Skip to main content
Dryad

Data from: Evidence for seasonal compensation of hunting mortalities in a long-lived migratory bird

Cite this dataset

LeTourneux, Frédéric et al. (2024). Data from: Evidence for seasonal compensation of hunting mortalities in a long-lived migratory bird [Dataset]. Dryad. https://doi.org/10.5061/dryad.x95x69pt7

Abstract

Understanding whether hunting mortality is additive to or compensated by other mortality sources is at the heart of managing harvested populations. Long-lived species are expected to exhibit hunting mortality additive to other sources of mortality, making them ideal candidates for population management through sport harvest. Previous studies on these processes have focussed on density-dependent natural mortality compensating for hunting mortality, but when harvest occurs in distinct periods of the year, heterogeneity in hunting vulnerability between individuals could also lead to compensatory mortality between these periods. We explore this new idea using the case of the greater snow goose (Anser caerulescens atlantica), a harvested species whose population became overabundant in the late 20th century. To control this population, wildlife agencies liberalized hunting regulations with unprecedented actions such as special hunting seasons implemented in spring 1999 in Canada and in winter 2009 in the USA. To determine the relative impact of each measure on survival, we estimated survival of adult geese on a seasonal basis using 30 years of capture-mark-reencounter data in a joint live-and-dead-encounter multievent model. We also used this quasi-experimental set-up to evaluate possible compensation in hunting mortality between seasons. We found that both special hunting seasons decreased goose survival in the seasons and periods in which they were implemented. However, survival increased during the spring hunting season after the establishment of the special winter hunting season in the USA in 2009. There was a negative relationship between annual spring and winter mortalities, suggesting that the increase in hunting mortality in winter was compensated by a reduction in spring mortality after 2009.

Synthesis and applications: To our knowledge, we report the first documented instance of hunting mortality in one season being compensated by a reduction in hunting mortality in a subsequent season. We suggest that heterogeneity in hunting vulnerability among individuals, possibly linked to the presence of juveniles, may explain this phenomenon. A better knowledge of seasonal patterns and relationships between mortality components is needed to improve our understanding of population dynamics and management of harvested populations.

README: Data for: Evidence for seasonal compensation of hunting mortalities in a long-lived migratory bird.

https://doi.org/10.5061/dryad.x95x69pt7

This repository contains all code needed to reproduce analyses of LeTourneux et al. 2024, manuscript entitled "Evidence for seasonal compensation of hunting mortalities in a long-lived migratory bird.". Data for all scripts and to reproduce all figures is provided in the repository. The compiled capture-recapture dataset to reproduce the E-surge analyses and the most parcimonious model obtained from this analysis is also provided.

All capture-mark-recapture analyses were conducted in the software E-SURGE and all figures and some analyses to generate the figures were made with R. The script to generate figures from E-SURGE model results are present in '*.R' files. There is one script for the figures in the main text, and each appendix has its own stand alone script for the figures in each appendix.

E-SURGE model output files are named in a self-explanatory way, e.g., "M6_this_study.csv" contains parameter estimates from model M6 of this study saved to a csv2 format (i.e., column delimiter=';', decimal delimiter=','). Parameter estimates are on the [0-1] scale, and are provided along with their SE and 95% confidence intervals. For files containing betas, betas are provided on the real scale (]-Inf,+Inf[) along with their SEs and 95% confidence intervals. Some files also contain model outputs from LeTourneux et al. 2022 J. Appl. Ecol. (doi: 10.1111/1365-2664.14268), for means of comparison.

The code for our most parsimonious E-surge model is provided in the file "E-Surge_code_model_M10.docx". Some basic knowledge of E-Surge is needed to fit the model, however, a detailed explanation is given on how to set the initial values as this step is not intuitive, even for experienced E-Surge users.

Description of the data and file structure

1. LeTourneux_et_al_2024_capture_histories.prn:

File saved to the PRN format, one of the few recognized by E-Surge. Can be read with any text editor. This file contains unique seasonal capture histories of greater snow geese captured and released on Bylot Island between 1990 and 2019. The first line contains the number of individual capture histories (first number) and the number of columns in the data (2nd number). Below the two lines containig a '$' are the capture histories. The first 119 columns are the seasonal capture-recapture data. The first column is a summer banding occasion. A '1' represents an individual initially captured, fitted with a legband, and released (if first encounter) and a recapture of that individual at subsequent summer occasions. A '2' at initial encounter represents an individual banded, marked with a lagband and collar and released. '2's then represent a live observation of that individual. A '1' in those lines where individuals were initially marked with a neck collar represent an individual that was recaptured alive without its collar. A '3' represents an individual that was recovered dead by a hunter either with or without its collar. The last two colums represent the number of individual males sharing this capture history (column #120) and the number of females sharing that capture history (column #121).

2. harvest_rate_GSG.csv:

File saved to csv2 format (column delimiter=';', decimal delimiter=','). Annual harvest rates of adult greater snow geese between 1990 and 2019. Annual harvest rates were obtained by dividing the number of adult greater snow geese harvested annually (estimated by national hunter surveys) by the fall population size (estimated by an aerial photographic census; details are provided in Appendix S2 of LeTourneux et al. 2022; J. Appl. Ecol. doi: 10.1111/1365-2664.14268).

