Data and code from: Cost of migration increased during an outbreak of highly pathogenic avian influenza virus
Data files
Mar 12, 2026 version files 10.76 MB
-
endpt_tstart-tstop-1locday.csv
10.72 MB
-
mapping_figures_DRYAD.R
7.17 KB
-
mortality_metadata.csv
12.78 KB
-
one-record-fate.csv
7.04 KB
-
README.md
3.36 KB
-
seasonal_survival_DRYAD.R
11.96 KB
Abstract
Migration is thought to be costly, such that challenges faced during migration likely affect how birds migrate through direct selection on migratory behavior. Survival throughout the annual cycle and conditions that affect the survival costs of migration are therefore critical to our understanding of the causes and consequences of migration. Estimating daily survival throughout the annual cycle of migratory birds has only recently become possible. We used over a decade of GPS tracking data from 186 migratory Buteo lagopus to quantify causes of mortality as well as demographic, seasonal, and spatial variation in daily survival and how survival consequences of migration changed during an outbreak of highly pathogenic avian influenza virus. The cost of migration via direct mortality varied between avian influenza and baseline study periods. Surprisingly, mortality risk during typical spring migrations was lower than in all other seasons, but mortality risk during the avian influenza outbreak was higher during both spring and fall migrations. Additionally, we found increased mortality risk in females during the avian influenza outbreak but no sex differences during baseline periods. Hence, the relative cost of migration was altered by the virus. During baseline periods, mortality risk was highest during late fall migration and winter, coinciding with increased anthropogenic-related mortalities at lower latitudes. However, the number of deaths caused by avian influenza (n = 11) in a single year equaled the total number of deaths caused by anthropogenic-related factors (n = 11; primarily collision trauma and persecution) across all 12 study years. Overall, our study highlights how added stressors such as disease outbreaks and anthropogenic factors may exacerbate existing seasonal constraints by directly increasing mortality risk.
Dataset DOI: 10.5061/dryad.j3tx95xvd
Description of the data and file structure
We captured and attached GPS transmitters to Buteo lagopus throughout their entire North American distribution from 2014–2026. Our survival analyses were based on 186 GPS-tracked birds for which we gathered data at capture to accurately determine age and sex.
Files and variables
File: mortality_metadata.csv
Description: Mortality metadata used to create Figure 1.
Variables
- tagid: transmitter (i.e., individual) ID.
- fate: final 4-level disposition of each individual. Definitions can be found within the manuscript itself.
- mortdate: date of mortality. NA if censored.
- last_lat: latitude of final (mortality) location.
- last_long: longitude of final (mortality) location.
- cod: exact cause of death
- cod_specific: caused of death used for Table 1.
- tagtype: transmitter type. PTT = satellite; GSM = cellular.
- mort-AI_year: did mortality occur during the 12-mo avian influenza study period (yes) or not (no)
File: one-record-fate.csv
Description: Used to create figures.
Variables
- tagid: transmitter (i.e., individual) ID.
- tstop_date: date of final location.
- AI_year: occurring with the avian influenza period (yes) or not (no).
- firstdate: date of first tracking location.
- endpt: disposition at tstop_date. 0 = censored; 1 = dead.
- cod_broad: broad cause of death used for plotting Figure 2B-C.
- tagtype: transmitter type. PTT = satellite; GSM = cellular.
File: mapping_figures_DRYAD.R
Description: R code used to create Figures 1A-B and Figures 2B-C.
File: seasonal_survival_DRYAD.R
Description: R code used for survival analyses and creating all other figures.
File: endpt_tstart-tstop-1locday.csv
Description: daily GPS location information for all individuals used in survival analyses.
Variables
- tstart_long: longitude on tstart_date.
- tstart_lat: latitude on tstart_date.
- tstop_long: longitude on tstop_date.
- tstop_lat: latitude on tstop_date.
- tagid: transmitter (i.e., individual) ID.
- tstart_date: date of tstart.
- tstop_date: date of tstop.
- dayofyear: tstart day of year (1-366).
- season: 4-level seasonal variable, defined within manuscript.
- firstdate: date of first tracking for each tagid.
- tstart: used for survival analyses. firstdate = day 0. Units are days since tracking began.
- tstop: used for survival analyses. Units are days since tracking began.
- endpt: disposition on tstop for each record. 0 = censored (alive or unknown); 1 = dead.
- tagtype: transmitter type. PTT = satellite; GSM = cellular.
- sex: sex of tagid. M = male; F = female.
- age: age during survival interval using cycle-based terminology. e.g., 2 = second-cycle. 4 = max age authors could assign at capture (all 4+ ages assigned to 4).
- minagedays: minimum age in days during survival interval.
Code/software
All survival analyses were conducted within the R statistical environment (v 4.5.0) using the coxph and Anova functions within the survival (v 3.8-3) and car (v 3.1-3) packages, respectively.
Note: Blank cells and N/A are used to indicate inapplicable or missing data.
