This work analyzes satellite tracking data of five migratory herbivorous waterfowl species in East Asia to investigate if and how herbivorous waterfowl track multiple resource waves during spring migration.

Description of data and file structure:

R Scripts:

• 01 - Analyzing migration parameters.R: R script that reads in the output of identified sites (calculated in SaTScan) and calculates migration parameters.
• 02 - Investigating resource utilization in migration.R: R script that analyses habitat use intensity on farmlands and natural habitats across time, and investigates habitat switch before and after the emergence of vegetation at core stopover sites in the Northeast China Plain.
• R scripts 03-06: Determine the bird arrival and resource availability relationship.
  03 - Testing differences in resource and migration timing.R: R script that shows resource and migration timing, and tests differences between the resource and migration timing in different regions for each species.
  04 - Resource selection function models.R: R script that executes models for each of five herbivorous waterfowl species in two regions.
  05 - Estimate optimal lags on resource availability.R: R script that reads in output of the model and estimates optimal lags on food availability/snowmelt conditions.
  06 - K-fold cross-validation.R: Cross-validation on the resource selection models.
  06b - K-fold.r: The source file input '06 - K-fold cross-validation.R' to run cross-validation.

Data:

Metadata

14 .csv files are included.

(1) siteinfo: identified wintering, stopover, and breeding sites

• species = abbreviated names of five herbivorous waterfowl species (GWFG, LWFG, TUBG, TS, SG)
• track = the identity of each track (track ID), named as BirdsID_Year
• BirdsID = the identity of each individual
• Year = the year of the track
• CLUSTER = the identity of identified sites for each track using the space-time permutation model in SaTScan
• LATITUDE = latitude of the site
• LONGITUDE = longitude of the site
• RADIUS = radius of the site (unit: km)
• START_DATE = Date of bird arrival at a stopover site
• END_DATE = Date of bird departure from a stopover site
• start = bird arrival - convert the date format into 'day of year' format
• end = bird departure - convert the date format into 'day of year' format
• site = site types (Wintering, Stopover, Breeding)

(2) PheEvents: Data used to generate Figure 2

• phe = the phenological events
  15 groups are included in this column:
  win_SOS, win_departure: vegetation emergence, departure at wintering sites
  ncp_25, ncp_50, ncp_75, ncp_100, ncp_SOS, ncp_arri, ncp_depar: 25%, 50%, 75%, 100% relative seed availability, vegetation emergence, arrival, departure at core stopover sites
  bre_25, bre_50, bre_75, bre_100, bre_SOS, bre_arri: 25%, 50%, 75%, 100% snowmelt, vegetation emergence, arrival at breeding sites
• Day = timing of phenological events (shown as day of year)

(3) WeeklyHabitatUse: Data used to generate Figure 3

• doy = day of bird occurrence
• habitat = Foraging habitat types (Farmlands, Natural - Grass/Shrub/Wetlands)
• percen_lc = Proportion of land cover in use at daytime
• week = week of bird occurrence (e.g., week 1 means 1-7 days of the year)

(4) HabitatSwitchNCP: Data used to investigate habitat switch before and after the emergence of vegetation in the Northeast China Plain and generate Figure S14

• group = during two different time windows ('Before'-from bird arrival to the emergence of vegetation; 'After'-from the emergence of vegetation to departure)

Tables (5) - (14): Input of resource selection function models. Data used to generate Figure 4, Figure S15, Table 1, Table S10 by R scripts 04-06.

Table names:
ncp_plant, ncp_x25, ncp_x50, ncp_75, ncp_100:  datasets for core stopover sites in the Northeast China Plain

bre_plant, bre_x25, bre_x50, bre_75, bre_100: datasets for breeding sites

Columns in tables:

• select = presence and absence (1-presence; 0-absence)
• Daysto = the number of days between resource availability/snowmelt and the bird arrival date.