Missing Release Data in Capture-Mark-Recovery Analyses: Consequences for Inference
Author:
Brook Rodney W.1, Dooley Joshua L.2, Brown Glen S.1, Abraham Kenneth F.1, Rockwell Robert F.3
Affiliation:
1. R.W. Brook, G.S. Brown, K.F. Abraham Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry, Wildlife Research and Monitoring Section, 2140 East Bank Drive, Peterborough, Ontario K9L 1Z8Present address of K.F. Abraham: 434 Manorhill Avenue, Peterborough, Ontario K9J 6H8 2. J.L. Dooley U.S. Fish and Wildlife Service, Division of Migratory Bird Management, Branch of Assessment and Decision Support, 1211 SE Cardinal Court, Suite 100 Vancouver, Washington 98683 3. R.F. Rockwell Division of Vertebrate Zoology, American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024
Abstract
Abstract
Demographic probabilities, such as annual survival and harvest probability, are key metrics used in research and for monitoring the health of wildlife populations and sustainability of harvest. For waterfowl populations, annual estimates of these probabilities come from mark-recovery analysis of data from coordinated banding operations. The Brownie model is the most commonly used parameterization for analyzing mark-recovery data from harvested species. However, if banded waterfowl are not released during a year of a multiyear banding operation, then estimating annual survival and recovery probabilities from a dead recovery model is a challenge. Due to coronavirus disease 2019, many wildlife monitoring efforts, including annual waterfowl banding programs, were canceled or reduced during 2020 and 2021, highlighting the need for wildlife managers to better understand the consequences of missing data on analyses and regulatory decisions. We summarized methods of model parameterization and use of alternative methods to explore the behavior of demographic parameter estimates when a year of release data was missing. Comparing constrained fixed-effect models (we set parameters during the missing year of data equal to parameters for years with release data) with random-effect models, we found that 1) bias of estimates during a year of missing release data was smaller when using a random-effect model, 2) the direction of the bias was unpredictable, but the expected range in bias could be generally known commensurate to the underlying variability in survival and recovery probabilities, and 3) potential bias was greatest if the missing year of releases occurred during the final year of a time series. We conclude that in some circumstances, various modeling approaches can provide reasonable estimates during a year of missing release data, particularly when underlying demographic parameters, or the parameter constrained in a model, vary little over time (e.g., adult survival in long-lived species), which would result in relatively little bias in the other estimated parameter (e.g., annual recovery probability). We also suggest that using alternative analytical techniques, such as random-effect models, may improve estimates for the demographic parameters of interest when release data are missing. Random-effect models also allowed us to estimate parameters, such as juvenile recovery probabilities, during the year of missing release data, which are not identifiable using standard modeling techniques. Where accurate and precise parameter estimation is important for making harvest management decisions and regardless of the model type or the data used, there is no analytical replacement for missing release data. We suggest that practitioners determine the potential consequences for missing data through simulation by using empirical data and simulated data with known demographic probabilities to determine the best actions to take for analyzing their capture-recovery data when release data are missing.
Publisher
U.S. Fish and Wildlife Service
Subject
Nature and Landscape Conservation,Animal Science and Zoology,Ecology,Ecology, Evolution, Behavior and Systematics
Reference46 articles.
1. Arnold
TW,
Afton
AD,
Anteau
MJ,
Koons
DN,
Nicolai
CA.
2016.
Temporal variation in survival and recovery rates of lesser scaup.
Journal of Wildlife Management80:
850–
861. 2. Arnold
TW,
Clark
RG,
Koons
DN,
Schaub
M.
2018.
Integrated population models facilitate ecological understanding and improved management decisions.
Journal of Wildlife Management.
82:
266–
274. 3. Barker
RJ.
1999.
Joint analysis of mark-recapture, resighting and ring-recovery data with age-dependence and marking-effect.
Bird Study46:
82–
91. 4. Brook
RW,
Leafloor
JO,
Abraham
KF,
Douglas
DC.
2015.
Density dependence and phenological mismatch: consequences for growth and survival of sub-arctic nesting Canada geese.
Avian Conservation and Ecology10:
1.
Available: https://doi.org/10.5751/ACE-00708-100101 5. Brownie
C,
Anderson
DR,
Burnham
KP,
Robson
DS.
1985.
Statistical inference from band-recovery data—a handbook. 2nd edition.
Washington, D.C.:
Fish & Wildlife Service, U.S.
Department of the Interior. Resource Publication No. 156. (see Supplemental Material, Reference S1).
|
|