Acoustic monitoring is a powerful technique for learning about the ecology of bats, but understanding sources of variation in the data collected is important for unbiased interpretation. The objectives of this dissertation were to investigate sources of variation in acoustic monitoring and make recommendations for acoustic survey design and analysis. I addressed this goal in three ways: i) variation resulting from differences in bat detectors, ii) methods for objective identification of peak activity, and iii) the use of stationary transects to address within-site spatial variation.
First, I compared variation of detection of echolocation calls among commonly available bat detectors and found significant differences in distance and angle of detection. Consequently, this source of variation should be taken into account when comparing datasets obtained with different systems. Furthermore, choice of detector should be taken into account when designing new studies.
Second, I investigated two statistical methods for identifying peaks in activity, percentile thresholds and space-time scan statistic (SaTScan). Acoustic monitoring provides a relative measure of activity levels and is rarely evaluated based on objective criteria, so describing bat activity as “high” or “low” is useful only in context of the studies in question. Percentile thresholds allow for peaks to be identified relative to a larger distribution of activity levels. SaTScan identifies peaks in space and time that are significantly higher than the background expectation of the dataset. Both methods are valuable tools for replicable and objective identification of peak activity that can be applied at various temporal and spatial scales.
Third, I examine how within-site spatial variation can impact estimates of bat activity. I used a stationary transect of bat detectors to i) assess variation in patterns of activity at each detector, ii) test whether spatial or temporal factors were more important for explaining variation in activity, iii) explore what sampling effort in space and time is required for species-specific activity levels. The picture of activity differs significantly within a site depending on detector placement so it is important to use multiple detectors simultaneously to collect accurate estimates of activity.