Data and R script to reproduce the modelling in the accompanying journal article (Part I). The models are stored for convenience. Furthermore, the code for the assessment of nest success based on the analysis of GPS tracking data is provided (Part II). All files are available in Curlew_COE_habitat.zip

Description of the data and file structure

File tree

• data
• models
• scripts

Part I – habitat use effects on nest success

Data

dt_hab_Curlew.csv dataset includes the raw data, arranged into the following variables:

• ID_seas = bird ID and year of breeding
• Bird_ID = ID of the individual
• Sex = male (M) of female (F)
• Non_gr = non-breeding grounds, ATL-S = southern Atlantic coast, ATL-C = central Atlantic coast, ATL-N = northern Atlantic coast, NSea = North Sea coast
• Br_gr = breeding grounds, W = western Europe, NE = northeastern Europe
• Nest_succ = nest was successfully incubated (1) or not (0)
• X_br = mean longitude of the breeding grounds (redacted as sensitive data)
• Y_br = mean latitude of the breeding grounds (redacted as sensitive data)
• X_non = mean longitude of the non-breeding grounds (redacted as sensitive data)
• Y_non = mean latitude of the non-breeding grounds (redacted as sensitive data)
• Dist = shortest migration distance from non-breeding grounds to breeding grounds (km)
• Prebr = length of the pre-breeding period, i.e. a period between arrival at the breeding grounds and the onset of incubation (day)
• non_arable = arable land use on non-breeding grounds (proportions of use)
• non_grass = grasslands use on non-breeding grounds (proportions of use)
• non_marsh_salt = salt marshes use on non-breeding grounds (proportions of use)
• non_mudflats = intertidal mudflats use on non-breeding grounds (proportions of use)
• non_salines = salines use on non-breeding grounds (proportions of use)
• non_wetland = inland wetlands use on non-breeding grounds (proportions of use)
• st_arable = arable land use at stopover sites (proportions of use)
• st_grass = grasslands use at stopover sites (proportions of use)
• st_marsh_salt = salt marshes use at stopover sites (proportions of use)
• st_mudflats = intertidal mudflats use at stopover sites (proportions of use)
• st_wetland = inland wetlands use at stopover sites (proportions of use)
• br_arable = arable land use on breeding grounds (proportions of use)
• br_grass = grasslands use on breeding grounds (proportions of use)
• br_mudflats = intertidal mudflats use on breeding grounds (proportions of use)
• br_wetland = inland wetlands use on breeding grounds (proportions of use)

In the variables describing proportions of use of particular habitat classes, NA means the absence of a habitat class in a given breeding ground, non-breeding ground, or stopover site.

land_use_classes.xlsx Habitat use classes on sheets LULC_Coastal_Zones, LULC_CLCplus_backbone, LULC_CGLS as defined in the original data sources, and the habitat classes used for analyses. Detailed metadata is included in a separate sheet within the file.

summary_table_prop_dirich.csv A table including proportions of habitat use derived from Dirichlet regression, including the 95% credible interval of the estimates. Data is used for plotting.

summary_table_reg_prop_dirich.csv A table including proportions of habitat use in non-breeding and breeding areas, derived from Dirichlet regression, including the 95% credible interval of the estimates. Data is used for plotting.

Models

Mod_PCR.rds Stored model testing sex-specific effects of habitat use on non-breeding and breeding grounds on nest success, including the length of the pre-breeding period and breeding grounds (categorical predictor) as covariates. The model provides estimates of the fixed effects of the following variables:

• Sex = male (M) of female (F)
• Prebr = length of the pre-breeding period, i.e. a period between arrival at the breeding grounds and the onset of incubation (day)
• Br_gr = breeding grounds, W = western Europe, NE = northeastern Europe
• PCX_non = gradient of habitat use on the non-breeding grounds derived from a principal component analysis; X denotes particular principal component
• PCX_br = gradient of habitat use on the breeding grounds derived from a principal component analysis; X denotes particular principal component
• Sex:PCX_non = sex-specific effects of a gradient of habitat use on the non-breeding grounds derived from a principal component analysis; X denotes particular principal component
• Sex:PCX_br = sex-specific effects of a gradient of habitat use on the breeding grounds derived from a principal component analysis; X denotes particular principal component

The model also provides estimate of the variance of Bird_ID random effect (ID of the individual).

