Non-Invasive Methods to Assess Co-Occurrence of Mammalian Carnivores

1. What they did

Carnivores are wide-ranging and hard to see, so researchers often rely on methods that do not need the animals to be caught or seen. Most models of where species live also look at one species at a time and ignore how species affect each other. The author set out to compare four such non-invasive methods and to see whether pairs of carnivore species avoid each other.

• Sites. 82 sites of 4 hectares (200 by 200 m) in forest on the North and South rims of Grand Canyon, at least 2 km apart, surveyed for two seasons on each rim between 2003 and 2005. Each site was visited three times per season, about 14 days apart.
• Transect searches. One observer walked four 200 m transects (800 m in all) per site, looking for scats, tracks, burrows and other evidence of carnivores. All evidence was measured, photographed and identified to species by size and appearance, using two field guides. Scats were collected.
• Cameras. One baited, infrared-triggered film camera per site.
• Hair traps. Baited with a scented lure, with an aluminium pie tin hung nearby to draw animals in.
• Track plates (from 2004). Aluminium plates with sticky shelf paper in the centre and a chalk and ethanol mixture on the ends, inside a plywood box, baited with cat food, added to pick up smaller carnivores. Each track was measured with calipers and matched to a published key.
• DNA. Scats and hairs were identified to species in the laboratory (cougar, coyote, bobcat and gray fox).
• Analysis. Occupancy models estimated how likely each method was to detect each species on one visit, how likely each species was to use a site, and whether pairs of species were found together more or less often than expected by chance.

2. What they found

Species detected

• Nine of the 11 carnivore species thought to live in the park’s forests were detected. American black bears and ringtails were not.
• All nine were detected by at least two methods, but raccoon, striped skunk, western spotted skunk and long-tailed weasel were each detected on five or fewer visits.

The four methods

• Transect searches gave the most detections; most were scats (77 scat detections). Only 65% of scats (50) were identified to species in the laboratory, and 62% of those were coyote.
• Transects and track plates detected the greatest number of species, especially smaller carnivores.
• Cameras detected coyotes most often and also recorded rarer species, including American badgers and both skunks. Photos gave unequivocal identifications.
• Hair traps gave the second-highest number of detections (38), but DNA failed for 61% of hair samples. The hair identified came from coyote (47%), gray fox (47%) and bobcat (7%), even though the lure was meant to attract cats.
• Per visit, cameras were the most likely method to detect coyotes (0.200) and gray foxes (0.240); track plates were most likely to detect bobcats (0.152).
• Using all four methods over eight sampling occasions, the chance of detecting a species at a site where it was present was 0.589 for bobcats, 0.623 for coyotes and 0.633 for gray foxes.

Use of habitat

• Modelled use of habitat was on average 71.5% higher than the observed detection rate on the North Rim and 64.6% higher on the South Rim.
• Bobcats and American badgers were detected more often on the North Rim; coyotes and gray foxes on the South Rim.
• Distance to a public road mattered most: coyotes and bobcats were more likely to use sites farther from roads, and gray foxes sites closer to roads. Distance to water mattered for no species.

Who avoided whom

• North Rim: coyotes and bobcats never occurred at the same site (species-interaction factor of zero), suggesting complete avoidance or exclusion.
• South Rim: coyotes and gray foxes occurred together less than half as often as expected if they were independent (factor 0.401).
• No avoidance was found for the other pairs, though the uncertainty around those estimates was high.

3. What they concluded

• No single non-invasive method is enough to survey a whole carnivore community. Pairing transect searches with more cameras and track-plate stations may be the most reliable and efficient approach in western North America. Scat-detection dogs could make transect searches more efficient.
• Hair traps were ineffective compared with the other methods.
• Results based only on detections underestimate the area a species uses, so imperfect detection needs to be accounted for.
• Co-occurrence and habitat should be modelled together, because responses to things like roads can make species seem to meet more or less often.
• Surveys should identify pairs of species likely to interact before fieldwork, collect data on co-occurring species in the same way as habitat data, and keep differences between sites small.

4. How tracking was or was not used

Tracks and signs were central to one of the four methods. On each transect, one observer searched for scats, tracks, burrows and other carnivore evidence, and measured, photographed and identified each sign to species from its size and appearance, using Murie (1975) and Halfpenny (1986). Scats of black bears, raccoons and skunks were assumed to be identifiable by appearance; scats of the four DNA-target species were confirmed in the laboratory. Track-plate prints were measured with dial calipers and matched to a published key (Taylor and Raphael 1988). No trailing was reported.

5. Tracker or observer certification

Not reported. The paper says one observer walked each transect and that one to two field technicians helped, but gives no information on their tracking experience. Sign identifications other than DNA-confirmed scats depend on this observer’s skill.

6. Other technologies, skills or innovations

• Infrared-triggered 35 mm film cameras, baited with a pork rib or chicken leg
• Hair traps with a scented lure (beaver castoreum, catnip oil and cougar or bobcat urine)
• Track plates (aluminium, sticky shelf paper and carpenter chalk) in a plywood box
• Species identification from scat and hair DNA (PCR and RFLP)
• Occupancy and two-species co-occurrence models in PRESENCE 3.0

7. Implications for conservation (from the paper)

• Measuring where species do and do not occur together matters for biodiversity monitoring and conservation.
• Surveys designed around the whole carnivore community, rather than one species, give more accurate estimates for every species.
• The survey was not intensive enough to show whether undetected species (black bears, ringtails) are truly absent; the author calls for more research on these species and for inventory data to help manage Grand Canyon National Park.