure 1. Representative hunt trials (10min) of dogs illustrating markedly different spatial patterns. (Brittany Spaniel: Yellow, English Pointer: Magenta, Non hunting breed:

Turquoise)

Figure 2. Individual variation characterized over three hunt trials across Hunting dog breeds (Brittany Spaniel: Blue, English Pointer: red), and control dogs (Violet)

Breed Trait Energetic Brittany Searching A T G T G T G Energetic Brittany Searching A T G T G T G Energetic English Pointer Searching A T G T G T G Energetic English Pointer Searching A T G T G T G Energetic English Pointer Searching A T G T G T G English Pointer Control G C A C A C A Labrador retriever Control G C A C A C A Labrador retriever Control G C A C A C A Labradoodle Control G C T A G C T A G C T A G Labradoodle Control G T G T G T G

Position on Chromosome 1 23.87 24.05 24.08 24.09 24.1 24.11 24.2

MBD2

Figure 3. SNP association based on the first principal component (Energetic searching) in 5 affected and 5 control dogs. The grouping of the dogs was based entirely on the behavioral characterization disregarding predefined specification of breeds.

Measurements PC1 PC2 PC3 PC4 Mean Speed 0.5 Mean Lateral Orientation Relative Orientation Consistency Fractal Dimension Index of Straightness Total Distance Traveled 0 Quartering Ability Mean Handler - Dog Distance Mean Dextral Orientation to Handler Orientation Consistency Relative to Handler Percent Area Searched -0.5 Percent Variance Explained 34.5 18.5 15.9 7.1 Cumulative Variance Explained 34.5 53 68.9 76 Explanation of Variables Search Linear Handler Zigzagged Eagerness Running Reliance Searching

Table 1. Factor loadings of a subset set of linear and circular dependent measures of dogs on the first four rotated principal components. Interpretation for each axis was made based on these factor loadings.

References:

1. Spady, T.C. & Ostrander, E.A. Canine behavioral genetics: pointing out the

phenotypes and herding up the genes. Am. J. Hum. Genet 82, 10-18 (2008).

2. Club, A.K. The Complete Dog Book: 20th Edition. (Ballantine Books: 2006).

3. Svartberg, K. Breed-typical behaviour in dogs—Historical remnants or recent

constructs? Applied Animal Behaviour Science 96, 293-313 (2006).

4. Lindblad-Toh, K. et al. Genome sequence, comparative analysis and haplotype

structure of the domestic dog. Nature 438, 803-819 (2005).

5. Karlsson, E.K. & Lindblad-Toh, K. Leader of the pack: gene mapping in dogs and

other model organisms. Nat Rev Genet 9, 713-725 (2008).

6. Parker, H.G., Shearin, A.L. & Ostrander, E.A. Man’s best friend becomes biology’s

best in show: genome analyses in the domestic dog. Annu. Rev. Genet 44, 309-336

(2010).

7. Jones, P. et al. Single-Nucleotide-Polymorphism-Based Association Mapping of Dog

Stereotypes. Genetics 179, 1033 -1044 (2008).

8. Miklosi, A. Dog Behaviour, Evolution, and Cognition. (Oxford University Press, USA:

2008).

9. Wolffe, A.P., Jones, P.L. & Wade, P.A. DNA demethylation. Proceedings of the

National Academy of Sciences 96, 5894 -5896 (1999).

10. Chen, N.-H., Reith, M.E.A. & Quick, M.W. Synaptic uptake and beyond: the sodium-

and chloride-dependent neurotransmitter transporter family SLC6. Pfl�gers Archiv

European Journal of Physiology 447, 519-531 (2004).

11. Gerlai, R. A guide to good behavior. Nat Neurosci 3, 1240 (2000).

12. Sutter, N.B. & Ostrander, E.A. Dog star rising: the canine genetic system. Nat Rev

Genet 5, 900-910 (2004).

13. Ostrander, E.A. & Kruglyak, L. Unleashing the canine genome. Genome Res. 10,

1271-1274 (2000).

14. Sutter, N.B. et al. A Single IGF1 Allele Is a Major Determinant of Small Size in Dogs.

Science 316, 112 -115 (2007).

15. Cadieu, E. et al. Coat variation in the domestic dog is governed by variants in three

genes. Science 326, 150-153 (2009).

16. Mosher, D.S. et al. A mutation in the myostatin gene increases muscle mass and

enhances racing performance in heterozygote dogs. PLoS Genet. 3, e79 (2007).

17. Clark, L.A., Wahl, J.M., Rees, C.A. & Murphy, K.E. Retrotransposon insertion in

SILV is responsible for merle patterning of the domestic dog. Proceedings of the

National Academy of Sciences of the United States of America 103, 1376 -1381

(2006).

18. Pelé, M., Tiret, L., Kessler, J.-L., Blot, S. & Panthier, J.-J. SINE exonic insertion in

the PTPLA gene leads to multiple splicing defects and segregates with the autosomal

recessive centronuclear myopathy in dogs. Hum. Mol. Genet. 14, 1417-1427 (2005).

19. Mignot, E. et al. Genetic linkage of autosomal recessive canine narcolepsy with a

mu immunoglobulin heavy-chain switch-like segment. Proc Natl Acad Sci U S A 88,

3475-3478 (1991).

20. Dodman, N.H. et al. A canine chromosome 7 locus confers compulsive disorder

susceptibility. Mol Psychiatry 15, 8-10 (2010).

21. J.N.C. What’s Wrong with My Mouse?: Behavioral Phenotyping of Transgenic and

Knockout Mice. (Wiley-Liss: 2000).

22. Serpell, J. The Domestic Dog: Its Evolution, Behaviour and Interactions with People.

(Cambridge University Press: 1996).

23. Chase, K. et al. Genetic basis for systems of skeletal quantitative traits: Principal

component analysis of the canid skeleton. Proc Natl Acad Sci U S A 99, 9930-9935

(2002).

24. Ito, H. et al. Allele frequency distribution of the canine dopamine receptor D4 gene

exon III and I in 23 breeds. J. Vet. Med. Sci. 66, 815-820 (2004).

25. Chase, K., Sargan, D., Miller, K., Ostrander, E.A. & Lark, K.G. Understanding the

genetics of autoimmune disease: two loci that regulate late onset Addison’s disease

in Portuguese Water Dogs. Int. J. Immunogenet. 33, 179-184 (2006).

CHAPTER III: QUANTITATIVE CHARACTERIZATION OF SPATIALLY EXPLICIT

HUNTING BEHAVIORS IN DOMESTIC DOGS, CANIS FAMILIARIS

Budhaditya Chowdhury, Moira van Staaden, and Robert Huber

Introduction

Dogs were among the first species to be domesticated around 15,000 years ago from multiple origins1–3. During the long course of time since then, they have been subjected to rigorous artificial selection for a number of morphological and behavioral characteristics giving rise to more than 300 different breeds, of which more than 170 are recognized by the American kennel club4. In “ancient” breeds, the genetic signs of admixture with wolves are evident2, and this has tailored some exceptional behavioral qualities into the dog. Some of these “ancient” breeds (originating >500 years ago) like

Basenji, Akita, and Afghan hound were selected for hunting expeditions. Some argue that selective breeding for increased hunting efficiency started at least 4000 years ago in the Middle East and North Africa5.

