This study attempted to examine three aspects of Equus caballus foal snapping behavior. First, it suggested that the previous theoretical explanation for snapping established by behavioral researchers was incorrect. Second, as a means of suggesting an alternative hypothesis, this study proposed that snapping behavior could be a modified play response that was elicited when foals were confused by the complex social signals displayed by conspecifics. Finally, this project tested the aforementioned hypothesis by recording interactions between foals and conspecifics and analyzing the snapping intensity with each subject's previous level of social experience. There were two indicators utilized to establish social experience. First, the "age" of the foal was employed as a measure of overall life experience and development. Second, the number of hours per day the foal was exposed to conspecifics gave an effective measure of the amount of time the subject had an opportunity to learn the complex Equine visual communication (referred to as "out-time"). / Master of Science

<p>Misunderstanding or misinterpreting specific animal behavior and welfare terminology, as well as principles of learning theory, may influence an individual’s perceptions of horse behavior. This error could potentially result in unnecessary applications of horse training principles and/or human interventions, which could potentially worsen the behavior or situation, leading to unnecessary welfare problems. </p><p>The purpose of this dissertation was to explore interpretations and understandings of specific animal behavior and welfare terminology, and learning theory principles, as applied to horses, among adults within the horse industry. Chapters 1 and 2 introduced, summarized, and linked the connection between horse behavior, horse welfare, learning theory, and schema theory. Chapter 3 and Chapter 4 each pilot-tested an online survey that was completed at the convenience of each participant to explore these interpretations and understandings. Chapter 3 (n = 46) utilized a survey containing general demographic questions, psychographic questions related to horse industry involvement, five videos of horse-human interactions (each with corresponding heart rate, HR, data), and 11 learning theory scenarios. Chapter 4 (n = 21) used a survey containing general demographic questions and five videos of various human-horse interactions, including the same five videos with horse HR information included. Using results from Chapter 3 and Chapter 4, further investigation of how individuals interpret and understand specific animal behavior and welfare terminology, and learning theory principles, as applied to horses was explored across a larger sample of adults within the horse industry. Chapter 5 (n = 1,145) utilized the same survey instrument as Chapter 3 among a more robust sample of adults within the horse industry. </p><p>Across all three studies, the main results suggest that although participants demonstrated a high level of agreement between each other when identifying fear, stress, and reactivity to describe a horse’s behavior, participants could not correctly define fear, stress, reactivity, or principles of learning theory, as related to horses. They also could not connect these states to an important physiological factor, i.e., HR, when identifying these states. Results across all three studies suggest that most participant demographics (such as age, gender, or race) did not influence participants’ abilities to correctly identify or define fear, stress, reactivity, or learning theory principles. Similarly, results from Chapters 3 and 5 suggest that most participant psychographics, such as horse ownership, or level of involvement with horses, did not affect ability to correctly identify key horse behaviors related to fear, stress, and reactivity, or understanding of learning theory principles, or ability to correctly define these states and principles. </p><p> Overall, this dissertation identified the need for additional education when it comes to clearly defining specific states such as fear, stress, and reactivity for individuals across the horse industry. Additional psychographic factors, such as an individual’s specific role or niche in the industry, or an individuals’ overall view of horses, should be further explored.</p>

Mammal species often live in social groups, but the factors that promote group cohesion can be difficult to analyze due to the prevalence of strong group affiliations. Feral horses maintain stable harems of one or two males and several females, and harem stability is strongly related to individual fitness. Anecdotal evidence and an early study in the non-breeding season suggest that management of the Shackleford Banks island horses with immunocontraception reduces harem stability in the population, providing an opportunity to study the factors that influence harem stability. I investigated the effects of the immunocontraceptive PZP on harem stability during the breeding season and examined mare activity budgets and harassment rates to determine if these factors influence harem stability. I hypothesized that 1) immunocontraception would increase the rates at which mares changed harems during the breeding season 2) activity budgets of contracepted individuals would differ significantly from those of uncontracepted individuals, and 3) contracepted mares would experience greater levels of harassment associated with changing harems than uncontracepted mares. I found that the immunocontraceptive does increase harem changes during the breeding season. I also found that contracepted mares have different activity budgets than uncontracepted mares; as predicted, contracepted mares grazed less and moved more than uncontracepted controls. The factors that influence mare activity budgets included immunocontraception, harem stallion, number of individuals in the harem, number of mares in the harem and body condition of the mare, as well as some interactions between factors. I found that high harassment rates by both harem stallions and other mares in the harem are correlated with higher harem change rates and that contracepted mares are harassed more than uncontracepted mares. These results indicate that the immunocontraceptive does influence harem stability in this feral horse population, potentially through alterations in activity budgets and harassment rates.

This thesis investigates whether Computer Vision can be a useful tool in interpreting the behaviors of monitored horses. In recent years, research in the field of Computer Vision has primarily focused on people, where pose estimation and action recognition are popular research areas. The thesis presents a pose classification network, where input features are described by estimated 2D key- points of horse body parts. The network output classifies three poses: ’Head above the wither’, ’Head aligned with the wither’ and ’Head below the wither’. The 2D reconstructions of keypoints are obtained using DeepLabCut applied to raw video surveillance data of a single horse. The estimated keypoints are then fed into a Multi-layer preceptron, which is trained to classify the mentioned classes. The network shows promising results with good performance. We found label noise when we spot-checked random samples of predicted poses and comparing them to the ground truth, as some of the labeled data consisted of false ground truth samples. Despite this fact, the conclusion is that satisfactory results are achieved with our method. Particularly, the keypoint estimates were sufficient enough for these poses for the model to succeed to classify a hold-out set of poses. / Uppsatsen undersöker främst om datorseende kan vara ett användbart verktyg för att tolka beteendet hos övervakade hästar. Under de senaste åren har forskning inom datorseende främst fokuserat på människor, där pose-estimering och händelseigenkänning är populära forskningsområden. Denna avhandling presenterar ett poseklassificeringsnätverk där indata beskrivs av uppskattade 2Dnyckelpunkter (eller så kallade intressepunkter) för hästkroppsdelar. Nätverket klassificerar tre poser: ’Huvud ovanför manken’, ’Huvud i linje med manken’ och ’Huvudet nedanför manken’. 2D-rekonstruktioner av nyckelpunkter erhålls med hjälp av DeepLabCut, applicerad på rå videoövervakningsdata för en häst. De uppskattade nyckelpunkterna matas sedan in i ett flerskikts- preceptron, som tränas för att klassificera de nämnda klasserna. Nätverket visar lovande resultat med bra prestanda. Vi hittade brus i etiketterna vid slumpmässiga stickprover av förutspådda poser som jämfördes med sanna etiketter där några etiketter bestod av falska sanna etiketter. Trots detta är slutsatsen att tillfredsställande resultat uppnås med vår metod. Speciellt var de estimerade nyckelpunkterna tillräckliga för dessa poser för att nätverket skulle lyckas med att klassificera ett separat dataset av samma osedda poser.