SNAMP Pub #25: Testing the consistency of wildlife data types before combining them: the case of camera traps and telemetry
Article Title: Testing the consistency of wildlife data types before combining them: the case of camera traps and telemetry
Authors: Viorel D. Popescu, Perry de Valpine and Rick A. Sweitzer
Wildlife data gathered by different monitoring techniques is often combined to estimate animal density. However, efforts to combine data types have in common the assumption that different data types are inherently consistent with each other and that they are measuring the same quantities, without any biases; and that therefore information from one data type can be used to predict responses in the other. This assumption has gone largely untested and, in general, should not be taken for granted. Thus, there is a need for investigating how different data types are related and testing whether they can be combined without creating biases. In this study, we ask whether information from either camera trap capture or telemetry datasets can predict responses in the other.
Extensive camera trap and telemetry data were gathered simultaneously for the Pacific fisher, a forest carnivore that is now uncommon and rare in the western part of its range in North America. The study area was the non-wilderness region of the Bass Lake Ranger District in the Sierra National Forest, near Oakhurst, California, and covered over 450 square miles (roughly1,150 square km). The study area is part of the Sierra Nevada Adaptive Management Project, which was formed to evaluate the impact of forest thinning on Pacific fisher among several other disciplines.
Our objectives were to provide insight into camera trap effectiveness and whether camera traps measure space use consistently with telemetry data. We asked two specific questions about the relationship between these types of data: (1) how does the number of telemetry relocations near camera traps predict camera detection probability, including differences between sexes and seasons and (2) does home range use estimated from telemetry data adequately predict camera detection probability?
We hypothesized that males are more likely to be detected at camera traps compared with females due to their larger home ranges. Because the camera traps were baited, we also hypothesize that detection probabilities would be higher during the winter season, when food is scarcer. Finally, we hypothesized that there may be differences among animal’s ability to be detected due to differences in their ability to discover camera traps and/or habituation to specific traps.
We only used 2 years of telemetry data from the larger study: October 1, 2008–September 30, 2009 (Year 1) and October 1, 2009–September 30, 2010 (Year 2). During this period, 52 fishers (32 females and 20 males) were radio-tracked, of which 19 females and 7 males were tracked during both years.
We first calculated the collared animal’s home range, both annual and seasonal. We then identified the locations of camera traps within their individual home ranges. Second, we determined which camera traps were available and when for each collared fisher by examining the overlap between the camera trap deployment windows and the period each fisher was actively radio-tracked. Then for each collared fisher and available camera, we extracted the number of telemetry locations within 250 and 500 meter circular neighborhoods centered on camera trap locations.
Our results showed that the number of telemetry relocations within 250 and 500 meters of camera traps predicted detection probability well during both years with differences based on season and sex. During the second year, all fishers were more likely to be detected during the fall/winter season, consistent with the hypothesis that animals are more likely to be captured at baited cameras during winter. Given the same utilization density males are more likely than females to find a camera given their larger home range and higher movement rates. Our models suggest that fishers have a weakly supported tendency to visit cameras more frequently in the central part of a home range.
Our results provide fundamental support for use of camera trap data to estimate animal space use by demonstrating that such data are largely consistent with more detailed telemetry data in the case of fishers in the Sierra Nevada. It is reasonable to conclude that in our case, camera trap data do reflect space use in a manner consistent with telemetry data. Our findings highlight several cautions for use of camera trap data only; to include possible biases in detections by sex, season, year and ecological conditions that make the case for integrating other sources of data, such as telemetry. In addition making use of complementary information from different data sources is a common theme in recent advances in statistical modeling in ecology.
In conclusion, our research represents the first empirical test for reconciling space use by animals using data gathered simultaneously using different sampling techniques and provides support for using camera traps for estimating space use with some cautions.
Popescu, Viorel D., Perry de Valpine and Rick A. Sweitzer. 2014. "Testing the consistency of wildlife data types before combining them: the case of camera traps and telemetry." Ecology and Evolution. 4(7):933–943.
The full paper is available here.
For more information about the SNAMP project and the Fisher team, please see the: Fisher Team Website.