Testing the precision and sensitivity of density estimates obtained with a camera-trap method revealed limitations and opportunities
St-Laurent, Martin-Hugues; Pettigrew, Pascal; Sigouin, Daniel (2022), Testing the precision and sensitivity of density estimates obtained with a camera-trap method revealed limitations and opportunities, Dryad, Dataset, https://doi.org/10.5061/dryad.g1jwstqqx
The use of camera traps in ecology helps affordably address questions about the distribution and density of cryptic and mobile species. The Random encounter model (REM) is a camera-trap method that has been developed to estimate population densities using unmarked individuals. However, few studies have evaluated its reliability in the field, especially considering that this method relies on parameters obtained from collared animals (i.e. average speed, in km/h), which can be difficult to acquire at low cost and effort. Our objectives were to (1) assess the reliability of this camera-trap method and (2) evaluate the influence of parameters coming from different populations on density estimates. We estimated a reference density of black bears (Ursus americanus) in Forillon National Park (Québec, Canada) using a spatial capture-recapture estimator based on hair-snag stations. We calculated average speed using telemetry data acquired from four different bear populations located outside our study area and estimated densities using the REM. The reference density, determined with a Bayesian spatial capture-recapture model, was 2.87 individuals/10km2 [95% CI: 2.41–3.45], which was slightly lower (although not significatively different) than the different densities estimated using REM (ranging from 4.06–5.38 bears/10km2 depending on the average speed value used). Average speed values obtained from different populations had minor impacts on REM estimates when the difference in average speed between populations was low. Bias in speed values for slow-moving species had more influence on REM density estimates than for fast-moving species. We pointed out that a potential overestimation of density occurs when average speed is underestimated, i.e. using GPS telemetry locations with large fix-rate intervals. Our study suggests that REM could be an affordable alternative to conventional spatial capture-recapture, but highlights the need for further research to control for potential bias associated with speed values determined using GPS telemetry data.
We estimated a “reference density” using a hair-snag genotyping survey and a DNA-based capture-recapture statistical method. We divided the study area into 37 irregular cells of ~7.5 km2 and placed a hair-snag station as close as possible to the center of 33 of these cells; the four remaining cells were discarded due to their inaccessibility and steep slopes. Hair-snag stations were composed of two barbwires fixed on tree trunks 35 cm and 65 cm aboveground to delineate an enclosure of 5 x 5 m. Scent lures were used to attract bears in the hair-snag stations in order to increase the probability of “capture”. Three types of scent lures were used: a mustelid lure in a pierced plastic bottle was hung in the air at the center of the station, while 100 mL of seal oil was applied to a wood pile in the middle of the station (on the ground), and a 1:1 mix of vegetable oil and anise oil was sprayed on tree trunks found within the barbwire section of the station. Lures were refreshed every week. Stations were sampled for 5 weeks, from 6 July to 18 August 2015, as black bears usually increase their movement rate at this time, increasing the probability of recapture and consequently yielding more precise density estimates. Stations were visited weekly and each hair sample was collected individually (i.e. multiple hairs tied on a single barb were considered one sample) and identified to record its location on the barbwire (lower vs. upper barbwire, and barb ID-#). After each visit, the barbwires were burned using a blowtorch to prevent DNA contamination between sessions. Each hair sample was stored in a paper envelope with silica desiccant and freeze-dried for 24 hours. All samples were sent to Wildlife Genetic International (hereafter WGI) for genotyping. Hair samples were subselected based on the “Mowat 1-in-3” method, developed by WGI ; this method offers a good compromise between genotyping many samples that belong to the same individual and missing genetic material from other individuals. DNA was extracted from the samples using QIAGEN DNeasy tissue kits, and the genotyping followed a standard three-phase approach (i.e. first pass, cleanup and error-check), using seven microsatellite markers (G10L, G10H, UarMU23, UarMU50, MSUT-2 and G10X) plus ZFX/ZFY for gender.
We used automated camera to estimate bear density with the Random Encounter Model. The encounter rate was assessed using 47 remote cameras (Spypoint model I-6: n = 13 and model Tiny: n = 22; Reconyx model RM45: n = 4; Moultrie model A-7i: n = 8). All cameras used passive infrared and movement sensors, and the trigger sensitivity was set to its maximum. The cameras were programmed to take three consecutive pictures when triggered, and were randomly distributed within the 7.5-km2 sampling cells by generating 10 random points in each cell; one to five of these points were then selected based on their accessibility (on foot or with an ATV). The radius (r) of the camera detection zone was assessed through several trials during which a person crossed the camera detection zone perpendicularly. The detection arc (θ) of the camera detection zone was assumed to be the value found in the specifications of each camera model. At micro-site scale, the cameras were installed in a direction that provides a relatively clear detection zone, without being faced deliberately towards a trail used by wildlife. We removed branches and stems in front of the cameras to maintain consistency across cameras regarding the detection zone area and to reduce the number of empty pictures triggered by branch movements that trigger the sensors. At each selected point, a camera was fixed on a tree 75 cm aboveground for a minimum of 21 days, leading to the placement of 110 cameras from 1 July to 9 September 2015. The average speed (i.e. movement rate, measured as the Euclidian distance between two successive relocations divided by the time elapsed between them) was not available for the study population (Forillon), so we used GPS telemetry data gathered from four different black bear populations in the province of Québec during the last 19 years (see Table 1 for more details). For each dataset, we calculated average speed (v) and its standard error using only the relocations collected from July to September (i.e. the same period during which of our camera-trap and hair-snag data were collected) with a minimum fix-rate of 8 locations/day (i.e. 1 location every 3 hours) to reduce bias in the estimation of average speed.
Natural Sciences and Engineering Research Council of Canada, Award: Discovery Grant #386661 to MH St-Laurent