whole body odor data.xlsx

Fox presence and food consumption.<br>To obtain the olfactory cue corresponding to the whole body odor of maremma sheep dog, we follow the methodology described by Leo et al. (2015), which consists of using white cotton towels of 700 gr/cm2 that were left for a month in the resting place of a set of maremma dogs, allowing the towels to be impregnated with skin oils, dandruff, sweat, saliva, anal excretions, remains of urine and feces. Then, towels were stored in airtight bags at -20 ° C until their use in the field. We used citronella oil (Cymbopogon winteranius) as a second olfactory treatment. This essence was sprayed on white cotton towels and subjected to the same treatment previous to store them until their use in the field. Control treatment consists of clean towels of the same material and stored in the same way. In the field, we stablish 9 experimental stations 400 m apart. At each station, towel pieces of 12 x 12 cm were placed inside a white plastic dispenser with holes in the base and adhered to a 60 cm wood pole, one pole placed in each cardinal point of the station, 5 meters away from a central container with food. We used only one odor treatment per station and they were randomly assigned every day, with 3 replicates for each treatment including controls. All experimental stations remained without olfactory cues and free food access during the first 48 hours to allow the habituation of foxes, as there is evidence of neophobic behavior in Lycalopex (Travaini et al., 2013). On the third day, towels with the odor treatments were placed, no station got the same treatment two nights in a row to avoid spatial habituation. A trap camera was placed pointing to each food container to record the visit of foxes during 24 hrs. <br>As an experimental approach to measure the repellency effect of odor treatments, we establish foxes giving up density (GUD) in presence of citronella oil, maremma whole body odor and controls. For this, each experimental station had a food source (plastic container staked to the ground) containing 500 g (646 pellets) of Pro Plan® adult food Medium breeds, which corresponds to twice the suggested portion for wild animals, presuming they could have a greater energy expenditure than that of domestic animals. Food was renewed daily for a total of 6 days; during that period food left behind on the previous night in each station was collected, counting the number of remaining pellets to stablish GUD.<br>To assess if habitat characteristics could influence the repellency effect of odor treatments, we used vegetation cover as an approximation. We measured the horizontal vegetation cover surrounding each experimental station around the food container and we estimated the total cover by shrubs and trees. For the vertical cover, a rod of 3 meters high was placed every one meter along the 5 meters of the tape (in the 4 cardinal points) and then measured the height of the vegetation touching the rod in centimeters.<br>For ethical reasons, this study was carried out in areas within the farm where sheep were recently excluded, to avoid attracting predators to sectors being used by sheep during the trial.<br>Statistical analysis.<br>We used Kruskal-Wallis test to stablish if there were differences in the presence of foxes associated to odor treatments, which we consider as the number of times a fox was seen on camera trap with at least 30 minutes apart. Then we used GLM with binomial distribution to stablish if the vegetation cover had any effect on the fox presence in the experimental stations.<br>To test if there was a repellency effect from maremma whole body odor or citronella oil reducing the number of pellets consumed by foxes compared to the control, we used the count of the number of pellets consumed as an approximation of GUD and zero-inflated Poisson (ZIP) models to include all 9 stations. We used the number of pellets eaten by foxes as the dependent variable, the fixed effect was the odor treatment (control, citronella oil, livestock guardian dog), and horizontal and vertical vegetation covers were included as covariates. Finally multi-model inference was used to assess the most plausible model explaining the change in food consumption.<br>Fox behavior. Based on Leo et al. (2015), we used the video recordings of fox visits to analyze the effect of odor treatments in the foraging behavior and risk assessment behavior of foxes. We constructed an ethogram (Table 1) and used it to analyzed all videos with fox presence with JWatcher software (Blumstein &amp; Daniel, 2007) were all scored behaviors were treated as mutually exclusive as suggested by Blumstein &amp; Daniel (2007), data was extracted as the mean proportion of time in sight an individual allocated to each behavior.<br>Statistical analysis.<br>PCA analysis was used to select the number of variables that explained most of the variation. Finally, we used Kruskal-Wallis stablish differences in foraging and risk assessment behavior between odor treatments. All analyses were conducted in the computer program R (R Development Core Team 2013).