Sampling Effects of Time and Season

Time and season are two of many factors known to influence avian singing behaviour and the probability of detection (Alldredge et al. 2007).  For example, songbirds generally sing most frequently in early morning, and males tend to sing more often in the early breeding season to establish territories and attract mates.  These factors affect the results of point counts; surveys conducted in periods of lower singing activity will detect fewer individuals, simply because birds are quieter, not necessarily because they are less abundant.  Our sampling effect graphs illustrate these relationships, using BAM point count data to show how time of day and season influence the singing rates of many bird species.  

To determine the best fit between bird abundance for individual species and the time of day/year when surveys were conducted, while controlling for other factors known to influence the mean count (e.g. count duration and distance), we used a Generalized Additive Model with a Poisson error family.  Using the predictions from this model, we created smoothed curves for time and season that set other variables in the model to constants.  

For reporting here and to make the models comparable among species, the predicted values for each time/season combination were divided by the maximum predicted count for that species.  This scaled all species to a maximum of 1, so that curves show the relative change in a species count over time.  The graphs do not account for repeated sampling at the same point count station, so locations with more survey effort have more influence on the predicted count (this enabled the creation of smoothed curves).  

We applied this information in our analyses to correct for observed sampling effects, so that all BAM survey data were standardised to account for these environmental factors that influence bird counts, based on the locations where they were collected.  We did this by adjusting all counts according to time since local sunrise and start of local spring.

Full technical details, including the assumptions of these models, will be available on our website after results have been published.

Reference:

Alldredge, M.W., T.R. Simons, and K.H. Pollock. 2007. Factors affecting aural detections of songbirds. Ecological Applications 17:948-955.