Modelling the abundance and distribution of marine birds accounting for uncertain species identification

In certain habitats, aerial surveys can monitor bird populations by taking high quality photos or videos. This is an example of how new technology is revolutionizing monitoring. Such passive survey techniques can often cover a larger area faster than human surveyors can, and may cause less disturbance to the animals concerned. However, these advantages may be offset by the loss of accuracy in identifying species or individuals, as the human ability to take in aspects of size, colour, behavior, habitat and group dynamics (all of which contribute to identification) has not yet been surpassed by technology.

A new study led by the BTO has analysed aerial photos of birds at sea taken to estimate species’ population sizes at a site where an offshore wind farm has been proposed. Traditionally birds at such sites have been counted by observers on boats, but digital aerial snapshots are quicker, and therefore avoid the issue of double counting that can arise from boat surveys. However, these aerial photos do not always create an image that is good enough to identify birds to species-level. Only 23% of photographed birds were identified to species-level with any confidence in this study, with some individuals classed only by family, e.g. “auk”. This low proportion identified presented a major hurdle to estimating population sizes.

To overcome this problem, the study incorporated data from boat surveys carried out at similar times and places to the aerial surveys. On boat surveys, observers routinely identify the species of up to 95% of the birds they record. Sophisticated statistical models were used to compare the proportions of different species observed on boats with that from planes, allowing “intelligent guesses” to be made on the species identification of birds in the aerial surveys. This is the first time population estimates have been calculated for individual species with uncertain identification. This approach could be applied in a range of situations that do not identify species with certainty, resulting in large-scale, quick, efficient and non-invasive monitoring that has obvious conservation benefits in today’s rapidly changing world.