Hunting behavior in canids has been suggested to partition into sequential motor patterns like predatory search, orienting towards prey, pointing, chasing, and biting6.

During domestication, some of these patterns were selectively amplified to fit specific anthropocentric needs. Pointing dogs, bred to assist humans in open grassland hunting, were subjected to selective breeding for amplified predatory search pattern, with the complete exclusion of the chasing sequence6–8. A pointing dog who has

detected game is expected to assume a characteristic frozen stance that alerts the hunter to the prey’s hiding location. The maintenance of this behavior in breed lines suggest a strong genetic basis making for an ideal phenotype that can be characterized towards a better understanding of inter-breed heterogeneity as well as intra-breed homogeneity.

Behavioral markers that have been used for canine characterization include the general categories of fearfulness9, emotionality10, tendency to approach and withdraw in novel situations11, learning abilities12, working aptitude13, playfulness and activity14,

“reactivity” and “immaturity”15, agonistic signaling16, “personality”17–19, and aggression20–

22. Although these studies broaden our understanding of canid behavior, the utility of this approach is weakened by its dependence on subjective categorical rating scales.

Moreover, the factor aggressiveness used in the personality test of sheltered dogs is not necessarily comparable to the “aggressive” factor in companion dogs23, or to the factor used to assess aggression in German Shepherds and Belgian Tervueren17. There seems to be little difference in definitions of “temperament”, “personality”, or

“character”24. Confounding definitions of “temperament”, range from “the degree of liveliness”13 to “physical flexibility and intensity of reaction to different environmental stimuli”24, and makes characterization of behavioral traits challenging and cross-study comparisons impossible. Subjective rating scales put the reliability and validity of the personality tests25 under question26 especially after the findings that owners are not skilled in observing dog behavior, and misinterpretations are a regular occurrence27.

Performance traits like hunting are robust, can be characterized by objective metrics, and are rendered accessible for quantitative behavioral assessment in dogs.

Professional hunting associations have provided a strict framework for judging performance of individual dogs in hunting contexts. The system in place judges the performance of pointing breeds by assessing success in finding hidden prey, posture of pointing stance, tail position, and overall performance4,7. Some of these assessments include a mixture of characters, confounding the observation, and many are purely cosmetic. Successful execution of hunting tasks places heavy demands on a variety of the dog's sensory, cognitive, and orientation skills. Among these are sampling of the environment for significant olfactory stimuli, matching perceptive signatures with a specific search image, and ultimately guiding the process towards a successful conclusion in the form of pointing behavior8. Selected to be performed in the company of human hunters, hunting also requires strong cooperative skills. The active searching area of the dog around the human hunter is known as the range of the dog28, whereas the location of the dog relative to the human hunter at any given time during the hunt is described as “quartering”29. Ranges have been suggested to vary considerably among pointing breeds of dogs4, and dogs with better quartering skills search areas in front and on either side of the human hunter instead of trailing behind. Such and other fine perceptual skills related to hunting are often subtle and run the risk of being overlooked by simple observations. Also, the variability in the dog–handler relationship in a hunt trial is not adequately captured with subjective categorization7,29. Pointing dog breeds exhibit faculties for the essential hunting traits, including efficient search strategies, spatial abilities, sensitive olfaction and strong aerobic performance. To date, however, a rigorous, quantitative dissection of the different behavioral phenotypes and their variability has been lacking in canine models.

The present study aimed to explore hunting strategies in pointing dogs with objective quantitative approaches. Characterization was done by a host of metrics quantifying a range of behavioral components in the dog’s hunting performance, obtained by satellite telemetry methods. We determined inter- as well as intra-breed variation in abilities, strategies, and addressed the social effect of the handler under hunting conditions. The goal of this project was to specifically develop measurement tools for characterizing: (i) basic features of running behavior and directionality for hunting - we explored the degree to which individual dogs are able to cover a given area in search of prey by characterizing measures of running speed, degree of turning, directional preference, net distance travelled, and changes in these measures over time during a hunting trial; (ii) different search strategies and their effectiveness by analyzing spatial metrics that reflect how efficiently a dog searches a given area for their quarry, and degree of zig-zagging or sinuosity; (iii) relations of dogs towards their handlers focusing on how the animals alter their hunting strategies with an eye on party members, how far the dogs range out from their handler, the consistency of such ranges, quartering efficiency, and predictability of running from the handler’s movements; and (iv) spatial cognitive performance with the overlap between successive passes and the degree to which the same areas is searched repeated

Methods:

Animals and field site:

Two different breeds of hunting dogs (English Pointer; n = 8 Brittany; n = 4) are used along with two different breeds of non hunting dogs (Labrador retriever; n = 2,

Labradoodle; n = 2) for comparison of behavioral phenotypes. Hunting dogs in the study range from young pups, trained dogs and field-trial champions (Table 1). The dogs are escorted to the field site by the owners for behavioral quantification.

For behavioral experiments, a 5 hectare square field was selected near

Tontogany, Ohio (41°24'34. 55"N, 83°46'24. 35"W). The field characteristics included even grass cover (0.5 meters), unobstructed wind flow, and limited environmental heterogeneity (no trees, no brush, and no water bodies). The dogs are held in their mobile kennels in the adjacent field behind a barn which provides a visual obstruction as the experimenter sets the field.

Experimental setup:

An square trail 150m X 150m) with stakes (height: 1.5m) in the center of the experimental field (figure 1a). Selecting from a number of predefined locations at random, live game birds are planted in the field to reliably replicate the hunting experience of the dogs. Prior to setting these target birds the experimental field is carefully surveyed to ensure that there are no resident birds. The target birds are held in cages and the cages are wrapped with plastic while transporting in the field. To minimize stray bird scents in the even grass cover while transportation, a John Deere

Gator is used. The Gator is driven to all four corners of the square trail before placing

the bird in the field to control for stray residual scents while transporting. The bird is hidden in the grass cover to minimize visual detection from a distance. The farthest stake from the “Start” is selected prior to the experimental runs and the target is placed

15 meters away from the farthest stake (Figure 1a).

Behavioral procedures:

Each dog is allowed to rest for 30 minutes in the mobile kennels prior to the experimental trials. For the trials we used a repeated measures design, with each dog

(n = 16) being tested three times separated by a period of at least 15 minutes. In the first scenario, one of the four stakes is randomly chosen as the start location with the target bird in place. In the second scenario, the stake diagonally opposite to the “Start” stake (in the first trial) is chosen as the start location with the target bird in place. In the third scenario, one of the two remaining stakes is randomly chosen as the start location.

The trials are terminated when the dog successfully finds the bird location, or, after a maximum duration of 10 minutes. The dogs’ collars are fitted with GPS receivers

(Garmin Forerunner 205, weight 90g) and the handler’s spatial movements are similarly tracked with an additional GPS receiver. During the experimental trial the owner walks the square path delineated by the four stakes at a steady pace with folded arms, and is asked to refrain from issuing any commands to the dogs. The dogs are allowed to run freely for the entire duration of experimental trials.

