Using faecal pellet counts along transects to estimate quokka (Setonix brachyurus) population density

1. What they did

Quokkas are very wary of traps, so counting them by trapping is slow, costly and difficult. The authors tested whether a simpler method, counting quokka scat (faecal pellet groups) along short transects, could estimate how many quokkas live in a swamp.

• Population estimates. At nine swamp sites in the northern jarrah forest of Western Australia, quokka numbers had been estimated by repeated live-trapping and mark-recapture over 2.5 years from 1998 (and by separate annual trapping at one site). Dividing the number of quokkas by the area of swamp vegetation, measured from aerial photographs, gave a density for each site. At three sites trapping was stopped early because no quokkas or evidence of them were found. Six sites had fox control (baiting) and three did not.
• Scat counts. In mid to late winter 2000, when cold weather slows scat decay, one observer (the first author) walked at least 10 transects of 30 to 50 m at each site, from the swamp edge to the centre, with more transects at larger sites. He searched only along the distinctive runways that quokkas make through the dense vegetation, and counted only fresh pellet groups: soft on the outside and green inside when broken. Older pellets are hard and black outside and grey inside.
• Comparison. For each site, the number of pellet groups divided by the number of transects gave a “pellet count value”. They then tested how well this predicted the density from trapping.

2. What they found

• The three sites where quokkas are now thought to be locally extinct (Hoffman, Holyoake and Wild Pig) had no pellets at all.
• Gervasse had the highest density (4.70 quokkas per ha) and Victor Road the highest pellet count value (0.60).
• Across all nine sites, scat counts and density were clearly related (r² = 0.56, P = 0.02), but the fit was only moderate.
• Victor Road, the only unbaited site that still had quokkas, did not fit: it had many more pellets than its density would suggest, and it broke the assumptions of the regression.
• Leaving out Victor Road, or leaving out all unbaited sites, gave an almost perfect fit (r² = 0.98 in both cases).
• The lines were not forced through zero, yet in both models the 95% confidence interval where the line meets the density axis included zero.

Telling quokka scats apart

Three other macropodids partly overlap with quokkas in this forest. The authors did not trust themselves to separate the scats by size and shape alone; they believe the pellets of adult male quokkas cannot be told apart from those of wallabies and young kangaroos. But only quokkas make runways about 15 cm by 20 cm through the dense swamp vegetation. The other species are much taller and do not use the densest parts of the swamps, so pellets of the right appearance found inside a runway can be confidently identified as quokka scats. Tammar wallabies are not known from the study sites, and the reintroduced woylie prefers open habitat and has different scats.

3. What they concluded

• Counting quokka scats along runways is an adequate and reliable way for managers to estimate quokka density over large areas, in any densely vegetated habitat in the quokka’s range.
• It takes far less time, effort and money than trapping, causes the animals far less stress and so may be more ethical, and can be combined with trapping and DNA methods.
• Some caution is needed, because the all-sites model broke a statistical assumption. The authors think this was caused by differences in behaviour: where foxes are not controlled and predation risk is high, quokkas rarely leave the swamp and so leave more scats inside it.
• Wrong identification, missed scats, and differences in how fast scats are dropped and decay can all cause errors. Here the uniform dense habitat, the cold weather and the short survey period are expected to have affected all sites similarly, but future studies should measure scat decay.
• Managers should weigh trapping against scat counts for each survey’s purpose. Adding distance sampling, with estimates of how fast scats are dropped and decay, may improve the method.

4. How tracking was or was not used

The whole method is based on finding and identifying signs. The observer walked transects through the swamps, followed the quokkas’ own runways, and counted scat groups there. Identification relied on the runways (a path made only by quokkas) combined with the appearance of the pellets, rather than on pellet size and shape alone. Pellets were also aged as fresh or old by their texture and colour inside and out. No trailing of individual animals was reported.

5. Tracker or observer certification

Not reported. All transects were walked by the same observer, the first author, who had trapped quokkas at these sites for 2.5 years, but no skill test is reported. Using one observer keeps any personal tendency to miss or misidentify scats the same at every site.

6. Other technologies, skills or innovations

• Mark-recapture estimates (Jolly-Seber method in Program MARK) from paired large and small cage traps baited with apples and “universal bait”.
• Swamp areas measured from aerial photographs with MapInfo mapping software.

7. Implications for conservation (from the paper)

• The quokka is listed as vulnerable (International Union for Conservation of Nature, IUCN, criteria), and its trap-shyness makes trapping surveys slow and expensive. Scat counts offer a cheaper, less stressful way to monitor this threatened species over large areas.
• Fox control may change where quokkas spend their time and so how many scats they leave in the swamps, which needs to be considered when interpreting scat counts.