Terrestrial mammal responses to edges in Amazonian forest patches: a study based on track stations

1. What they did

Brazil’s Amazon forest is being cut into patches surrounded by cattle pasture. The edges of these patches change the forest, but little was known about how large ground-dwelling mammals respond to them, because these animals are rare, secretive and hard to study. The authors used sand track stations to find out.

• Where. Around Alta Floresta, Mato Grosso, Brazil, they surveyed 17 forest patches (5 to 4714 ha) and two areas of continuous forest, grouped into four size classes: small (under 50 ha), medium (80 to 400 ha), large (over 900 ha) and continuous forest.
• Which animals. Five plant-eating mammals: two large rodents (red-rumped agouti and paca) and three hoofed mammals, or ungulates (gray brocket deer, collared peccary and lowland tapir).
• How. They cut 49 transects, each 150 m long, straight into the forest from the pasture edge. Along each one they set 10 sand track stations, from the edge (0 m) to 150 m in. For each station they cleared a 1.5 by 1.5 m square and spread about 6 kg of sand over a 0.75 by 0.75 m area: coarse sand below and a fine layer on top so tracks showed clearly. Each station was baited daily with a slice of banana.
• Checking. Every day, local hunters who knew the tracks of all the species helped identify each track; identifications were then double-checked against field guides. The sand was smoothed and re-baited. Stations ran for 10 days per transect (with a two-day break in the middle), giving 4900 track station days in total. They watered the sand in dry weather and repeated days lost to heavy rain.
• Analysis. They compared how often each species visited stations (visitation rate) across patch sizes and distances from the edge, removed the effect of patch area statistically, and tested each species’ trend with distance.

2. What they found

Visits

• 917 records of the five species. The red-rumped agouti was by far the most frequent in every size class, and visited most in small and medium patches. All the other species visited more in large patches and continuous forest.
• Stations were statistically independent of one another: visits at nearby stations were no more alike than those at distant ones.
• Recording all five species took 501 track station days in large patches and 710 in continuous forest, but 1450 in small and 1220 in medium patches.

Patch size

• Patch size affected both groups, in opposite ways: rodent visits fell, and ungulate visits rose, as patches got bigger.

Distance from the edge

• Neither rodents nor ungulates showed an overall effect of distance from the edge.
• For ungulates, the response to the edge depended on patch size. In large patches, visits at 150 m (mean 0.69) were much higher than at the edge (0.18). In small patches the pattern reversed (0.05 at 150 m against 0.21 at the edge). This still held after the effect of patch area was removed.
• At species level, lowland tapirs showed a significant fall in visits away from the edge in small patches and a significant rise in large patches. Agoutis showed no trend in either. There were too few records to test paca, brocket deer and peccary.
• Data from five core transects 350 to 500 m from the edge (unpublished) suggest edge effects for ungulates may reach about 365 m into large patches.

Cost and effort

• Sand and bananas for one transect of 10 stations cost about US$1.20; with labour and transport, about US$38.20. Running the same transect with camera traps would cost about US$2255 to US$3990.
• Labour was the main cost: 6 person-days to set up three transects, including carrying about 200 kg of sand per patch, and stations had to be checked daily. Camera traps are lighter and need checking only now and then.

3. What they concluded

• The species groups showed no direct response to edges independent of patch size. They seem to be neither “forest interior” nor edge-sensitive species: either they depend on forest but are not sensitive to edges, or they tolerate and may even do well at edges.
• Agoutis may thrive in small patches because large predators (jaguar, puma, ocelot) are unlikely to occur in patches under 400 ha, and hunting does not much affect them in this region. As seed predators, high agouti numbers could affect forest regeneration.
• Ungulates may use small patches only now and then, for food or shelter, moving through the edges between patches and across pasture. How patches are arranged in the landscape, how connected they are, and how close to continuous forest may matter more for mobile species such as tapirs than the edge alone.
• Edge effects for large mammals likely reach beyond the usual 150 m.
• Sand track stations are cheap, repeatable and give a lot of data over a large area in a short time. Despite uncertainties in track identification, they have not been used to their full potential. Combining them with camera traps and direct counts could help calibrate indices of relative abundance.

4. How tracking was or was not used

The whole study rests on tracks and signs identification: footprints left in prepared sand at baited track stations were identified to species every day and counted as visits. No trailing was reported. The authors note the “uncertainties inherent in track identification” and the care needed to keep the sand at the right dampness for clear, consistent tracks.

5. Tracker or observer certification

Reported but not validated. Tracks were identified “with the assistance of local hunters who were familiar with tracks of all species”, then double-checked with field guides (Becker and Dalponte 1991; Emmons and Feer 1997). The hunters’ skill was not tested, and they are not named in the paper. The acknowledgements thank Geraldo, Jorge and Emma Flaxman for field assistance.

6. Other technologies, skills or innovations

• A detailed recipe for sand track stations: a cleared, levelled 1.5 by 1.5 m square; about 4 kg of coarse sand (0.5 to 1 mm) as a base, topped with 2 kg of fine sand (0.1 mm); a slice of banana as bait; 1 to 2 litres of water every 2 to 3 days in dry weather.
• A sampling schedule to reduce repeat visits being counted as independent: 5 days of monitoring, a 2-day break without bait, re-baiting on day 8, then 5 more days.
• A cost comparison of track stations against camera traps.

7. Implications for conservation (from the paper)

• Understanding the impact of agoutis and pacas on seeds and seedlings should be a priority for managing and restoring forest in the region.
• Models of edge effects should include the spatial arrangement of patches in the landscape, and studies should sample the full range of patch sizes relevant to the species.
• Research on forest fragmentation needs more focus on medium and large mammals, and on cheap, time-effective survey methods for highly mobile species.