Data Analysis

GPS Data

Individual GPS tracking were used to collect spatial data from both dogs and handlers. Spatial track points are recorded by the GPS receivers at an interval of one second and at an accuracy of ± 2 meters (Figure 1b). Spatial track points recorded by the GPS receivers are transferred into a computer (Mac OS X ver 10.5.8) on the day of collection. Prior to analysis every input file is plotted in Google Earth Pro ver 5.0 to ensure validity of the tracks (and to ascertain the quality of the GPS record). Spatial analyzis of the track coordinates was conducted using Java DataGrinders

Paths For the re-projection and plotting of GPS tracks we used Google Earth pro

5.0. Visualized tracks were screened for artifacts and irregularities in the data. In one of the tracks several data points were missing, presumably due to an intermittent loss of satellite signal, and the track was excluded from further analysis. The initial processing of the raw GPS data included Haversine transformation of geographic coordinate system to metric measurements.

The dog’s movements during each trial were characterized by a number of linear and circular metrics (Figure 3). The linear variables extracted from each experimental run included the dog’s (1) mean speed, (2) variance in speed, (3) maximum speed attained, (4) mean index of straightness, (5) variance in index of straightness, (6) fractal dimension, (7) total distance traveled, (8) mean distance to the handler, (9) maximum distance to handler, (10) quartering ability, and (11) percent of total area searched.

Circular descriptive measures included the dog’s (12) mean lateral orientation, (13) relative orientation consistency r, (14) mean dextral orientation to handler, and (15) orientation consistency relative to the handler.

Speed variables were calculated across successive moves between GPS coordinates. Straightness index was calculated by an algorithm that considered the angle between consecutive 1m sections around each point. The The actual distance (D) was calculated between the start of the preceding and the end point of the following segment. The D/2 ratio measured how much shorter this bee-line was compared to the maximum possible distance of 2m The fractal dimension was calculated for each path using the traditional dividers method30, but incorporated replications by remeasuring the path multiple times with a range of divider sizes (1 meter – 20 meters, with even increment of 1 meter in divider sizes) and taking a mean of all the measures. Distance traveled was calculated by the total path length of the dog’s run. Handler – dog distance was calculated for successive locations of handler and dog over the entire trial.

Quartering ability was estimated as the ratio of time the dog spent within -45 t0 +45 degrees in front of the handler’s location over the total duration of the trial. The experimental field was divided into 400 imaginary grid spaces. A measure of the area searched was estimated via the total number of grid cells visited by the dog in the trial per time.

The dogs’ mean relative angle of orientation, angular variance, length of mean vector, mean relative angle between handler’s move direction and dog, and the variance as well as vector strength were calculated using standard methods for circular descriptive statistics31.

Rayleigh circular statistics analyzed the relative angular directions of movement that the animals made over the course of individual trials. Because the initial positions of the dogs were randomized, the relative angular direction was calculated with 0 degree signifying forward direction and 180 degrees signifying a complete back tracking. A second order analysis of the Rayleigh movement angles was performed using a Java

Applet from the JavaGrinders library, which provided a grand mean angle and direction for each of the trial conditions. The mean circular measures were then transformed to fit within -180 degrees and + 180 degrees to avoid circular aggregation during statistical analysis. We fitted a linear regression model with means of the behavioral measures, and sexes, as well as training of dogs as predictor variables to estimate to what extent these variables affect behavior. To assess differences within and between breeds based on behavioral phenotypes, and to identify traits that most effectively distinguish between these breeds, we employed nested multivariate analysis of variance

(MANOVA), and Discriminant Function Analysis (DFA) on standardized data. DFA evaluates all measurements together, and derives linear combinations that maximize variation between groups and minimize variation within groups

Results:

Assessment of innate abilities

All hunting dogs (n = 12) attained the characteristic pointing posture at the end of hunt trials signifying a successful hunt expedition. The English pointers took 5 min +/-

2.4, the Brittany Spaniels took 2.91 min +/- 0.2 to locate game birds. Training has been suggested to affect efficiency in hunting and thus impact behavioral measurements.

However, multiple regression failed to identify any single metric that could have been affected by training, or sex of the dogs (Table 1). Means for all measured variables are listed in table 2. Table 3 summarizes behavioral differences between three different types of dogs viz, Brittany Spaniels, English Pointers, and non hunting dogs. Significant differences were found in Speed variables (Mean Speed, Variance in Speeed), Handler related distance variables (Mean Distance to handler, Variance in Handler Distance, and Maximum distance to Handler), Variance in Straightness Index, and in orientation consistency (Table 3).

Breed comparison on innate traits

The type specific behavioral attributes of the three dog groups differed significantly from each other (MANOVA Wilk’s Lambda p<<0.001). The Canonical centroid plot (Figure 3) distinguished dog types on the basis of Search Strategy, Aerobic

Competence, and Socio-spatial Relationships. Non-hunting dogs distinctly segregated from Hunting dogs on the first canonical axis formed from measures of orientation, and speed. The second canonical axis separated the two hunting dog breeds, viz., English

pointers and Brittany spaniels, with measures of Straightness index and maximum distance to the handler.

Factor analysis

Principal Components analysis of the innate tendencies produced four PC – axes, cumulatively accounting for 75 % of the total variation among individuals. Different modalities showed high loadings (table 1, Chapter II) on different PC – axes. Search

Eagerness loaded high on PC1, Linear Running on PC2, Handler Reliance on PC3, and

Zigzagged Searching on PC4. Search Eagerness behavior was characterized by running at high speeds, covering a wide search area, and ranging far from the human handler. This principal component explained 34.5 % of the variance between the dog breeds. Linear Running was majorly characterized by the straightness of the dog’s search path, whereas Handler Reliance was explained by a strong orientation consistency relative to the handler. Zigzagged Searching loaded high fractal dimension values and a relatively smaller active search area.

Discussion:

Scott and Fuller10 layed out a foundation of comparative dog behavior, however, that promise is still to be met. Coding of dog behavior can be deceptively easy, and the widespread disparity9-19 in assessing emotional traits makes breed comparison even tougher. Using a host of objective measures we tried to circumvent four major issues – i) the unreliability of questionnaire studies from owners with a wide variety of experience27, ii) the problem of causality where environmental factors are correlated with behavior without enough evidence32, iii) the issue with owner biases8, and iv) the

problem of the pervasive folk knowledge which assesses dogs with ambiguous traits like “intelligence”33 .

This work has shown that dog breeds differ in their innate behavioral tendencies in the context of hunting. We focused on the entire behavioral experience rather than the culmination of it in the form of pointing behavior. This allowed us to tease apart behavioral tendencies in a broader context of the total sensory background. Additionally this enabled us to compare hunting and non hunting breeds under similar conditions.

We found that English Pointers and Brittany Spaniels differ in their behavioral responses in hunt trials, and both these upland hunting breeds differ from non hunting breeds under similar socio – environmental conditions in key traits. We found running speed, relative angular orientation, and variation in such angular running preferences in a trial to be the major distinguishing factor between upland hunting dogs and non hunting dogs. Hunting dog breeds, on the other hand, differed in their ranging capabilities, consistency of maintaining such range parameters, and straightness of their running path. Upland hunting breeds tended to be faster, more variable in their running speed, and more likely to turn while maintaining a relatively straight running path than the non hunting breeds.

Interestingly, quartering capabilities, search coverage of the field, relative angular orientation from the handler, and maintenance of such angular preference did not seem to differ between English Pointers, Brittany Spaniels, and the non hunting breeds.

