Rodent density indices from tracking tunnels, snap-traps and Fenn traps: do they tell the same story?

1. What they did

In New Zealand, footprint tracking tunnels are widely used to gauge how many introduced rats and mice are in an area. A tracking tunnel is a small baited tunnel with an ink pad and paper inside: an animal walking through leaves its footprints, and the index is the percentage of tunnels tracked by a given species. The method had largely replaced kill-trapping, but it had rarely been tested. The authors asked whether tracking tunnels and two kinds of kill traps tell the same story about ship rat numbers.

The study took place on the Puketukutuku peninsula at Lake Waikaremoana, in Te Urewera National Park, as part of a larger project on rats, stoats and the threatened northern brown kiwi. They ran four tests:

• Removal grid (June 1998). A 300 by 300 m grid of 169 snap-traps 25 m apart, with 28 tracking tunnels at 50 m intervals, each tunnel within 1 to 2 m of a trap. Traps were checked daily for five nights, and every rat caught was removed and taken to the laboratory to be weighed, measured, sexed and aged.
• Snap-trap line (1997). 40 snap-traps at 50 m intervals along an existing trap-line, run for three nights every six weeks, with 18 tracking tunnels at 100 m intervals run for one night shortly before each trapping session.
• Tracking tunnels and Fenn traps in tawa-podocarp forest (January 1996 to March 1998). Monthly one-night tracking rates were compared with rats caught as by-catch in Fenn traps set for stoats along predator control trap-lines.
• Tracking tunnels and Fenn traps in beech forest (same period). A second tunnel line at the far end of the peninsula was compared with nearby Fenn trap-lines, and the two forest types were compared by each index.

2. What they found

Removal grid

• 121 ship rats and no house mice were caught in five nights. Males and females, and juveniles and adults, were caught at similar rates; most rats were immature.
• The minimum density was 7.1 rats per hectare, and the removal estimate was 8.2 rats per hectare (95% confidence interval 5.9 to 10.53).
• Tracking rate followed the falling number of rats left on the grid (72% of the variation explained), but the relationship was not significant (P = 0.07). It was heavily shaped by the last night, when the estimated density was 1.25 rats per hectare and the tracking rate was 19.8%.
• Captures on the edge of the grid rose through the five nights, to 89% on the final night, showing that rats were moving in from outside. This breaks one of the assumptions of the removal method.
• One tunnel recorded mouse prints on the last night, though no mice were caught.

Snap-trap line

• 75 ship rats and 7 house mice were caught over 1997.
• Tracking rate was significantly related to the number of rats caught per session, but not to captures per 100 trap-nights.

Tracking tunnels and Fenn traps

• In tawa-podocarp forest, where rats were at moderate to high density, the two indices were significantly correlated. The tunnels showed bigger swings and stayed high for longer in 1996. Both indices showed a significant difference between 1996 and 1997.
• In beech forest, the two indices did not agree. The tunnels showed a peak in March to May 1996, a low in March to April 1997, and a new peak in March 1998. The Fenn traps showed no 1996 peak and stayed low until captures rose from June 1997. There was no long-term agreement between the two.

Comparing the two forests

The answer depended on the index. Tracking tunnels showed significantly more rats in beech forest in 1997. The number of rats caught per month was significantly higher in tawa-podocarp forest in both years. Captures per 100 trap-nights were higher in tawa-podocarp forest in 1996, but significantly higher in beech forest in 1997.

3. What they concluded

• Ship rat density indices varied a great deal with the type of index and the study design, so the reliability of tracking tunnels needs closer study before decisions rest on them.
• Where rats are at medium to high density, changes in tracking rate appear to follow changes in rat numbers.
• Tracking tunnels are particularly sensitive to how active rats are, because one rat can track several tunnels but can only be caught once in a kill trap. Rats in low-density populations may be more active and range more widely, which could explain the poor agreement in beech forest. Rodent behaviour must be considered whenever tracking tunnels are used.
• Mice were probably present at very low density but may have avoided the tunnels, or been kept out of them, while rats were present.
• The authors recommend four steps: (1) compare populations with tracking tunnels only within the same habitat type, with a consistent protocol, running treatment and non-treatment areas on the same nights; (2) run enough tunnels to detect a difference, ideally checked with a power analysis before the study begins; (3) consider the animals’ behaviour: with male home ranges of around 1 ha, the common 50 m spacing lets one rat track several tunnels, and 100 m spacing may give a more reliable index; (4) always pair tracking tunnels with a second density index.
• The relationship between the commonly used indices and true rodent density needs urgent attention.

4. How tracking was or was not used

The whole study tests a tracks-based index. Rats left their footprints on inked paper in baited tunnels set overnight, and each tunnel was scored as tracked or not tracked by a given species, so the prints had to be identified to species (rat or mouse). The index was the percentage of tunnels tracked in one night. No trailing was involved. The paper does not describe how prints were identified or checked.

5. Tracker or observer certification

Not reported. The paper does not say who read the tracking papers or how prints were told apart, though the mouse prints recorded on the grid show that species identification was part of the work. Six named people and the Lake Waikaremoana Conservation Corps are thanked for help collecting the Fenn and snap-trapping data.

6. Other technologies, skills or innovations

• Footprint tracking tunnels (King and Edgar 1977 design) baited with peanut butter.
• Ezeset Supreme snap-traps under plastic covers to keep out non-target animals.
• Mk 4 and Mk 6 Fenn kill traps under wooden covers, baited with hen eggs or rabbit meat, from an existing predator control programme, with the rat by-catch used here as a density index.
• Ageing rats by reproductive condition and tooth wear.
• Time-series methods (stationarity tests and co-integration) to compare two monitoring records that both change over time.

7. Implications for conservation (from the paper)

• Research and management decisions in New Zealand often rest on tracking tunnel indices of introduced rodents, so their reliability, repeatability and applicability need further testing.
• Tracking tunnels should only be used to compare relative abundance within similar habitat types, and should always be backed by a second density index.
• Confidence in these indices can only grow by calibrating them against true density estimates, including the effects of index type, activity, catchability, tunnel placement and sampling effort.