Understanding how wildlife responds to the spread of human-dominated habitats is a major challenge in ecology. It is still poorly understood how urban areas affect wildlife space-use patterns and consistent intra-specific behavioral differences (i.e., behavioral types; BTs), which in turn shape various ecological processes. To address these questions, we investigated the movements of a common resident wader, the spur-winged lapwing (Vanellus spinosus), hypothesizing that urban individuals will be more mobile than rural ones. We used an ATLAS tracking system to track many (n=135) individuals at a high resolution (8 second-fix interval) over several months each. We first established that daily movement indices show consistent differences among individuals, acting as spatial-BTs. Then focusing on the two main principle-components of lapwings’ daily-movements – mobility and position along the exploration-exploitation gradient– we investigated how these BTs are shaped by urbanization, season (nesting vs. non-nesting), and sex. We found that urban lapwings were indeed more mobile in both seasons. Furthermore, urban females were less explorative than rural females, especially during the nesting season. These results highlight how urbanization affects wildlife behavior, even of apparently urban-resilient avian residents. This underscores the need to consider possible behavioral consequences that are only apparent through advanced tracking methods.

For movement tracking we captured nesting lapwings during 2019-2022. We searched the study area for nests and placed a walk-in trap on them. When possible, sex was determined from spur length (males > 10 mm, females < 7 mm), and otherwise by means of genetic markers from a commercial provider, using feather samples collected from all individuals. We used 7g tracking device tags attached in a leg-loop configuration (< 4% of body weight, including harness). A total of 211 transmitters were deployed on 194 individual adult lapwings (2020 = 68, 2021 = 74 and 2022 = 69), including 17 re-captures in subsequent years.

To obtain thetracking data at the high-frequency we used ATLAS (Advanced Tracking and Localization of Animals in real-life Systems). This system is based on a reverse-GPS, in which the bird-borne tags emit a unique-ID signal at set intervals that is detected by an array of fixed base-stations (tower-mounted antennas) deployed at high vantage points in the region. Signals detected by three or more base stations enable localization of the tag from their differential times of arrival. The tags were set to transmit at 8-sec intervals (1/8Hz), with a practical lifetime of up to 315 days. The deployment of our particular ATLAS system comprised 19 base-stations covering ~350 km². Tracking always started during the nesting season (March-August), but many of the trajectories extended into the non-nesting season (September-February), allowing us to include seasonal differences in the analyses, while addressing individual consistency in behavior, on an urbanization gradient.

To determine whether lapwing movements present spatial-BTs, we characterized their daily movements using several movement indices. Center of activity (COA) was defined with a 100x100 m grid, as the area where the lapwings spent more than 75% of their total tracking time. Using the flight segments we calculated daily: (1) total time in COA (minutes); (2) maximum displacement from the COA (the distance to the furthest location in the daily trajectory, in meters); (3) total flying distance (in meters); (4) number of flying segments, (5)  flying speed (meters/second); (6) max diameter (the distance between the two most distant locations in a daily trajectory, in meters); (7) daily locations visited; (8) daily unique locations visited; (9) re-visitation rate (total visited locations/unique locations visited on the map); and (10) total time on the ground.

To determine whether the affinity to urbanization affects the lapwings' movement behaviors, we estimated the coverage of urban habitat within their core home range (HR). As an index of their core HR (where they spent most of the time) we used the median kernel density estimation (KDE50%). For our analysis we consider urban habitats as areas where people live, such as small villages or cities including lawns and roundabouts, or open areas within settlement border. Rural habitats, in contrast, include fields and ponds with much lower human infrastructure (e.g. buildings) and activity.

For determining the core HR (KDE50%) we subsampled our data into 1-hour intervals. We considered the non-flight locations only. The KDE analysis was implemented with the package 'adehabitatHR' with parameters of h = href, extent = 1 and grid = 500. We combined habitat polygon layers of urban areas and fish ponds, with agricultural fields, and calculated the percentage of built-up in each lapwing’s core home-range. This index (ranging from 0 to 100%) was used as a proxy for the bird's level of urbanization, and individuals could readily be classified into ‘urban’ and ‘rural’ individuals. The index showed a clear bi-modal distribution with one peak below 30% (rural lapwings) and one above 60% (urbans).

We calculated each one with repeatability using the package 'rptR', using 1000 bootstraps for confidence interval estimation, including the individual identity as a random effect.  To reduce the dimensionality of movement, we then performed a Principal Component Analysis (PCA) that included all indices that were significantly repeatable (P < 0.05). The PCA was implemented with the 'prcomp' package from a base 'stats' package in R. The first two principal components (PC1 and PC2) were used as the main dependent variables.

For each PC we used a set of generalized linear mixed models (GLMMs), in a model comparison framework. All models included bird ID as random factors. Fixed effects (predictors) varied among models and included (1) our main predictor of interest - the urbanization level (urban or rural), as well as the additional predictors of: (2) sex (male or female); (3) season (nesting or non-nesting); and (4) initial weight (scaled). We also considered the date in the model as temporal autocorrelation, by using 'corAR1' in the model for each ID. Because the effect of urbanization on movement can be mediated by these predictors, we also included models with two- and three-way interactions of urbanization with the other three predictors.  We implemented model selection with the 'MuMIn' package, and included only models with Delta_AICc ≤ 4  in the model averaging.