Dogs and socialized foxes are extremely capable of picking up human cues under smaller experimental settings34. This phenomenon has been explained by the correlated by-product hypothesis which states that a selection for canids that will approach

humans fearlessly and non-aggressively will eventually lead to a selection of a correlated trait enabling skillful perception of human gestures35. Belyaev’s foxes which were selected for tameness did equally well in perceiving human cues when compared to domestic dogs. Another idea on canine sociocognitive skills proposed by Miklosi36 suggests that in order to be successful in an anthropogenic environment, dogs were selected for heightened social behavior. These two views are not incompatible. Here we show that in a broader environmental setting, and under a performance trait context, domestic dogs play close attention to human movement patterns even when no vocal or physical cues are given.

Our results strongly suggest that complex behavioral phenotypes can be best studied by organizing them into linear combination. Search Eagerness, an essential phenotype related with hunting was properly characterized only after grouping the multivariate data into latent phenotypes. Indeed, this component explained most variance observed among the dogs. Handler Reliance was another interesting factor describing the socio-cognitive aspect of domestic dogs.

Breed creation is a relatively new phenomenon which started in the 1900s37. Our results suggest that perception of human cues has been adaptively selected before the creation of hunting breeds and has not changed in the dogs selectively bred to hunt.

Currently dogs are subjected to “a dangerous game”8 where the “form” takes the driver’s seat with a complete ignorance of behavior. As the breeders are encouraged to create the perfect form, many breed typical behaviors are getting lost. Without a careful comparative estimate spanning multiple breeds, it becomes challenging to assess the behaviors that played a defining role in selection of the dogs as working partners. We

anticipate that our study will emphasize the importance of performance traits that made dogs ideally suited to work with humans. We also predict that the ability to quantify behavior with objective measures at an individual level will also aid the hunt for genes underlying innate behaviors.

Figure 1a. Google Earth image showing the experimental field (blue), walking trail of the handler (yellow), stakes at four corners (black) and the position of target bird for the run which commences at “Start”. b. Hunt trials (10min) of dogs illustrating markedly different spatial patterns. (Brittany Spaniel: Yellow, English Pointer: Magenta, Non hunting breed: Turquoise)

Figure 2. Mean speed, mean handler-dog distance (a), mean handler-dog angle (b), quartering ability, relative orientation of the dog (c), orientation consistency, fractal dimension, total area searched, and straightness of path at every move (D/W) was estimated from each trial.

VHD

Figure 3. Canonical centroid plot describing the separation of behavioral attributes between different types of dog breeds. Canonical axes 1 and 2 represent linear combinations of standardized behavioral measures plotted as vectors, extending outward in a direction that conveys correlation. The length and direction of vector lines signifies their ability to distinguish the body types. Positively correlated vectors face the same direction, negatively correlated in opposite directions, whereas uncorrelated vectors are orthogonal. The clustering of the data points for each dog type, represented as 95% confidence limit circles with a central multivariate mean, differ in size according to the sample size (small circle = large sample size), and are spatially separated by the most effective characters.

Training Sex

Parameter Parameter

estimate (p – estimate (p –

value) value)

Dependent measures

Mean speed -0.015 (0.91) -0.364 (0.56)

Variance in speed -0.37 (0.43) -2.71 (0.17)

Maximum speed -0.32 (0.55) -2.53 (0.26)

Index of straightness 0.009 (0.16) 0.021 (0.44)

Variance in Index of straightness -0.003(0.3) -0.009 (0.5)

Fractal dimension -0.0002 (0.21) 0.0005 (0.45)

Distance traveled -6.63 (0.9) -143.2 (0.51)

Mean distance to handler 2.84 (0.33) 3.57 (0.76)

Maximum Handler distance 4.56 (0.49) -3.13 (0.9)

Quartering ability -0.66 (0.59) -6.94 (0.18)

Total area searched -0.0009 (0.9) -0.029 (0.34)

Angular orientation -0.145 (0.79) -2.69 (0.25)

Relative orientation consistency 0.006 (0.62) 0.025 (0.63)

Mean dextral orientation to handler -4.958 (0.34) 4.88 (0.81)

Orientation consistency relative to -0.002 (0.82) 0.052 (0.34) handler

Table 1. Standardized partial regression coefficients from multivariate regression analyses conducted on two separate measures that could influence performance

(training and sex) on behavioral characteristics.

Breed N MS VS MHD VH MXH SI VSI FD Q ML OC MDOH OCRH D D O

Brittan 4 2.0 3.12 29.6 287 66.3 0.7 0.1 1.01 52.75 3.4 0.73 3.57 0.37 y 3 5 8

English 8 2.5 5.17 34.8 447 89.11 0.7 0.1 1.01 46.6 3.07 0.73 44.69 0.32 pointer 8 4 7

Control 4 1.0 0.73 5.65 20.8 17.98 0.7 0.07 1.01 48.83 4.28 0.84 19.45 0.41 6 9

Table 2. Summary statistics for three types of dogs examined. The statistics for each type are Sample size (N), Mean Speed (MS), Variance in Speed (VS), Mean Distance to Handler (MHD), Variance in distance to Handler (VHD), Maximum Handler Distance

(MXHD), Straightness Index (SI), Variance in Straightness index (VSI), Fractal

Dimension (FD), Quartering Ability (Q), Mean Lateral Orientation (MLO), Orientation

Consistency (OC), Mean Dextral Orientation to Handler (MDOH), and Orientation

Consistency Relative to Handler (OCRH). Speed variables are in meters/sec, distance variables in meters, and angular measurements are in degrees. SI, VSI, FD, Q, OC, and

OCRH are unit less (as these represent ratios).

Variables F p-value

MS 8.71 0.01

VS 7.77 0.005

MHD 8.47 0.003

VHD 7.11 0.007

MXHD 10.81 0.005

SI 0.37 0.55

VSI 7.79 0.01

FD 0.16 0.69

Q 0.05 0.8

MLO 0.94 0.33

OC 8.44 0.01

MDOH 0.39 0.54

OCRH 1.81 0.19

Table 3. ANOVA table summarizing differences between different types of dogs based on the different objective metrics assessed from individual run of hunting and non hunting dogs.

References:

1. Vilà, C. et al. Multiple and Ancient Origins of the Domestic Dog. Science 276, 1687 -

1689 (1997).

2. vonHoldt, B.M. et al. Genome-wide SNP and haplotype analyses reveal a rich history

underlying dog domestication. Nature 464, 898-902 (2010).

3. Savolainen, P., Zhang, Y.-ping, Luo, J., Lundeberg, J. & Leitner, T. Genetic Evidence

for an East Asian Origin of Domestic Dogs. Science 298, 1610 -1613 (2002).

4. Club, A.K. The Complete Dog Book: 20th Edition. (Ballantine Books: 2006).

5. Serpell, J. The Domestic Dog: Its Evolution, Behaviour and Interactions with People.

(Cambridge University Press: 1996).

6. Coppinger, R. & Coppinger, L. Dogs: A New Understanding of Canine Origin,

Behavior and Evolution. (University Of Chicago Press: 2002).

7. Arkwright, W. The Pointer and His Predecessors. (Derry Argue: 1999).

8. Miklosi, A. Dog Behaviour, Evolution, and Cognition. (Oxford University Press, USA:

2008).

9. MAHUT, H. Breed differences in the dog’s emotional behaviour. Can J Psychol 12,

35-44 (1958).

