Monitoring large mammals in high Andean prairies and cloud forests in Bolivia

Abstract

Monitoring programs of population trends frequently have design flaws which might result in data insufficient to statistically test whether a population trend exists or not. This fact weakens decisions regarding the conservation status of the species involved. Monitoring programs with designs that include a power analysis can be adjusted in a way to ensure obtaining useful data and simultaneously avoid wasting field efforts and scarce resources. In this paper, I use field data to design a monitoring program of population trends for five species of mammals, considering the results of a power analysis conducted on a pilot survey, and discuss the need of considering survey features that help the decision making process for the conservation of species. The pilot survey included counts of visitation rates in groups of lines with six track plots each. Separation between track plots within a line was 50 m, and lines were separated 600 m from each other. Two groups of six lines of track plots were placed in sites separated 10 km from each other, covering high Andean prairies and cloud forests within the Cotapata National Park in Bolivia. Track plots were either baited with different lures for carnivores or lacked any bait for four consecutive days, and were revised 24 hs after preparation. Mean and standard deviations calculated from counts of the proportion of lines with tracks (using days as replicates) were used to simulate the power of a monitoring program based on such design. As different lures or none was used in consecutive days, I considered as true replicates only those days when the effect of the lure or the lack of it did not significantly affect visitation rates. Monitoring programs aimed for an 80% power to detect a declining trend of 3% per year in visitation rates (assuming that this track-based index can be used to monitor true population trends), with a sampling period of 10 years to be considered acceptable. If the simulated monitoring program did not reach those requirements using the original field-based design, further simulations were run changing the number of groups of track plot lines, and the number of counts done on each one, until the requirements were fulfilled. Results indicate that an acceptable monitoring program for Hippocamelus antisensis would require at least nine counts on eight groups of track plot lines, while a monitoring program for Mazama chunyi would require nine counts in six groups of track plot lines to be acceptable. Seven counts on 10 groups of track plot lines would be enough to monitor Leopardus tigrinus, 14 counts on 10 groups of lines would be required to monitor Oreailurus jacobita, along with Lynchailurus pajeros populations (these two simpatric species cannot be identified by their tracks). Finally, a monitoring program for Pseudalopex culpaeus would require five counts on six groups of track plot lines to be acceptable. The implications of the lack of a power analysis in monitoring efforts is discussed, as well as considerations regarding the separation between lines of track plots when monitoring different species.