Mod_PCR_stop.rds Stored model testing sex-specific effects of habitat use at stopover sites on nest success, including the length of the pre-breeding period and breeding grounds (categorical predictor) as covariates. The model provides estimates of the effects of the following variables:

• Sex = male (M) of female (F)
• Prebr = length of the pre-breeding period, i.e., a period between arrival at the breeding grounds and the onset of incubation (day)
• Br_gr = breeding grounds, W = western Europe, NE = northeastern Europe
• PCX_st = gradient of habitat use on the stopover sites derived from a principal component analysis; X denotes the number of a principal component
• Sex:PCX_st = sex-specific effects of a gradient of habitat use on the stopover sites derived from a principal component analysis; X denotes particular principal component

The model also provides estimate of the variance of Bird_ID random effect (ID of the individual).

Mod_hab_dirich_all.rds Stored model investigating the proportions of habitat types used on non-breeding grounds, stopovers, and breeding grounds. The model provides estimates of the proportions of habitat use:

• munonHAB_propmatrixnon_Intercept = log odds of habitat use proportions (HAB) on non-breeding grounds, where HAB is grass (grasslands), marshsalt (salt marshes), mudflats (intertidal mudflats), salines, or wetlands (inland wetlands)
• mubrHAB_propmatrixbr_Intercept = log odds of habitat use proportions (HAB) on breeding grounds, where HAB is grass (grasslands), mudflats (intertidal mudflats), or wetlands (inland wetlands)
• mustHAB_propmatrixst_Intercept = log odds of habitat use proportions (HAB) on stopover sites, where HAB is grass (grasslands), marshsalt (salt marshes), mudflats (intertidal mudflats), or wetlands (inland wetlands)

Mod_hab_dirich_reg.rds Stored model investigating the proportions of habitat types used on non-breeding grounds, stopovers, and breeding grounds, grouped by the non-breeding areas and breeding areas. The model provides grouped estimates of the proportions (log odds) of habitat use:

• munonHAB_propmatrixnon_Intercept = habitat use proportions (log odds) on the non-breeding grounds for Curlews using the reference non-breeding (central Atlantic coast) and breeding areas (northeastern Europe), where HAB is grass (grasslands), marshsalt (salt marshes), mudflats (intertidal mudflats), salines, or wetlands (inland wetlands)
• mubrHAB_propmatrixbr_Intercept = habitat use proportions (log odds) on the breeding grounds for Curlews using the reference non-breeding (central Atlantic coast) and breeding areas (northeastern Europe), where HAB is grass (grasslands), mudflats (intertidal mudflats), or wetlands (inland wetlands)
• mustHAB_propmatrixst_Intercept = habitat use proportions (log odds) on the stopover sites for Curlews using the reference non-breeding (central Atlantic coast) and breeding areas (northeastern Europe), where HAB is grass (grasslands), marshsalt (salt marshes), mudflats (intertidal mudflats), or wetlands (inland wetlands)
• munonHAB_propmatrixnon_Non_grAREA = difference in habitat use proportions (log odds) of a given non-breeding habitat type (HAB) between Curlews using a given (AREA) and the reference non-breeding area (central Atlantic coast), where HAB is grass (grasslands), marshsalt (salt marshes), mudflats (intertidal mudflats), salines, or wetlands (inland wetlands), and AREA is ATLMN (northern Atlantic coast), ATLMS (southern Atlantic coast), or NSea (North Sea coast)
• munonHAB_propmatrixnon_Br_grW = difference in habitat use proportions (log odds) of a given non-breeding habitat type (HAB) between Curlews using western Europe and the reference breeding area (northeastern Europe), where HAB is grass (grasslands), marshsalt (salt marshes), mudflats (intertidal mudflats), salines, or wetlands (inland wetlands), and W is western European breeding ground
• munbrHAB_propmatrixbr_Non_grAREA = difference in habitat use proportions (log odds) of a given breeding habitat type (HAB) between Curlews using a given (AREA) and the reference non-breeding area (central Atlantic coast), where HAB is grass (grasslands), mudflats (intertidal mudflats), or wetlands (inland wetlands), and AREA is ATLMN (northern Atlantic coast), ATLMS (southern Atlantic coast), or NSea (North Sea coast)
• mubrHAB_propmatrixbr_Br_grW = difference in habitat use proportions (log odds) of a given breeding habitat type (HAB) between Curlews using western European and the reference breeding area (northeastern Europe), where HAB is grass (grasslands), mudflats (intertidal mudflats), or wetlands (inland wetlands), and W is western European breeding ground
• mustHAB_propmatrixst_Non_grAREA = difference in habitat use proportions (log odds) of a given stopover habitat type (HAB) between Curlews using a given (AREA) and the reference non-breeding area (central Atlantic coast), where HAB is grass (grasslands), marshsalt (salt marshes), mudflats (intertidal mudflats), or wetlands (inland wetlands), and AREA is ATLMN (northern Atlantic coast), ATLMS (southern Atlantic coast), or NSea (North Sea coast)
• mustHAB_propmatrixst_Br_grW = difference in habitat use proportions (log odds) of a given stopover habitat type (HAB) between Curlews using western European and the reference breeding area (northeastern Europe), where HAB is grass (grasslands), mudflats (intertidal mudflats), or wetlands (inland wetlands), and W is western European breeding ground

