|
nif:isString
|
-
Data on dog behavior and environmental factors was collected using a validated owner-filled questionnaire, (S1 Appendix) which has been earlier shown to correlate with dogs' behavior in test situations (external validity) and to have good test-retest reliability [32]. The questionnaire includes 35 questions (S1 Appendix) altogether. The potential fearful reaction and frequency (0–4) towards strangers, unfamiliar dogs, and new situations were asked (S1 Appendix) [32]. To reduce the possible subjectivity of the owner’s judgment, we added a question for the owners to describe ‘how exactly’ the dog behaved in a specific situation. If the owner felt that the dog was fearful when meeting a stranger (or strange dogs, and/or novel situations), the owner had to indicate a specific reaction: how did the dog behave (for example, the dog withdraws when meeting a stranger). Similarly, if the owner felt that the dog did not show fear towards a stranger, a more specific description of the reactions was required to enable our own evaluation of the situation. During data collection, the questionnaire was modified three times, resulting in four slightly different versions of the questionnaire (the first one being a paper version, the three others—online questionnaires). The main questions regarding our target traits, fearfulness (towards persons, dogs, and new situations), noise sensitivity, and separation anxiety have not changed between the versions. The main difference comes from adding further background questions to versions three and four (maternal care, place of birth, type of food, extra nutrients, time spent alone/day, daily exercise) to better document the early life experiences and conditions of the dogs. We have also removed one unclear question that continuously confused owners (question on fearfulness towards novel objects), which is also not included in the following analysis. The latest version also has more detailed questions on ‘bold’ behavior. Several background questions regarding early life experiences, such as socialization and the puppy period, were included, as well as more general questions concerning daily routines and diet, with food types and extra nutrients. Instead of trying to capture the entire spectrum of phenotypic variation in fearfulness, we aimed to structure the questionnaire so that it would find the most fearful individuals as 'cases', and those with no marked fear reactions as 'controls'. We derived several behavioral variables from the questionnaire data (Table 1), which were used as response variables in analyzing the associated environmental factors.
Table data removed from full text. Table identifier and caption: 10.1371/journal.pone.0141907.t001 Behavioral and environmental variables derived from questionnaire data. We advertised the questionnaire to owners of all breeds, but especially Great Danes, German Shepherds, Belgium Shepherds, Staffordshire Bull terriers, Lagotto Romagnolos and Salukis. This is because we aimed for large sample sizes from breeds with a large number of existing blood samples already in our Dog DNA bank. The questionnaire was advertised via breed clubs and Facebook. Both shy and bold individuals, as well as dogs with fearful reactions to loud noises, and also dogs with no marked behavioral reactions to loud noises were invited to participate.
Statistical analysis and used behavioral variables: We analyzed the possible association of environmental factors with several anxieties using the generalized linear mixed model (GLMMs) with binomial distribution (PROC GLIMMIX, SAS version 9.3), using breed and the questionnaire-version (1–4) as a random variable (all breeds were pooled). The GLIMMIX procedure allows the analysis of fixed and random variables in the same model, and it also allows non-normal data and correlation among the variables. The response variable for fearfulness was Fear-status, for noise sensitivity—Noise sensitivity –status and for separation anxiety we used Separation anxiety –status (referred hereafter as A analysis) (Table 1). Fear –status was determined such that as case dogs we had individuals experiencing fear more than in 40% of situations met (unfamiliar people and situations); correspondingly, control dogs did not show fear towards unfamiliar people, dogs or situations (Table 1). All dogs having Noise reactivity > 0 were assigned to Noise sensitivity cases (Table 1). That means that any dogs showing any fearful reaction, even mild, towards thunder, fireworks and gunshots were considered to belong to the noise sensitivity case group. Often the dog, which is generally fearful, also has noise sensitivity, and/or separation anxiety. Thus, in the A analysis, a case dog for one anxiety may also have other anxieties as well. To investigate those environmental factors that contribute to one specific anxiety-type, we re-analyzed the data. In these B analyses, we included dogs having only one specific anxiety-type (either fearfulness, noise sensitivity, or separation anxiety) as cases, and controls were dogs that had no anxieties (dogs with Fear-status = 0, Noise sensitivity = 0 and Separation anxiety = 0). Finally, we compared whether environmental factors would differ between those dogs that have all three comorbid anxieties (dogs have Fear-status = 1, Noise sensitivity = 0 and Separation anxiety = 1) and between dogs having no anxieties (dogs have Fear-status = 0, Noise sensitivity = 0 and Separation anxiety = 0). This analysis is referred to as C analysis. In all analyses, the following explanatory variables were included as fixed variables in the analysis: sex, size of the dog, age of arrival to a new home (in weeks), socialization (amount of socialization after the puppy has arrived into the new home, in Finland the average age for this is 7–8 weeks, see Table 1), number of children in the household, number of adults, number of dogs in the household, neutering status, kennel/indoor dog, was the dog the owner’s first, second, etc, the time the dog has to spend alone during a normal day, amount of daily exercise (see Table 1), amount of activities done with the dog, dietary supplements (does the dog get any dietary supplements yes/no), type of food (industrial/home made), quality of maternal care, age of separation from the mother, problems at birth (yes/no). Non-significant terms were dropped out individually and only the statistically significant terms were kept. The size of the dog was estimated for each dog as the average breed-specific height of that particular sex of particular breed. Therefore, no height was estimated for mixed breeds dogs. Multicolinearity was assessed in all our final generalized linear mixed models using PROC REG analysis. Factor place of birth (at the dog’s permanent home/ breeder’s place) was not included into the generalized linear mixed model, as although included in all versions of the questionnaire, it included many missing values or “don’t know” answers and therefore including this variable into the model would have reduced the sample size significantly. We therefore analyzed the place of birth–variable separately using Wilcoxon rank sum test, comparing the Shyness and Noise reactivity -scores (see Table 1) between dogs that were born at their permanent home or at the breeder’s place. Kruskall-Wallis test and Spearman correlation were used to investigate the association between daily exercise and other variables.
The data in this study is collected using an online questionnaire. Before completing the questionnaire, owners were informed of the following: “all filled questionnaires are included in the study, which focuses on the phenotypic characteristics of fear and noise phobia, and possible associated environmental factors. All information submitted is kept strictly confidential. Neither you nor your dog(s) will be identified during the study”. By answering the questionnaire, owners thus gave permission to use the data in studying the genetic and environmental factors affecting anxiety-related traits."
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}