Drought-driven foraging adjustments in breeding white storks (Ciconia ciconia): GPS tracking insights from two French marshes
Data files
Dec 30, 2025 version files 83.64 KB
-
matrice_ciccic_2025all.csv
15.46 KB
-
README.md
6.07 KB
-
Stats_JABarticle_BEGASSATetal_2025.Rmd
62.11 KB
Abstract
In the context of climate change, the increasing frequency of severe meteorological events, such as floods or droughts, is expected to impact various life history traits in organisms, primarily by altering the availability and quality of their trophic resources. Our study aimed to quantify the effects of meteorological conditions on the fine-scale space use of breeding white storks (Ciconia ciconia). Birds were equipped with GPS tags and accelerometer sensors in two breeding areas within the marshes of the French Atlantic coast and monitored over four years, including one year of drought. Specifically, we examined variations in home-range size, daily foraging distances, the proportion of time spent foraging, and daily activity levels in relation to drought conditions and individual state (sex, brood age and brood size). Our findings reveal that under drier conditions, storks increased their daily foraging distances, home-range size, and time spent foraging. Individuals with smaller broods travelled greater distances from the nest, and tended to exploit larger home-ranges. Their activity levels and time spent foraging increased with brood age and brood size, and were higher in females than in males. Our findings reveal how climate change, particularly drought, affects the foraging behaviour of a wetland top predator, and highlight the conservation challenges faced by wetland ecosystems.
Dataset DOI: 10.5061/dryad.tdz08kqct
Description of the data and file structure
White Storks (Ciconia ciconia) were equipped with GPS tags and accelerometer sensors in two breeding areas within the marshes of the French Atlantic coast and monitored over four years, including one year of drought. Specifically, we examined variations in home-range size, daily foraging distances, the proportion of time spent foraging, and daily activity levels in relation to drought conditions and individual state (sex, brood age and brood size).
Files and variables
File: matrice_ciccic_2025all.csv
Description: Filtered and aggregated data matrix used for the statistical models.
Variables
- device: GPS device ID
- id_30: 30-day period ID (1st or 2nd part of the rearing period)
- Deviceyear: concatenation of the device ID and the study year
- num_unique: unique no. per Deviceyear
- ID: shorter ID, concatenation of num_unique, the two last number of the year and the period number (1 or 2)
- YEAR: study year
- POP1: French administrative department code of the bird (44: Loire site, 17: Gironde site)
- POP: same as for POP, except that 44 is replaced by 1 and 17 is replaced by 2.
- SEXE: sex of the individuals
- PRODUCTION_envol: number of fledglings
- START: start date of the recordings used in the analyses
- END: end date of the recordings used in the analyses
- ECLO: hatching date
- ENVOL: fledging date
- PERIOD: 1st or 2nd 30-day period of the chick rearing
- CHICKAGE: 0 when in the 1st period, 1 when in the 2nd period of chick rearing
- n_all_bhv: number of GPS locations (all behaviours)
- days_available: number of days with valid GPS locations
- n_foraging: number of GPS locations classified as "foraging" ("walking")
- ratio_foraging_all: ratio between n_foraging/ n_all_behaviours
- HRker95_allhadjust: Home range size calculated using Kernel density estimator (95%) using adjusted h
- HRker95_foragingHref: Home range size calculated using Kernel density estimator (95%) using h-ref
- Hrker_foraging_hadjust: Home range size only based on foraging locations, calculated using Kernel density estimator (95%) using adjusted h
- Mean_nest_dist: distances between the nest and each foraging location were calculated for each individual-period using the R package ‘geosphere’ (Hijmans et al. 2022), and averaged for each day to obtain mean daily distance.
- prop_for_month: the proportion of foraging activity calculated by dividing the number of walking locations by the total number of GPS locations for each 30-day period.
- ODBA.mG.: daily mean Overall Dynamic Body Acceleration was calculated first by summing the absolute dynamic acceleration across the three spatial axes for each 10 s sequence, and then by averaging these values across all sequences recorded within the day (Qasem et al. 2012). Daily mean ODBA was then averaged for each 30-day rearing period.
- daily_mean_ODBA: same as ODBA.mG. but in G
- SPEI_har_pond: the standardized precipitation-evapotranspiration index (SPEI). This index was calculated using the functions spei and hargreaves (modified form of the Hargreave’s equation; Droogers and Allen 2002) from the ‘SPEI’ package (Beguería et al. 2014, Beguería and Vicente-Serrano 2023). It was computed on a monthly basis, incorporating data from the two preceding months, to reflect time-integrated drought conditions experienced by breeding storks. It was then weighted for each individual 30-day period based on the proportion of days from each overlapping month that fell within the period.
File: Stats_JABarticle_BEGASSATetal_2025.Rmd
Description: Script containing the analyses performed in the manuscript.
Code/software
Analyses were conducted using R ver. 4.3.2 (www.r-project.org) with the ‘RStudio’ interface.
We used the following R packages: dplyr, corrplot, lme4, nlme, lmerTest, Matrix, MuMIn, car, lubridate, GGally, mgcv, lmtest, effects, flexmix, performance, gridExtra, emmeans, ggeffects, and multcomp.
Full package citations are provided below.
- dplyr — Wickham et al. (2023). dplyr: A Grammar of Data Manipulation.
- corrplot — Wei & Simko (2024). corrplot: Visualization of a Correlation Matrix.
- lme4 — Bates et al. (2015). Fitting Linear Mixed-Effects Models Using lme4.
- nlme — Pinheiro & Bates (2000); Pinheiro et al. (2025). nlme: Linear and Nonlinear Mixed Effects Models.
- lmerTest — Kuznetsova et al. (2017). lmerTest Package: Tests in Linear Mixed Effects Models.
- Matrix — Bates et al. (2025). Matrix: Sparse and Dense Matrix Classes and Methods.
- MuMIn — Bartoń (2025). MuMIn: Multi-Model Inference.
- car — Fox & Weisberg (2019). An R Companion to Applied Regression.
- lubridate — Grolemund & Wickham (2011). Dates and Times Made Easy with lubridate.
- GGally — Schloerke et al. (2025). GGally: Extension to ggplot2.
- mgcv — Wood (2003, 2004, 2011, 2016, 2017) — foundational GAM references.
- lmtest — Zeileis & Hothorn (2002). Diagnostic Checking in Regression Relationships.
- effects — Fox & Weisberg (2019), plus Fox (2003), Fox & Hong (2009), Fox & Weisberg (2018).
- flexmix — Grün & Leisch (2025; 2004; 2007; 2008). flexmix: Flexible Mixture Modeling.
- performance — Lüdecke et al. (2021). performance: Assessment and Testing of Statistical Models.
- gridExtra — Auguie (2017). gridExtra: Miscellaneous Functions for Grid Graphics.
- emmeans — Lenth (2025). emmeans: Estimated Marginal Means.
- ggeffects — Lüdecke (2018). ggeffects: Marginal Effects from Regression Models.
- multcomp — Hothorn et al. (2008). Simultaneous Inference in General Parametric Models.