Wildlife species and trash interactions. The Leatherback Trust (C)

Description

With the constant urban growth, food subsidies production increases, consequently attracting opportunistic wildlife species. Food subsidies favor opportunistic species proliferation, resulting in strong competition and/or predation on other species. High availabilitity of these resources near protected areas, can threaten the target species conservation. In Costa Rica, within Parque Nacional Marino Las Baulas buffer zone, urban development has growth exponentially in recent years, resulting in Northern Raccon population growths. Raccoons started predation on endangered sea turtles within the National Park, predating 100% of all Olive Ridley (Lepidochelys olivacea) nest in a single season. Given these negative effects, we used camera traps to assess diferent trash containers in the buffer zone of the National Park and determine their effect on wildlife species attraction. Also, we assess the behaviour of these species in relation to trash collection dynamics to generate management recommendations. Our hipothesis was that depending on the desing/material of the trash container and trash dynamics, wildlife species can increase their access success and feed on Food Subsidies inside the containers. This data base contents the raw data on the observations made with camera traps during the study period.

Steps to reproduce

To identify and characterize opportunistic predator behavior, we deployed infrared automatic camera traps in 24 distinct garbage containers, between July and October 2021. Each camera trap was attached to a tree at a height of ~40 cm pointing toward the garbage container. Camera traps were set to be active via passive infrared sensor and video mode, with a minimum delay of 30 seconds between consecutive triggers. Camera traps were only set to be active at night (between 17:00-06:00). After one week of data collection around a particular container, the camera trap was moved to a different location. We identified species that interacted with garbage and the times during the night when they were actively foraging. We estimated access success to the container by dividing the number of times the animal successfully entered the garbage container by the total number of attempts it made to get in or if they removed garbage from the containers. We considered each video as one detection, independently of the group size. To ensure that each video corresponded to a different event, we considered triggers within 60 s of each other as one detection, following Kays et al. (2020). We assessed the influence of location, day of the week, and type of container on group size (number of individuals of a species per group) and number of independent detections per night. We conduted a Generalized Linear Model (GLM) with a link log function, as customary for count data, with a negative binomial error distribution family due to data overdispersion. We used the Tukey HSD post-hoc test to compare group sizes and the number of detections between days of the week, type of containers, and locations. Since raccoons were the predominant species, and the number of detections of the other species was relatively low and occasional, we did not include other species in this analysis. To assess the influence of location, day of the week, and type of container on the trash access success for all opportunistic species, we used a GLM with binomial error distribution. Success was treated as binary data (success/no success). We used a multiple-comparison analysis with a Tukey HSD post-hoc test, to compare access success between locations, days of the week, and type of containers. In all models, we used the logarithm of the number of camera trap nights as an offset to standardize counts to account for variation in sampling effort at the camera trap location (Montalvo et al., 2019). We used the AICc values to determine the best-fit model. We tested for deviance of the model using null and residual deviances to verify its adjustment. We tested for data Overdispersion using the Goodness-of-Fit Model with the aods3 R package (v0.4.1.2; Lesnoff & Lancelot, 2022). The number of detections per hour was plotted using the Circular Statistics R package (v0.5.0; Agostinelli & Lund, 2023). All statistical tests were conducted with an α = 0.05 in R Statistical Software (v4.3.2; R Core Team, 2023).

Categories

Funding

Licence