Long-term use of non-invasive sampling methods: does successful sampling of brown bears by hair snares and camera traps change over time?

Context Non-invasive sampling methods are widely used by ecologists to collect animal hair, images, tissue or signs. Sampling devices are imperfect, and collection success may vary over time owing to behavioural changes in study organisms or other factors. If collection success decreases, the utility of non-invasive sampling devices for longitudinal studies that rely on consistency may be compromised. Aims Our primary objectives were to evaluate whether collection success of brown bear (Ursus arctos) hair by using hair snares and camera traps changed over time, and whether hair- and image-collection success was influenced by bear activity around the sampling site. Methods We paired non-invasive sampling by hair snares with motion-activated cameras at six streams in Alaska over 4–6 years, so as to evaluate how often brown bears left samples on wires or were photographed by cameras, and whether this sampling success changed over time. Changes in sampling success were evaluated in the context of bear activity per sampling period as determined by camera data (number of bear–wire encounters) or hair snare (number of barbs with hair); genetic analyses allowed us to evaluate whether the same bears were sampled repeatedly. Key results Overall, hair was collected in 78% and images in 73% of 2-day sampling periods when bears visited sites, and we observed no substantial change in the probability of successful sampling over time at 11 sites. The number of bear–wire encounters was positively correlated with the number of hair samples collected, as would be expected if sampling rates remained constant over time, and individual bears with previous wire experience were sampled in multiple years. Conclusions Overall, the results indicated that sampling success by using hair snare and camera trap showed substantial interannual variability, but changes over time were not consistently identified across sites. Among-site variation in sampling success highlighted the importance of accounting for site-specific differences in sampling success, and neither method sampled unfailingly. Implications Sampling by wires and cameras remained effective over time, suggesting that these non-invasive sampling methods may be successfully employed in long-term studies.

Estimation of wolf population density using spatial capture-recapture: refining methods for monitoring cryptic species

Wolves (Canis lupus) in Southeast Alaska have been proposed for listing under the U.S. Endangered Species Act first in 1993, and more recently in 2011. Reports of declining wolf populations sparked concern, in addition to high rates of logging and broad-scale succession patterns predicted to negatively impact Sitka black-tailed deer (Odocoileus hemionus sitkensis), the primary ungulate prey species of wolves in Southeast Alaska. Given the recurring interest of wolf viability in the region, and in order to manage wolves and their prey sustainably, it is imperative to obtain regular and reliable population estimates. However, monitoring wolves in temperate rainforests is challenging because the landscape obscures visibility and lowers success of traditional methods such as aerial surveys and radio collar mark-recapture. We used hair snares to collect DNA samples and spatially-explicit capture-recapture (SECR) models to estimate fall wolf density during 2012–2015 on northcentral Prince of Wales Island (POW), Alaska. We incorporated covariates including sex, behavioral responses, and site-specific changes in effectiveness of detection probability by fitting hybrid mixture models to the data. We also incorporated into our models landscape variables including forest habitats in various management conditions and succession stages to relate to wolf habitat selection. We concurrently implemented a traditional approach for comparison to the DNA-based SECR method using radiocollared wolf data to estimate population abundance with minimum counts and the size of wolf packs and pack territories. The results of the DNA-based SECR method proved to be more reliable, efficient, cost-effective, and robust than the traditional method, which was sensitive to violations of model assumptions. Our efforts to improve SECR density estimate precision by increasing the hair sampling intensity and area resulted in more wolf hair detections and redetections, and increased the number of unique wolves redetected. Based on multiple lines of evidence, we report a decline in wolf population abundance over the past 2 decades in northcentral POW. We conclude that DNA-based SECR is an effective tool for regular population monitoring, as is required in situations of elevated concern for the persistence of a population, and may simultaneously provide information on heterogeneous landscape use, an important wildlife management consideration in fragmented forests.

Estimation of wolf population density using spatial capture-recapture: refining methods for monitoring cryptic species