10. Scott, J.P. & Fuller, J.L. Genetics and the Social Behavior of the Dog. (University Of

Chicago Press: 1998).

11. Plutchik, R. Individual and Breed Differences in Approach and Withdrawal in Dogs.

Behaviour 40, 302-311 (1971).

12. Pongrácz, P., Miklósi, Á., Vida, V. & Csányi, V. The pet dogs ability for learning from

a human demonstrator in a detour task is independent from the breed and age.

Applied Animal Behaviour Science 90, 309-323 (2005).

13. Slabbert, J.. & Odendaal, J.S.. Early prediction of adult police dog efficiency—a

longitudinal study. 64, 269-288 (1999).

14. Hart, B.L. & Miller, M.F. Behavioral profiles of dog breeds. J. Am. Vet. Med. Assoc.

186, 1175-1180 (1985).

15. Bradshaw, J.W. & Goodwin, D. Determination of behavioural traits of pure-bred

dogs using factor analysis and cluster analysis; a comparison of studies in the USA

and UK. Res. Vet. Sci. 66, 73-76 (1999).

16. Goodwin, D., Bradshaw, J.W.S. & Wickens, S.M. Paedomorphosis affects agonistic

visual signals of domestic dogs. Animal Behaviour 53, 297-304 (1997).

17. Svartberg, K. & Forkman, B. Personality traits in the domestic dog (Canis familiaris).

79, 133-155 (2002).

18. Jones, A.C. & Gosling, S.D. Temperament and personality in dogs (Canis

familiaris): A review and evaluation of past research. 95, 1-53 (2005).

19. Kubinyi, E. Comparative Social Cognition: From wolf and dog to humans. ccbr 2,

(2006).

20. Cattell, R.B. & Korth, B. The isolation of temperament dimensions in dogs. Behav

Biol 9, 15-30 (1973).

21. Netto, W.J. & Planta, D.J.U. Behavioural testing for aggression in the domestic dog.

52, 243-263 (1997).

22. van der Borg, J.A.M., Netto, W.J. & Planta, D.J.U. Behavioural testing of dogs in

animal shelters to predict problem behaviour. Applied Animal Behaviour Science 32,

237-251 (1991).

23. Goodloe, L.P. & Borchelt, P.L. Companion dog temperament traits. J Appl Anim

Welf Sci 1, 303-338 (1998).

24. Ruefenacht, S., Gebhardt-Henrich, S., Miyake, T. & Gaillard, C. A behaviour test on

German Shepherd dogs: heritability of seven different traits. 79, 113-132 (2002).

25. Martin, P. & Bateson, P. Measuring Behaviour: An Introductory Guide. (Cambridge

University Press: 2007).

26. Diederich, C. & Giffroy, J.-M. Behavioural testing in dogs: A review of methodology

in search for standardisation. 97, 51-72 (2006).

27. van den Berg, L., Schilder, M.B.H. & Knol, B.W. Behavior genetics of canine

aggression: behavioral phenotyping of golden retrievers by means of an aggression

test. Behav. Genet. 33, 469-483 (2003).

28. Wolters, R.A. Gun Dog: Revolutionary Rapid Training Method. (Dutton Adult: 1961).

29. Lamb, V. The Ultimate Hunting Dog Reference Book: A Comprehensive Guide to

More Than 60 Sporting Breeds. (Globe Pequot: 2006).

30. Mandelbrot, B. How Long Is the Coast of Britain? Statistical Self-Similarity and

Fractional Dimension. Science 156, 636 -638 (1967).

31. Batschelet, E. Circular Statistics in Biology. (Academic Press Inc: 1981).

32. Podberscek, A.L. & Serpell, J.A. The English Cocker Spaniel: preliminary findings

on aggressive behaviour. 47, 75-89 (1996).

33. Coren, S. The Intelligence of Dogs: Canine Conciousness and Capabilities. (The

Free Press/Macmillan: 1994).

34. Hare, B., Brown, M., Williamson, C. & Tomasello, M. The Domestication of Social

Cognition in Dogs. Science 298, 1634 -1636 (2002).

35. Hare, B. et al. Social cognitive evolution in captive foxes is a correlated by-product

of experimental domestication. Curr. Biol. 15, 226-230 (2005).

36. Miklósi, Á., Topál, J. & Csányi, V. Comparative social cognition: what can dogs

teach us? Animal Behaviour 67, 995-1004 (2004).

37. Parker, H.G. et al. Genetic Structure of the Purebred Domestic Dog. Science 304,

1160 -1164 (2004).

CHAPTER IV: QUANTITATIVE ETHO-GENETIC DISSECTION OF HUNTING

BEHAVIOR IN DOMESTIC DOGS, CANIS FAMILIARIS

Budhaditya Chowdhury, Danika Bannasch, Paul Morris, Moira van Staaden, and Robert

Huber

Introduction

Models of behavioral variation that are accessible to experimental and molecular dissection are in glaringly short supply. Domestic dogs represent an ideally system for such an endeavor as they (1) exhibit extremes in form and function, (2) have a unique genome organization, and (3) have had their complete genome sequenced with the development of molecular tools that make genetic dissections feasible.

Recent times have seen domestic dogs becoming the scientists’ pet model to explore genetic – phenotypic interactions1. Indeed, to the delight of the molecular biologist, thousands of years of selection on rarified traits are represented today by over

400 closed breeding populations2, with unique patterns of morphology and behavior3.

Making gene hunting even more feasible are the two bottleneck events that shaped the dog genome, one at domestication and one at breed creation4. Hunting was one of the primary traits of interest during the domestication of dogs6,7. Selective breeding for increased hunting efficiency started at least 4000 years ago in the Middle East and

North Africa8. Notable among resulting breeds are genetic lines of hunting dogs for detecting, localizing, and retrieving grassland prey. Intense selection over thousands of years5 has enabled these dogs with robust behavioral traits with considerable intra– as

well as inter-breed plasticity; Ideal from the standpoint of rigorous quantitative assessment and molecular characterization. The fixed sequential patterns of predatory behavior in wolves partition into several, more or less independent units9 and transitions between these patterns were disrupted during domestication10. In Pointing dogs predatory search patterns (orienting or pointing towards the prey upon detection) became more pronounced while the final chasing sequence disappeared9–11. The characteristic frozen pointing stance signals the end of the active exploratory mode9 and has a strong genetic basis2,11.

Successful execution of hunting tasks places heavy demands on a variety of the dog's sensory, cognitive, and orientation skills10. Pointing dog breeds exhibit stunning faculties for the essential hunting traits, including efficient search strategies, spatial abilities, sensitive olfaction and strong aerobic performance. Among these are sampling of the environment for significant olfactory stimuli, matching perceptive signatures with a specific search image, and ultimately guiding the process towards a successful conclusion in the form of pointing behavior. Selected to perform in the company of human hunters, hunting also requires strong cooperative skills. “Range” describes the active searching area of the dog relative to the hunter12 whereas “Quartering” denotes the location of the dog relative to that of the handler over the entire hunt trial13. Ranges have been suggested to vary considerably among pointing breeds of dogs. Dogs with better quartering skills search areas in front and on either side of the human hunter instead of trailing behind. Fine perceptual skills related to hunting are often subtle and run the risk of being overlooked by simple observations, whereas the variability in dog – handler relationship in a hunt trial is not adequately captured with current practices.