Scripts

R-script_COE_habitat_v3.R Code for preparing the datasets, fitting the models, and plotting the figures for the analysis of the effects of habitat use on nest success in the Eurasian Curlew.

Scripts below are used for the assessment of nest success based on the analysis of GPS tracking data

Nest_succ_user.R Code to run the workflow for the assessment of nest success based on the analysis of GPS tracking data: loading packages and data, setting parameters for the analyses, running all the analyses, and saving the outputs of the analyses and plots for inferences.

Nest_succ_dt_dbscan.R Data preparation and DBSCAN.

Nest_succ_helpers.R Helper functions and code for creating folders with the outputs

Nest_succ_main.R Main code for the assessment of nest success

Nest_succ_dist.R Nest distance analysis

Part II - assessment of nest success based on the analysis of GPS tracking data

Analysis of bird GPS tracking data in the breeding season, allowing to identify nesting attempts and assess nest success. Birds are analyzed one by one because the user needs to adjust the parameters based on the outcomes of individual steps of the workflow.

Documentation

The workflow run by Nest_succ_user.R can export bird tracking data with identified clusters of possible nest locations, summary table of the nest cluster analysis, and coordinates of the possible nests. Plots depicting the results of the analysis of nest clusters and distance to nest analysis are exported. The following parameters are defined before running the analyses:

bird_name = species name (shortened for usage in file names and plot titles)

bird_ID = ID of the analysed individual, usually loaded from a list of the assessed bird species

dt_nest = dataset including tracking data. Created by the classification application "Migration and stationary sites", available at "https://www.moveapps.org/apps/browser/164b6064-206a-4e44-9e6c-8516a288e18a"

Year_br = breeding season for the analysis, a single year

thin_time = thinning interval for the data (in min), used when getRecursions() function has not enough memory (default = 10)

cl_eps = epsilon (in m), the maximum distance between two points to be considered neighbours in DBSCAN, usually values of 2, 5, 10, or 20 m

cl_pts = minimum number of points in the cluster (default = 100)

tot_w = total weight for DBSCAN weighted by residence time (in hours), usually between 1 and 30 hours

bcp_thres = minimum posterior probability to consider given change points in the analysis (default = 0.3)

radius = radius (in m) used for calculation of the pattern by recurse() (default = 20)

run_recurse = run trajectory and revisits analysis, disable it if speed of computation is preferred (default TRUE)

dbscan_time = time-weighted DBSCAN is run (TRUE), or a classical DBSCAN (FALSE)

first_date = the first date of an assessed period, format YYYY-MM-DD

date_range = length of the assessed period (in days), starting from the first_date

cl_no = number of the cluster assessed, see "clus_summary" created within the workflow

Interpretation of the plots

See the Supporting Information related to the article for details on the interpretation of the plots generated by the algorithm.

Usage notes

R is required to run all the scripts, as well as to load the stored models (.rds files). The models can be loaded using the base R function readRDS(), but we recommend following the code provided in the R scripts in order to assign names to the models anticipated within the workflow.

Microsoft Excel or a similar spreadsheet editor can be used to open the data sets. All CSV files are semicolon-delimited.