Introduction

Understanding the dynamics of animal populations requires data on key demographic processes such as mortality, productivity, and dispersal. The widespread adoption of miniaturized automated tracking devices has facilitated the study of seasonal migration patterns of many animal species in great detail (Kays et al. 2015, Panuccio et al. 2021). So far, however, relatively few studies have used these large-scale GPS tracking data to extract information on survival (Klaassen et al. 2014, Sergio et al. 2019, Swift et al. 2020, Buechley et al. 2021) or reproduction (Picardi et al. 2020, Schreven et al. 2021, Bowgen et al. 2022, Overton et al. 2022, Ozsanlav-Harris et al. 2022, Eisaguirre et al. 2023).
Reproductive performance in long-lived species generally improves with age, and the timing and outcome of the first reproduction can have major implications for population dynamics (Blas et al. 2009, Sergio et al. 2011, Weimerskirch 2018). As young birds recruit into the breeding population, their first breeding attempts often fail at an early stage and can therefore be difficult to detect through field observations (Péron et al. 2014, Kidd et al. 2015). Because the collection of reproductive information from individuals in the field can be very time- and labour-intensive, using tracking data to determine when the first reproduction occurs and how successful it is may therefore offer additional detail to help understand the population dynamics of long-lived species (Murgatroyd et al. 2023).
Information on reproduction can be derived from animal tracking data if individuals follow repeatable patterns that can be reliably identified and distinguished. For example, the nest as a central location will be visited frequently during the breeding phase, resulting in typical central-place foraging movement patterns that can be detected in tracking data (van der Wal et al. 2015, Picardi et al. 2020, Bowgen et al. 2022, Overton et al. 2022, Eisaguirre et al. 2023). The identification of nesting behaviour can be further improved by using both GPS and accelerometery data, as periods of incubation lead to unique patterns with near-zero movement (Schreven et al. 2021, Ozsanlav-Harris et al. 2022). However, accelerometery data are not ubiquitously available, and it is so far not clear whether models can be developed that reliably identify nesting and breeding success of different populations of the same species based solely on location data (Picardi et al. 2020, Eisaguirre et al. 2023).
The red kite (Milvus milvus ) is a widespread European raptor species that has recently recovered from historic persecution in central Europe (Mammen et al. 2017, Aebischer and Scherler 2021), but is still vulnerable to anthropogenic causes of mortality in south-western Europe (Mateo-Tomás et al. 2020, Sergio et al. 2021). Despite being a long-lived species with delayed maturity, the temporal fluctuation of productivity was the most important demographic driver of the trajectory of red kite populations (Sergio et al. 2021, Pfeiffer and Schaub 2023). A better understanding of the age at which young red kites recruit into the breeding population and contribute to the reproductive output of a population is therefore a key demographic parameter that is so far poorly studied (Katzenberger et al. 2021, Literák et al. 2022, Pfeiffer and Schaub 2023). Developing a tool that allows an objective assessment of this recruitment process would therefore be a useful advance for studying red kite population dynamics.
Besides population dynamics, the movements of red kites have attracted much research interest and > 2000 individuals have been equipped with tracking devices in Germany, Austria, Switzerland, Spain and France (Aebischer and Scherler 2021, Mattsson et al. 2022, Pfeiffer and Meyburg 2022). These tracking efforts have revealed information on mortality and survival (Sergio et al. 2019, Katzenberger et al. 2021, Mattsson et al. 2022), migration and dispersal (Maciorowski et al. 2019, García-Macía et al. 2022a, Literák et al. 2022, Scherler et al. 2023b), and revealed that successfully breeding red kites showed smaller home ranges than unsuccessfully breeding birds (Spatz et al. 2022). Therefore, telemetry data obtained from individuals during the breeding season may be sufficient to extract information on breeding propensity and success. However, the ranging behaviour of red kites varies markedly between countries and years, with territory sizes dependent on landscape composition, habitat structure, and food availability (Pfeiffer and Meyburg 2015, Aebischer and Scherler 2021, Spatz et al. 2022). There is an urgent need to test whether the predictions of certain demographic parameters based on movement patterns are transferrable across populations (Mattsson et al. 2022).
We built on existing approaches (Picardi et al. 2020, Schreven et al. 2021, Overton et al. 2022) to develop a transferable algorithm (hereafter referred to as ‘NestTool’) predicting important demographic parameters (territory acquisition, breeding propensity, and breeding success) of red kites using only GPS tracking data. We trained and tested this model using data from 697 known-outcome individual breeding seasons in Switzerland, and validated the utility of the algorithm with data from two separate populations in Germany. We provide guidance on what movement patterns indicate successful breeding and provide software to rapidly annotate existing tracking data with information about breeding propensity and success. We further examine the conditions under which an accurate prediction of nest outcome is not possible, and caution researchers that typical tracking data will not be able to classify all breeding episodes correctly. Our NestTool facilitates the rapid estimation of breeding propensity and success, and will therefore contribute to a better understanding of important demographic processes that drive red kite population dynamics (Pfeiffer and Schaub 2023).