Molecular data suggest considerable variation within pointing breeds of dogs7 and kennel clubs recognize a similar level of behavioral variation. Dogs have emerged as an excellent system for analyzing the genetic underpinnings of phenotypic traits5,14–

17.The unique structure of purebred dog lines reduces genetic heterogeneity making them ideal for gene hunting. Genetically, dogs are placed in four basic clusters - (i)

Asian breeds (Chow, Akita and sled dogs such as husky); (ii) Guard dogs (mastiffs, bull terriers, boxers and Bernese mountain dog); (iii) Herding breeds (collie, shetland sheepdog, some sight hounds like the greyhound); and (iv) Modern hunting breeds

(English Pointer, German short-haired pointer, Brittanys) hounds and terriers15. The complete canine genome was sequenced in 20054, followed shortly by the identification of candidate genes for a range of morphological and behavioral heterogeneities18.

Normal morphological traits such as size differences19, coat variation20, muscular builds21, and pathologies such as chondrodysplasia22, have been successfully associated with underlying genetic makeup. Among the few studies that have attempted to correlate genes and behavior, the majority have focused on canine aggression. Such studies have suggested differential heritability for aggression directed towards humans or conspecifics23. Systematic searches for single nucleotide polymorphisms (snps) in genes for neurotransmitters related to aggression have suggested a differential allele of monoamine oxidase as one of the major players in golden retrievers24.

Similar to the issues imposed by the discipline25, the challenges in studying canine behavior genetics are twofold - a pervasive subjective bias often accompanies behavioral assessments, and the quantification of dog behavior can be deceptively easy. Therefore, not surprisingly, studies have suggested pointing ability to range from

being non heritable to being highly heritable26. Currently, the most high-throughput approach for dog behavioral phenotyping are owner directed surveys27, as it offers the most flexible method of data collection via phone, mail, or internet. Though useful in assessing binary traits like presence or absence of diseases, breed specific phenotypes of large differences (e.g. herding or pointing), a drawback of this approach is the complete inadequacy in estimating individual variation. Marker based association techniques, which are becoming increasingly common in detecting molecular signatures underlying morphology in dogs, however, depends on these within breed variations.

To address this challenge, we took a decidedly quantitative approach aimed at understanding innate tendencies related to hunting. Implementing a dual strategy whereby we concentrated on traits under high selective pressure, and used refined satellite telemetry techniques, we subjected the behaviors to rigorous quantification.

This allowed us to combine a quantitative characterization of individual hunting behavior phenotypes, with a whole-genome attempt to identify the genetic signatures for such behaviors.

Materials and Methods

Animals and field site:

Two hunting breeds (English Pointer; n = 8 Brittany; n = 4) are used along with two non hunting breeds (Labrador retriever; n = 2, Labradoodle; n = 2) for comparison of behavioral phenotypes(Table 1). For behavioral experiments, a 5 hectare square field is chosen in (41°24'34. 55"N, 83°46'24. 35"W). The field characteristics included even grass cover (0.5 meters), unobstructed wind flow, and limited environmental heterogeneity (no trees, no brush, and no water bodies).

Behavioral procedures:

A square trail (150 m X 150 m) is marked with stakes (height: 1.5 m) in the center of the experimental field (figure 1a, Chapter III). Alive game birds held in cages are planted in the field to replicate the hunting experience of the dogs. The bird is hidden in the grass cover to minimize visual detection from a distance.

For the trials we used a repeated measures design, with each dog (n = 16) being tested three times separated by a period of at least 15 minutes. In the first scenario, one of the four stakes is randomly chosen as the start location with the target bird in place. In the second scenario, the stake diagonally opposite to the “Start” stake (in the first trial) is chosen as the start location with the target bird in place. In the third scenario, one of the two remaining stakes is randomly chosen as the start location. The trials are terminated when the dog successfully finds the bird location, or, after a maximum duration of 10 minutes. The dogs’ collars are fitted with GPS receivers (Garmin Forerunner 205, weight 90 grams) and the owners’ spatial movements are tracked with GPS receivers

as well. As an experimental trial is in motion, the owners walk the square path delineated by the four stakes at a steady pace with folded arms, and refrain from issuing any commands to the dogs. The dogs are allowed to run freely for the entire duration of experimental trials.

Data collection and processing:

Behavioral Data:

The GPS receivers are used to collect spatial data from both dogs and handlers.

Spatial track points are recorded by the GPS receivers at an interval of one second and at an accuracy of ± 2 meters. Spatial track points recorded by the GPS receivers are transferred into a computer (Mac OS X ver 10.5.8) on the day of collection. Prior to analysis every input file is plotted in Google Earth Pro ver 5.0 to ensure validity of the tracks (and to make sure the GPS receivers didn’t record any noise). The tracks are then analyzed using Java DataGrinders http://caspar.bgsu.edu/~software/Java/). 15 different measurements were then extracted from each track (for details of path analysis, see chapter 2).

Genetic Data:

Blood samples were acquired from dog owners (blood drawn by family veterinarians). For each dog owners donated 6ml of blood in anticoagulant treated

Vacutainer tubes. The blood samples were stored at -80 degrees till genomic DNA extraction and analysis.

Molecular analysis:

Genomic DNA was isolated using the QiaAmp DNA blood extraction kit (Qiagen

Inc). DNA samples were purified according to Illumina sample preparation guidelines

(http://www.genomics.princeton.edu/microarray/downloadables/

ILLUMINA/TruSeq_DNA_SamplePrep__Guide_15005180_A.pdf) or through the use of

Centricon spin columns (Millipore), and eluted in reduced EDTA –TE buffer (10 mM

Tris, pH 8.0, 0.1 mM EDTA, pH 8.0).

Illumina HD custom canine SNP arrays were used to obtain genotype calls

(Illumina). Microarray work was performed by the UC Davis Cancer Center Genomics and Expression Shared Resources, Sacramento, CA. DNA was amplified and labeled according to the manufacturer’s protocol, and arrays were washed and stained on a

Fluidics Station 450 and were scanned on a GeneChip Scanner 3000.

Only genotype calls with a p value of <0.01 were used for analysis. The program PLINK

30 was used for the genome wide association analysis using the v2 Platinum SNP set with minor allele frequencies >0.05 , 75% genotype calls and no more than 10% missing genotypes per individual. Permutation testing for whole genome association with

100,000 permutations was performed using PLINK. Haplotype and association analysis for single chromosomes was performed using the program Haploview 31.

For each association analysis, behavioral measurements were used to group dogs into affected and controls (based on high and low values, respectively), disregarding predefined breed characterization.

Results:

We classified all the behavior measures into four independent correlated groups which explained 76 % variance (table 1, Chapter II). The four latent dispositions described Search Eagerness, Linear Running, Handler Reliance, and Zigzagged

Searching. Energetic searching behavior was characterized by running at high speeds, covering a wide search area, and ranging far from the human handler. This principal component explained 34.5 % of the variance between the dog breeds. Linear Running was majorly characterized by the straightness of the dog’s search path, whereas

Handler Reliance was explained by a strong orientation consistency relative to the handler. Zigzagged Searching loaded high fractal dimension values and a relatively smaller active search area. Each dog was then assigned a binary value of “affected” or

“control” based on high or low values on the rotated principal component axis (Table 2).

