Line-transect versus point-transect sampling: The effects of survey area and survey effort on method efficiency for Geoffroy's spider monkeys

Context: Line-transect sampling is considered to be a more efficient survey method than point-transect sampling to estimate population densities and abundance of many animal species. Aims: In the present study, we tested whether this claim holds true when surveying arboreal fast-moving primate species occurring at low densities, and whether the potential difference in efficiency can be explained by the difference in the size of the survey area between the methods. We further examined the impact of survey effort for point-transect sampling. Methods: We conducted line- and point-transect sampling for Geoffroy's spider monkeys (Ateles geoffroyi) in the same locations and compared the numbers of detected individual monkeys and the probability of their occurrence per survey between the two methods. We further compared the data from point-transect sampling gathered within three different waiting periods. Key results: We found a higher probability to detect monkeys and a higher number of monkeys during line-transect sampling than during point-transect sampling, but more spider monkeys were detected at point transects when controlling for the size of the survey area. More monkey detections were made during the first 10 min than during the second and third 10-min periods of point-transect surveys. Conclusions: We showed that line-transect sampling is more efficient than point-transect sampling when surveying Geoffroy's spider monkeys in a flat landscape of tropical forest with homogenous visibility. We discuss factors influencing survey results and recommend 20 min as the maximum waiting time at point transects when surveying arboreal mammals. Implications: Our study has provided a quantitative approach to compare efficiency across survey methods for fast-moving arboreal animals that occur at low densities, and supports the use of point-transect sampling in sites where line-transect sampling is not feasible, such as in human-modified landscapes.