Column description:

  • Year: year;
  • HR: harvest rate (adults only), see details in main text for calculation of harvest rate.

3. Model output files I (M6_this_study.csv, M10_this_study.csv, M14_LeTourneux_et_al_2022.csv, M17_LeTourneux_et_al_2022.csv):

Files saved to csv2 format (column delimiter=';', decimal delimiter=',').

Column description:

  • Parameters: Refers to the different matrices and steps. IS: Initial states, C: 1st transition matrix (collar loss), S: 2nd transition matrix (survival); E: 1st event matrix, E2: 2nd event matrix.
  • From: State of origin (i.e., matrix lines). See Appendix 1 in this study and LeTourneux et al. 2022 for details.
  • To: State of arrival (i.e., matrix columns)
  • Time: Capture occasion.
  • Age: Age class, used to model change in collar loss probabilities through time (See LeTourneux et al. 2022 for details).
  • Group: Grouping factor for the data. Not used in this study.
  • Step: Matrix step (information already included in 'Parameters').
  • Estimates: Parameter estimates on the [0-1] scale.
  • CI-: Lower 95% confidence interval of parameter.
  • CI+: Upper 95% confidence interval of parameter.
  • SE: Standard error of parameter.

4. Model output files II (M6_this_study_betas.csv, M10_this_study_betas.csv):

Files saved to csv2 format (column delimiter=';', decimal delimiter=',').

Column description:

  • Index: Line number
  • Value: beta of parameter on the real scale ]-Inf,+Inf[.
  • CI-: Lower 95% confidence interval of beta.
  • CI+: Upper 95% confidence interval of beta.
  • SE: Standard error of beta.
  • beta: Description of beta represented by each line. Mostly used to extract betas for script and figures
  • IS1: Prob. of being ringed and weakly observable (Index=1); Prob. of being ringed and highly observable (Index=2); Prob. of being collared and highly observable in summer (Index=3);
  • IS2: Prob. of being collared and highly observable during fall
  • IS3: Prob. of being collared and highly observable during winter
  • IS4: PProb. of being collared and highly observable during spring
  • perte_col: Prob. of losing collar
  • S_ete: Survival ringed only summer
  • S_fall: Survival ringed only fall
  • S_wint: Survival ringed only winter
  • S_spr: Survival ringed only spring
  • col_ete_d2: Collar effect on survival summer hunting period 2
  • col_ete_d3: Collar effect on survival summer hunting period 3
  • col_fall_d2: Collar effect on survival fall hunting period 2
  • col_fall_d3: Collar effect on survival fall hunting period 3
  • col_wint_d2: Collar effect on survival winter hunting period 2
  • col_wint_d3: Collar effect on survival winter hunting period 3
  • col_spr_d2: Collar effect on survival spring hunting period 2
  • col_spr_d3: Collar effect on survival spring hunting period 3
  • pcap_heterog: Effect of lowly heterogeneous group on capture probability
  • pcap_sex: Effect of sex on pcap heterogeneity (beta female)
  • pcap: capture probability (1/summer occasion)
  • recov_fall1: recovery probability for fall during hunting period 1
  • recov_fall2: recovery probability for fall during hunting period 2
  • recov_fall3: recovery probability for fall during hunting period 3
  • recov_winter1: recovery probability for winter during hunting period 1
  • recov_winter2: recovery probability for winter during hunting period 2
  • recov_winter3: recovery probability for winter during hunting period 3
  • recov_sp_d2: recovery probability for spring during hunting period 2
  • recov_sd_d3: recovery probability for spring during hunting period 3
  • col_obs: collar observation probability (Index=174-202: summer, 204-233: fall, 235-264: winter, 266-294: spring)
  • het_obs_summer: effect of lowly observable group on observation probability for summer
  • het_obs_fall: effect of lowly observable group on observation probability for fall
  • het_obs_winter: effect of lowly observable group on observation probability for winter
  • het_obs_spring: effect of lowly observable group on observation probability for spring
  • band_recov_p1: effect of wearing a legband on recovery probability hunting period 1(vs. collars)
  • band_recov_p2: effect of wearing a legband on recovery probability hunting period 2(vs. collars)
  • band_recov_p3: effect of wearing a legband on recovery probability hunting period 3(vs. collars)
  • t: year (for survival estimates only)

Sharing/Access information

Links to other publicly accessible locations of the data:

Code/Software

All files ending in '*.R' can be run using the R statistical environment. All scripts are thoroughly commented and were tested and run on June 5th 2024 with latest package versions and R version 4.4.0.

The file 'E-Surge_code_model_M10.docx' contains all the information needed to fit the analysis from model M10 in the E-Surge software.

Methods

This capture-recapture dataset of greater snow geese was collected on Bylot Island, Nunavut (73N, 80W) between 1990 and 2019. Geese are captured at the end of the breeding season (August), when they are moulting and flightless. Please refer to the published manuscript for further details on the methods used and to the readME file for details on files provided along with this dataset.

Funding

Canada First Research Excellence Fund

Sentinelle Nord

Arctic Goose Joint Venture

Canadian Wildlife Service

U.S. Fish and Wildlife Service

Atlantic Flyway Council

Polar Knowledge Canada

Natural Sciences and Engineering Research Council

ArcticNet

Université Laval

Polar Continental Shelf Program (PCSP, Natural Resources Canada)