Wolves (Canis lupus) in Southeast Alaska have been proposed for listing under the U.S. Endangered Species Act first in 1993, and more recently in 2011. Reports of declining wolf populations sparked concern, in addition to high rates of logging and broad-scale succession patterns predicted to negatively impact Sitka black-tailed deer (Odocoileus hemionus sitkensis), the primary ungulate prey species of wolves in Southeast Alaska. Given the recurring interest of wolf viability in the region, and in order to manage wolves and their prey sustainably, it is imperative to obtain regular and reliable population estimates. However, monitoring wolves in temperate rainforests is challenging because the landscape obscures visibility and lowers success of traditional methods such as aerial surveys and radio collar mark-recapture. We used hair snares to collect DNA samples and spatially-explicit capture-recapture (SECR) models to estimate fall wolf density during 2012–2015 on northcentral Prince of Wales Island (POW), Alaska. We incorporated covariates including sex, behavioral responses, and site-specific changes in effectiveness of detection probability by fitting hybrid mixture models to the data. We also incorporated into our models landscape variables including forest habitats in various management conditions and succession stages to relate to wolf habitat selection. We concurrently implemented a traditional approach for comparison to the DNA-based SECR method using radiocollared wolf data to estimate population abundance with minimum counts and the size of wolf packs and pack territories. The results of the DNA-based SECR method proved to be more reliable, efficient, cost-effective, and robust than the traditional method, which was sensitive to violations of model assumptions. Our efforts to improve SECR density estimate precision by increasing the hair sampling intensity and area resulted in more wolf hair detections and redetections, and increased the number of unique wolves redetected. Based on multiple lines of evidence, we report a decline in wolf population abundance over the past 2 decades in northcentral POW. We conclude that DNA-based SECR is an effective tool for regular population monitoring, as is required in situations of elevated concern for the persistence of a population, and may simultaneously provide information on heterogeneous landscape use, an important wildlife management consideration in fragmented forests.

Complementary use of motion-activated cameras and unbaited wire snares for DNA sampling reveals diel and seasonal activity patterns of brown bears (Ursus arctos) foraging on adult sockeye salmon (Oncorhynchus nerka)

The seasonal and diel movements of predators to take advantage of shifts in prey availability are fundamental elements of their foraging ecology, and also have consequences for the prey populations. In this study, we used complementary noninvasive techniques (motion-activated cameras and hair snares) to investigate seasonal and diel activity of brown bears (Ursus arctos L., 1758) along six proximate streams supporting spawning populations of sockeye salmon (Oncorhynchus nerka (Walbaum in Artedi, 1792)) in southwestern Alaska. Camera records over 3 years showed a rapid increase in bear activity around the time salmon arrived in the streams, with differences among streams corresponding to differences in salmon phenology. Bears were active throughout the day and night, but there were clear crepuscular peaks when camera data were pooled. When wire snares (to collect hair samples) were paired with cameras, the data showed similar seasonal patterns, but each technique detected bears missed by the other. Roughly equal numbers of bears left hair but no camera image, and images but no hair, at paired sites. Taken together, the results indicated a close correspondence between bear activity and salmon timing, differences in diel timing among streams, and the complementarity of data obtained by motion-activated cameras and hair snares.

Sniffing out the stakes: hair-snares for wild cats in arid environments

Context Wild cats (Felis spp.) are difficult to monitor because of their cryptic lifestyle and usually low numbers. Hair-snaring is a promising non-invasive method being used increasingly to estimate mammal populations. Aims Our aim was to carry out pilot trials of a simple hair-snare designed to capture hair from wild cats in arid environments. Methods Roughened wooden stakes were set at multiple sites on the crests of sand dunes and in swales in western Queensland, Australia, and in mostly sandy habitats of the Namib and Kalahari Deserts, Namibia. In Australia, stakes were sprayed with cat urine, extracts of catnip or valerian herbs as lures, or left untreated; in Namibia, alternate stakes were sprayed with a food lure of tuna emulsion oil. The stakes were checked for hair, usually daily, for 2–14 days, and the surrounding ground was inspected for tracks. Remote cameras also were used at some sites to confirm the identity of visitors to stakes. Key results In Australia, feral cats (Felis catus) were attracted to, and left hairs on, stakes sprayed with cat urine six times more frequently than to unsprayed stakes irrespective of whether snares were on dune crests or in swales, and showed no response to catnip or valerian. Tracks and photos showed that cats, dingoes or wild dogs (Canis lupus ssp.) and foxes (Vulpes vulpes) also approached and sniffed the stakes. In Namibia, F. catus, F. lybica and F. nigripes left hair on stakes, with deposition rates two and a half-fold higher at stakes with the food lure than without it. At least five other species of predators visited the hair-snare sites. Conclusions Simple wooden stakes provide a cheap and simple method of snaring hairs from wild cats, especially if used in conjunction with appropriate lures. Our results broadly support previous work, and extend the utility of the method to different Felis spp. in arid habitats. Implications Further research is needed on snares to investigate the seasonal efficiency of different lures. If DNA also is to be extracted to identify individuals, more work is needed to confirm that snares yield hair of sufficient quality to allow this.