Extracting reproductive parameters from GPS tracking data -- a new tool
for a nesting raptor in Europe
Abstract
Understanding population dynamics requires estimation of demographic
parameters like mortality and productivity. Because obtaining the
necessary data for such parameters can be labour-intensive in the field,
alternative approaches that estimate demographic parameters from
existing data can be useful. High-resolution biologging data are now
frequently available for large-bodied bird species, and can be used to
estimate survival and productivity. We build on existing approaches to
develop a new tool (‘NestTool’) that uses GPS tracking data at hourly
resolution to estimate important productivity parameters such as
territory acquisition, breeding propensity and breeding success. We
developed NestTool with data from 258 individual red kites (Milvus
milvus) from Switzerland tracked for up to 7 years. NestTool first
extracts 42 movement metrics such as time within a user-specified
radius, number of revisits, home range size, and distances between most
frequently used day and night locations from the raw tracking data for
each individual breeding season. These variables are then used in three
successive random forest models to predict whether individuals exhibited
home range behaviour, initiated a nesting attempt, and successfully
raised fledglings. The models achieved > 95% accurate
classification of home range and nesting behaviour in cross-validation
data, but slightly lower (> 80%) accuracy in classifying
the outcome of nesting attempts, because some individuals frequently
returned to nests despite having failed. NestTool provides a graphical
user interface to manually annotate those individual seasons for which
model predictions fall below a user-defined threshold of certainty. When
applied to tracking data from different red kite populations in Germany,
NestTool yielded accurate predictions with > 80% accuracy
in all parameters. NestTool is available as R package at
https://github.com/Vogelwarte/NestTool and we encourage ornithologists
to adapt it for different populations and species. NestTool will
facilitate the more widespread estimation of demographic parameters from
tracking data to inform population assessments