Subsequent marker associations were performed on such assessment (See appendix

I). Using genomic array data (from the Illumina HD 27K array) for 5 energetic searchers and 5 control dogs which failed to attain a state of invigorated searching, genome wide association mapping found a 175 kb haplotype block on chromosome 1 (CFA1)

2 (Genome wide p value 100,000 permutations = 0.07; Chromosome wide significance, X p – value 100,000 permutations = 0.016*) and another SNP on chromosome 27 (CFA27) (Genome

2 wide p value 100,000 permutations = 0.07; Chromosome wide significance, X p – value 100,000 permutations = 0.014*) (Figure 1). The associated SNPs on CFA 1 lies within 200 kb of a brain specific demethylase gene (MBD2 – methyl cpGbinding domain protein 2) (Figure

3, Chapter II), whereas the SNP on CFA 27 is adjacent to sodium and chloride dependent GABA transporter genes expressed in the brain (BGT1, GAT2, and GAT3).

Discussion:

We found latent dispositions in innate hunting behaviors in domestic dogs, viz.,

Search Eagerness, Linear Running, Handler Reliance and Zigzagged Searching. The experience of hunting is thus a function of motivation, aerobic capability, socio cognitive skills, and strategy. A separation of innate tendencies facilitated the grouping of individual dogs based on performance and aided in the marker based association.

Natural behavioral traits like energetic searching are likely to predate breed creation – selection on specific traits might have merely enhanced part of the behavioral repertoire in breeds (like pointing) and not the original behavioral sequence. From a cross breed association point of view, this makes it more challenging as the haplotype blocks will be much shorter and call for much denser SNP chips. We accounted for the population stratification effects and cleaned up the association by 100,000 permutations. Although we did not attain genome wide significance value of 0.05, but considering the modest sample size of our association study, a significance of 0.07 is nonetheless a compelling signal along with the significant chromosome wide association.

Behavioral variation has been suggested in the past to arise from altered gene copies of regulatory27 and neurotransmitter related genes32. Coding for transcriptional repressor proteins (MBD2)33, and regulating levels of extracellular neurotransmitters in the brain (BGT1, GAT2, GAT3)34, these genes can be strong candidates for such behavioral modulation.

We have shown the existence of within breed variation at a behavioral level and identified interesting genomic regions which could potentially play a role in the maintenance of such behaviors. The logistical constraints of behavioral assessment from an ethological standpoint are often a small sample size, because all the animals need to be tested under similar natural environment. This is a major issue with the potential to vitiate experimental power. However, here we show that a robust multidimensional characterization on carefully selected innate traits can potentially overcome such limitations. Our results strongly suggest that a molecular understanding of complex behavior has to begin with a robust quantification at an individual level, and the best candidates for such assessments are innate traits under strong selection pressure.

Figure 1. Genome Wide Association Study (GWAS) showing regions of strong association with Search Eagerness on Chromosome 1 (a), and on Chromosome 27 (b).

The figures show both the raw association and the permuted associations.

Dog Breed Sex Age Training Obtained (years) Allie Brittany F 2 0 Ben Brittany M 4 2 Jake Brittany M 4 2 Maddie Brittany F 2 0 Browser English Pointer M 10 8 Gradie English Pointer F 3 1 Mickey English Pointer F 2 0 Nugget English Pointer F 2 0 Riley English Pointer M 3 1 Stormy English Pointer M 2 0 Tank English Pointer M 5 3 Thunder English Pointer M 4 2 Rumi Non Hunting Dog F 3 0 Lola Non Hunting Dog F 3 0 Hera Non Hunting Dog F 6 0 Zeus Non Hunting Dog M 9 0

Table 1. Summary of dogs used in the experimental trials. Rumi and Lola are

Labradoodle breeds; Hera and Zeus are Labrador Retrievers. Training obtained refers to active hunting training.

Dog Type PC1 PC2 PC3 PC4 Allie Brittany Maddie Brittany Browser English Pointer Nugget English Pointer Stormy English Pointer Thunder English Pointer Rumi Non Hunting Dog Lola Non Hunting Dog Hera Non Hunting Dog Zeus Non Hunting Dog Variables Search Linear Handler Zigzagged Eagerness Running Reliance Searching

Affected Control

Table 2. Subgroups of affected and control dogs created on the basis of loading of the four principal components.

References:

1. Sutter, N.B. & Ostrander, E.A. Dog star rising: the canine genetic system. Nat Rev

Genet 5, 900-910 (2004).

2. Club, A.K. The Complete Dog Book: 20th Edition. (Ballantine Books: 2006).

3. Svartberg, K. Breed-typical behaviour in dogs—Historical remnants or recent

constructs? Applied Animal Behaviour Science 96, 293-313 (2006).

4. Lindblad-Toh, K. et al. Genome sequence, comparative analysis and haplotype

structure of the domestic dog. Nature 438, 803-819 (2005).

5. vonHoldt, B.M. et al. Genome-wide SNP and haplotype analyses reveal a rich history

underlying dog domestication. Nature 464, 898-902 (2010).

6. Lindsay, S.R. Handbook of Applied Dog Behavior and Training, Vol. 1: Adaptation

and Learning. (Iowa State University Press: 2000).

7. Parra, D., Méndez, S., Cañón, J. & Dunner, S. Genetic differentiation in pointing dog

breeds inferred from microsatellites and mitochondrial DNA sequence. Anim. Genet.

39, 1-7 (2008).

8. Clutton-Brock, J. A Natural History of Domesticated Mammals. (Cambridge University

Press: 1999).

9. Coppinger, R. & Coppinger, L. Dogs: A New Understanding of Canine Origin,

Behavior and Evolution. (University Of Chicago Press: 2002).

10. Miklosi, A. Dog Behaviour, Evolution, and Cognition. (Oxford University Press, USA:

2008).

11. Arkwright, W. The Pointer and His Predecessors. (Derry Argue: 1999).

12. Wolters, R.A. Gun Dog: Revolutionary Rapid Training Method. (Dutton Adult: 1961).

13. Lamb, V. The Ultimate Hunting Dog Reference Book: A Comprehensive Guide to

More Than 60 Sporting Breeds. (Globe Pequot: 2006).

14. Leonard, J.A. et al. Ancient DNA evidence for Old World origin of New World dogs.

Science 298, 1613-1616 (2002).

15. Ostrander, E.A. & Kruglyak, L. Unleashing the canine genome. Genome Res. 10,

1271-1274 (2000).

16. Sutter, N.B. et al. A Single IGF1 Allele Is a Major Determinant of Small Size in Dogs.

Science 316, 112 -115 (2007).

17. Parker, H.G. & Ostrander, E.A. Canine Genomics and Genetics: Running with the

Pack. PLoS Genet 1, e58 (2005).

18. Ostrander, E.A., Giger, U. & Lindblad-Toh, K. The Dog and Its Genome. (Cold

Spring Harbor Laboratory Press: 2005).

19. Sutter, N.B. et al. A Single IGF1 Allele Is a Major Determinant of Small Size in Dogs.

Science 316, 112 -115 (2007).

20. Cadieu, E. et al. Coat variation in the domestic dog is governed by variants in three

genes. Science 326, 150-153 (2009).

