Density Estimates

The BAM project uses its national database to generate population density estimates, which can be an important tool for the conservation of boreal birds.  These results are stratified by Bird Conservation Regions, the ecologically defined planning units for bird population management in North America, and by provincial/territorial jurisdictions overlapping the boreal forest in Canada. 

To estimate density, first we had to address the basic problem that surveys rarely detect all individuals, leading to an underestimate of animal abundances.  As we have such a large amount of survey data, from thousands of point counts conducted by different researchers across the boreal forest, our estimates can correct for many of the factors that typically introduce ‘variation’ or ‘error’ into survey results.

Note, there is no single “correct” approach to calculating population sizes.  Our density values, like all estimates, include an inherent level of uncertainty. They were calculated using one possible approach to correcting for the “nuisance factors” that influence bird counts, based on these assumptions:

• Observers accurately estimated the distance to birds.
• Effective detection radius (EDR) was constant among habitats and jurisdiction.
• Bird survey locations were spatially and temporally representative of jurisdictions.
• Habitat strata were classified correctly.

Density estimates were calculated using a statistical method called hierarchical regression or Generalized Linear Mixed Models.  Our correction process involved a complex series of calculations, starting with the raw survey data from each study (i.e. the number of individuals observed).  We produced density estimates using the following series of steps:

• Correction for differences among the point count survey methods used by various researchers (e.g. count duration or distance);
• Correction for environmental factors that affect counts (e.g. time of day and year; refer to graphs showing influence of time and season on number of birds detected);
• Account for inter-species differences in the detectability of bird sounds, because some species sing more loudly or frequently than others;
• Calculation of bird density in each habitat type, because factors like vegetation and human land-use affect the habitat preferences and behaviour of birds; and
• Calculation of “stratified density estimates” at the BCR-province level: multiplied the species’ density estimate per habitat type by the amount of each habitat within each BCR-province class, to obtain the total density of the species in the area.

This is the formula used to calculate a density estimate, D (birds per hectare), where C is the number of birds per point count, and r is the sampling area over which detection error is 0.  The numerator includes corrections for variation in count length, distance, time of day/year, and number of visits per station, while the denominator includes correction for detectability (EDR).

Maps of relative density estimates:  The resulting colour-coded maps compare the relative density estimates for each bird species (number of males per hectare) by Bird Conservation Regions within each province/territory.  They are scaled so that the darkest colours represent the highest density for that species and the lightest colours represent the lowest.  This scaling is intended to show the relative change in density across the boreal forest in Canada.

Example:

Tables of absolute density estimates: As a companion to the colour-coded map of relative densities, these tables show a species’ absolute density in the boreal Bird Conservation Regions within each province and territory across the boreal forest in Canada.

Example:

Full technical details of the BAM estimation method will be available on our website after results have been published.