Animals that Hide and the Challenges of Accurate Assessment

Our conclusions about anthropogenic impacts on the environment, and our decisions of how to manage these impacts, are only as good as the methods we use to identify and measure them.  In studies of complex interactions between tropical subsistence hunters and their game, identifying hunting-induced changes in animal populations remains technically challenging and influences everything from management decisions to downstream ecological studies such as cascading effects and disruption of food webs. The most commonly used method to assess vertebrate populations in the tropics is visual sightings along line transects.  Statistical models and software programs use these counts to derive measures of occupancy, abundance, diversity, density, biomass and population size. Errors or biases that enter during data collection influence these derived applications. We assess the consequences of using sighting counts to detect presence-absence, spatial distribution, population parameters, and hunting impacts on vertebrate populations by comparing counts of sightings vs. sign encountered along stratified linear transect arrays for 255 species in 16 naturally vegetated areas of a 30,000 km² region of Amazonia.  We walked a total of 12,400 km over 128 transects, over 12 to 17 months per site.  For 6 ungulate species we recorded 6 to 70 times more sign than sightings, for 4 armadillo species from 84 to 166 times more sign, and similar trends occurred for 3 out of 5 rodent, 6 out of 8 carnivores, and 6 out of 6 feral/domestic species.  The relationship was reversed for 6 out of 7 primate species and for almost all bird, reptile and amphibian species.  25,000 sign data points were collected vs. 12,000 sightings; numbers were highly positively correlated for 11 of 12 of the most hunted animal species.  Landscape level abundance and distribution patterns for many species differed when described using sign vs. sighting data; animals were not only present but abundant in areas were they would have reported as rare or extinct based only on sighting data.  In Amazonian ecosystems, the use of sighting data alone may thus result in underestimates of the presence, rarity and abundance of many species, and misinterpretation of hunting impacts.