21. Mosher, D.S. et al. A Mutation in the Myostatin Gene Increases Muscle Mass and

Enhances Racing Performance in Heterozygote Dogs. PLoS Genet 3, e79 (2007).

22. Parker, H.G. et al. An Expressed Fgf4 Retrogene Is Associated with Breed-Defining

Chondrodysplasia in Domestic Dogs. Science 325, 995 -998 (2009).

23. Mackenzie, S.A., Oltenacu, E.A.B. & Leighton, E. Heritability estimate for

temperament scores in German shepherd dogs and its genetic correlation with hip

dysplasia. Behavior Genetics 15, 475-482 (1985).

24. Hashizume, C. et al. Identification of an cysteine-to-arginine substitution caused by

a single nucleotide polymorphism in the canine monoamine oxidase B gene. The

Journal of veterinary medical science the Japanese Society of Veterinary Science 67,

199-201 (2005).

25.J.N.C. What’s Wrong with My Mouse?: Behavioral Phenotyping of Transgenic and

Knockout Mice. (Wiley-Liss: 2000).

26. Serpell, J. The Domestic Dog: Its Evolution, Behaviour and Interactions with People.

(Cambridge University Press: 1996).

27. Spady, T.C. & Ostrander, E.A. Canine behavioral genetics: pointing out the

phenotypes and herding up the genes. Am. J. Hum. Genet 82, 10-18 (2008).

28. Mandelbrot, B. How Long Is the Coast of Britain? Statistical Self-Similarity and

Fractional Dimension. Science 156, 636 -638 (1967).

29. Batschelet, E. Circular Statistics in Biology. (Academic Press Inc: 1981).

30. Purcell, S. et al. PLINK: A Tool Set for Whole-Genome Association and Population-

Based Linkage Analyses. Am J Hum Genet 81, 559-575 (2007).

31 Barrett, J.C., Fry, B., Maller, J. & Daly, M.J. Haploview: analysis and visualization of

LD and haplotype maps. Bioinformatics 21, 263-265 (2005).

32. Ito, H. et al. Allele frequency distribution of the canine dopamine receptor D4 gene

exon III and I in 23 breeds. J. Vet. Med. Sci. 66, 815-820 (2004).

33. Wolffe, A.P., Jones, P.L. & Wade, P.A. DNA demethylation. Proceedings of the

National Academy of Sciences 96, 5894 -5896 (1999).

34. Chen, N.-H., Reith, M.E.A. & Quick, M.W. Synaptic uptake and beyond: the sodium-

and chloride-dependent neurotransmitter transporter family SLC6. Pfl�gers Archiv

European Journal of Physiology 447, 519-531 (2004).

35. Chase, K., Sargan, D., Miller, K., Ostrander, E.A. & Lark, K.G. Understanding the

genetics of autoimmune disease: two loci that regulate late onset Addison’s disease

in Portuguese Water Dogs. Int. J. Immunogenet. 33, 179-184 (2006).

APPENDIX I: SUPPLEMENTARY MATERIAL FOR CHAPTER IV

(Genome Wide Association for Principal Components describing innate tendencies related to hunting)

Principal Component 1 (Search Eagerness)

CHR SNP EMP1 EMP2 1.0 1_23875145 0.008398 0.07219 27.0 27_44946112 0.008398 0.07219 1.0 1_20282233 0.008398 0.5613 1.0 1_52992334 0.008398 0.5613 2.0 2_78874898 0.008398 0.5613 7.0 7_22946670 0.008398 0.5613 13.0 13_64890114 0.008398 0.5613 14.0 14_40566613 0.008398 0.5613 15.0 15_11276527 0.008398 0.5613 15.0 15_11320580 0.008398 0.5613 15.0 15_45150033 0.008398 0.5613 15.0 15_57663105 0.008398 0.5613 24.0 24_18255061 0.008398 0.5613 25.0 25_52359817 0.008398 0.5613 25.0 25_54318201 0.008398 0.5613 27.0 27_42071868 0.008398 0.5613 35.0 35_20459838 0.008398 0.5613 37.0 37_4703208 0.008398 0.5613 38.0 38_15242360 0.008398 0.5613 38.0 38_18386084 0.008398 0.5613 23.0 23_26178110 0.0154 0.5699 39.0 39_28827055 0.008398 0.6039 39.0 39_1129524 0.008398 0.6039

Principal Component 2 (Linear Running)

CHR SNP EMP1 EMP2 38.0 38_18355898 0.004799 0.4415 15.0 15_11754363 0.004799 0.5889 18.0 18_20195036 0.004799 0.5889 38.0 38_25191494 0.004799 0.5889 20.0 20_16599062 0.004799 0.6427 1.0 1_23875145 0.04419 0.8202 2.0 2_79859243 0.009998 0.8202 6.0 6_65937260 0.009398 0.8202 6.0 6_65957038 0.009398 0.8202 9.0 9_25327939 0.009998 0.8202 14.0 14_46856584 0.009398 0.8202 14.0 14_46866547 0.009398 0.8202 15.0 15_12272543 0.008798 0.8202 15.0 15_12289457 0.008798 0.8202 15.0 15_12300343 0.008798 0.8202 27.0 27_44946112 0.04419 0.8202 2.0 2_76297100 0.004799 0.9542 2.0 2_76317182 0.004799 0.9542 2.0 2_77782920 0.004799 0.9542 2.0 2_79626069 0.004799 0.9542 3.0 3_45231951 0.004799 0.9542 3.0 3_89935922 0.004799 0.9542 4.0 4_6720333 0.004799 0.9542 4.0 4_6801582 0.004799 0.9542 4.0 4_6818455 0.004799 0.9542 4.0 4_6834620 0.004799 0.9542 4.0 4_6842536 0.004799 0.9542

Principal Component 3 (Handler Reliance)

CHR SNP EMP1 EMP2 1.0 1_65625421 0.005399 0.8912 11.0 11_23811264 0.005399 0.8912 11.0 11_23823002 0.005399 0.8912 15.0 15_40597459 0.005399 0.8912 16.0 16_48203816 0.005399 0.8912 19.0 19_27924678 0.005399 0.8912 31.0 31_30550087 0.04739 0.8912 32.0 32_6242993 0.04739 0.8912 15.0 15_40696301 0.0146 0.9678 15.0 15_40716317 0.0146 0.9678 21.0 21_4235951 0.014 0.9678 21.0 21_42510675 0.014 0.9678 31.0 31_8636765 0.0138 0.9678 31.0 31_29673583 0.0126 0.9678 35.0 35_7844652 0.0124 0.9678 35.0 35_7867523 0.0124 0.9678

Principal Component 4 (Zigzagged Searching)

CHR SNP EMP1 EMP2 2.0 2_34453955 0.006799 0.5495 13.0 13_19994508 0.006799 0.5495 14.0 14_48793006 0.006799 0.5495 14.0 14_48814097 0.006799 0.5495 31.0 31_10812425 0.006799 0.5495 31.0 31_10827836 0.006799 0.5495 9.0 9_64222745 0.006799 0.5579 36.0 36_18828374 0.0142 0.5579 36.0 36_18838099 0.006799 0.5579 1.0 1_25146731 0.006799 0.8778 1.0 1_25150760 0.006799 0.8778 1.0 1_50254591 0.04559 0.8778 1.0 1_50255732 0.04559 0.8778

APPENDIX II: IACUC APPROVAL