The impacts of camera placement on species detectability in an Indonesian terrestrial community
Ecologists have increasingly favoured the use of camera traps in studies of animal populations and their behaviour. Because camera trap study design commonly implements non-random selective placement, we must consider how this placement strategy affects the integrity of our data collection. Selective placement of camera traps have the benefits of 1) maximizing the probability of encounter events by sampling habitats or microhabitats of known significance to a focus or closely-related species and 2) reducing data collection and maintenance effort in the field by situating cameras along more easily-accessible landscape features. Introducing a non-random survey method, such as selective placement, into a project studying a species or community that also expresses non-random habitat use may lead to unintentionally biased data and inaccurate results. By using a paired on-trail/off-trail camera-trap study design, my aim is to investigate potential differences in popular ecological indices, species detection probability (p) using multi-method occupancy models, and intraspecific temporal activity for a terrestrial community in Gunung Palung National Park in Indonesian Borneo. Differences in detection probability between on and off-trail cameras were compared against species characteristics (including body size, diet, and taxonomic group) to find potential correlations. While several species exhibited a significant difference in detection probability between cameras placed on foot trails and those placed randomly off-trail, there was no measured community trend. This stresses my conclusion further that a non-random study design leaves results open to bias from unknown patterns in detection due to underlying variation in behaviour and microhabitat use. Selective placement may be effective for increasing detection probability for some species but can also lead to substantial bias if the features selected for are not explicitly taken into account within the analysis or balanced with a control in the study design. In addition, a positive interactive effect was found between on trail species detection and body size for the terrestrial omnivore guild, and three species presented significant variation in temporal activity between camera placement types. This provides evidence that camera placement not only affects species state parameters and indices but has a noticeable impact on behavioural observations that require accountability as well.