Comparing methods for sampling large- and medium-sized mammals: camera traps and track plots

1. What they did

Monitoring wildlife is usually limited by money, so choosing an efficient method matters. The authors compared two common, non-invasive ways of surveying medium and large mammals: track plots (patches of ground where footprints are recorded) and camera traps. They asked how well each detects mammals, whether either favours animals of a certain size, and what each costs.

At Jataí Ecological Station in São Paulo State, Brazil, they made ten track plots, each 1.5 m long across a dirt road about 4 m wide, from the road’s own sandy soil, spaced every 700 m. A camera trap with an infra-red motion sensor and a film camera faced each plot, so each pair formed one sampling unit. For ten days in July 2006 (the dry season), they checked every plot and camera once a day, noted footprints and photos, and then cleaned the plots. The same species in the same plot or camera on the same day counted once. Two researchers identified the footprints on site.

They also estimated the day-by-day costs of each method for a 10-day and a 30-day survey, counting only the costs that differ between methods (equipment, film, batteries, fuel and daily allowances for staff).

2. What they found

Records

• Track plots gave 28 records of six taxa; camera traps gave 17 records of six species (Table 1 lists the same six taxa for both).
• Puma: 6 by track plot, 8 by camera. Maned wolf: 6 and 4. Brocket deer: 2 and 1. Striped hog-nosed skunk: 6 and 2. Armadillo: 4 and 1. Tapeti: 4 and 1.
• The two methods sampled the species differently (Wilcoxon test).
• Detection rates were 0.117 records per hour for track plots and 0.071 for camera traps, so track plots detected about 1.65 times more.

Body size

• Camera trap detections depended on body mass: the larger species (puma, maned wolf, brocket deer) were recorded well by cameras, the smaller ones much less often.
• The authors read the Chi-square results as showing the two methods work similarly for larger animals but differently for smaller ones.

Identification

• Footprints of brocket deer and armadillos could only be identified to genus, because more than one species of each lives in the area.
• Some days the soil was so dry that even easily recognised footprints, such as puma, could not be securely identified. Wind and rain can also wipe out tracks.
• A photographed mammal can easily be identified to species, sometimes to individual, and the photo records the time.

Costs (per day, variable costs only)

• 10-day survey: camera traps US$305.13, track plots US$149.90, so cameras cost about 2.04 times more.
• 30-day survey: camera traps US$101.71, track plots US$145.89.
• Cameras are mostly an up-front cost and can be left for up to a month, so their daily cost falls over time. Track plots need the researcher in the field every day to identify footprints, so their daily cost stays about the same.

Camera problems

• Cameras took about one second to fire, so some animals may have left before the photo.
• On hot days, when the air is close to body temperature, the sensor works less well; most detections came at night. The national camera traps coped better with heat and humidity than the imported ones.
• Camera shyness was seen in the authors’ 18-month study in the same region, but not in this short one.

3. What they concluded

• Track plots are good enough for quick surveys of local species richness on a limited budget.
• Camera traps cost more at first and favour larger animals, but give precise species identification, activity patterns and sometimes population density. They suit solitary, wide-ranging or low-density animals and longer studies.
• Used together, the two methods complement each other and improve data quality.

4. How tracking was or was not used

Footprints in sand track plots on a dirt road were recorded and identified to species or genus on site each day, then the plots were swept clean. Each species was counted once per plot per day. No trailing was reported. The authors note that this method depends strongly on the researcher’s skill at identifying footprints, that soft ground makes the same animal’s footprint look larger, that dry, hard ground and wind or rain spoil tracks, and that similar species (such as the two brocket deer) are hard to tell apart.

5. Tracker or observer certification

Reported but not validated. The footprints were identified by “two skilled researchers with large experience in the task”. Their experience was not validated by any certification or test, and the paper does not say whether they agreed on identifications.

6. Other technologies, skills or innovations

• Infra-red motion-sensor camera traps with 36-exposure film cameras (national make, Trapa-camera), covering 90° horizontally, 30° vertically and up to 8 m, with a delay of about one second.
• A comparison of a Brazilian camera trap with an imported Stealth Cam MC2-GV, which was harder to calibrate in hot weather and less well sealed.
• A simple cost model separating fixed and variable expenses.

7. Implications for conservation (from the paper)

• Knowing how well survey methods work and what they cost helps researchers plan conservation strategies and monitoring programmes.
• The best method depends on the aim, time and budget: track plots for quick richness surveys, camera traps for longer studies and accurate identification. Combining methods gives better data, especially for rare or secretive species.