Comparing GPS collar and fecal sampling using detection dogs for habitat selection analysis in brocket deer (Mazama)

Abstract Habitat use data are key to understanding species ecology and extinction risk. However, such information is lacking for the elusive deer species of Neotropical region. In this context, fecal sampling has emerged as an alternative tool, in which development and evaluation are essential to obtaining unbiased ecological data. We aimed to compare data from GPS-tracked animals and fecal sampling using scat detection dogs to evaluate the noninvasive performance of this method in habitat selection analysis. We carried out the study in the Brazilian Pantanal, where we monitored six free-living Gray Brocket Deer (Mazama gouazoubira) with GPS collars for 1 year (average of 584 GPS locations/animal) and collected fecal samples (n = 649) simultaneously along a set of transects designed for a scat detection dog survey. We evaluated habitat selection using the chi-square test in an availability/utilization analysis and submitted both data to a bootstrap procedure to assess its precision and accuracy with increasing sample size. GPS data indicated habitat selection at a fine utilization scale, in which savanna and cerrado were preferred and open grassland habitat was avoided. Exclusive fecal sampling also indicated habitat selection, revealing the preference for cerrado and avoidance of open grassland. The GPS and fecal habitat utilization estimates did not differ significantly and fecal sampling increased precision and accuracy with increased sample size, reaching minimal values once n = 200 which should be considered a sufficient survey effort. The similarity between the two methods suggested the reliability of fecal sampling, as long as a standardized sampling design is used. This noninvasive sampling framework can provide previously unavailable ecological data for threatened Neotropical deer as well as other elusive species.