A Meta‐Analysis of Band Reporting Probabilities for North American Waterfowl

Abstract

Knowledge of band reporting is important for converting band encounter data into estimates of harvest probabilities, which can then be used to assess harvest management goals or estimate population size and other vital rates. Historical estimates of band reporting probabilities have come from reward‐band studies or joint analysis of band recovery and harvest survey data, but there are long gaps between estimates, and most studies have focused exclusively on mallards (Anas platyrhynchos). We compiled 337 estimates of band reporting probabilities for North American waterfowl and conducted a Bayesian state‐space analysis to provide a continuous time series of estimated reporting probability from 1948 to 2010. Band reporting probability increased sharply between 1996 and 2000 when toll‐free phone numbers were added to band inscriptions and agencies implemented electronic methods for band reporting, but our analysis also identified gradual long‐term trends in reporting probability throughout the time series. We found little evidence for among‐species variation in reporting probability, but a few species that are widely regarded as trophies by waterfowl hunters (canvasbacks [Aythya valisineria], redheads [Aythya americana], and northern pintails [Anas acuta]) had higher historical reporting probabilities than mallards. We also found little evidence of geographic variation in reporting probabilities, although we confirmed lower reporting probabilities for eastern Canada. We recommend using our estimates of band reporting probabilities and their variances as informed priors in future analyses of band recovery data to fully embrace uncertainty about how this parameter affects estimates of other population parameters. Retrospective studies using parts collection data are needed to explore potential among‐species variation in reporting probabilities during recent decades. © 2019 The Wildlife Society.

Variation in Survival and Harvest Rates in Florida Mottled Duck

“…During our study, banded samples averaged approximately 600 ducks/year, similar to the banding effort during 1983; however, the higher band-reporting rates (Boomer et al 2013, Arnold et al 2020 and greater numbers of overall recoveries under liberalized season frameworks during the time frame we evaluated effectively increased the number of band recoveries per total birds banded, which reduced variances in our model estimates. We also had bandreporting probabilities (Boomer et al 2013, Garrettson et al 2014, Arnold et al 2020) estimated from reward bands and uncomplicated by band solicitation over the time frame we considered (P. Garrettson, unpublished data). This allowed us to calculate robust harvest probabilities for Florida mottled ducks, whereas previous researchers (Johnson et al 1984(Johnson et al , 1995 could report only unadjusted band recovery rates because of the variation in band solicitation rates over time and space.…”

“…We used survival estimates from the Burnham model and recovery estimates from the equivalent Brownie model. To estimate harvest probabilities, we expanded recovery probabilities by a reporting probability of 0.757 (SD = 0.027), which is the average annual value for waterfowl in the United States from Arnold et al (2020;2000-2010 estimates and log-linear predictions of these data for 2011-2013). south; for juveniles in the north, our estimates were lower.…”

“…We modeled survival (S) and recovery (f ) probabilities as a function of age (juvenile and adult), banding area (north and south), sex (male and female), and temporal effects (cubic [T 3 ] trend on year [e.g., T 3 = intercept + T + T 2 + T 3 ]). Harvest probabilities were band recovery probabilities (f ) from the highest-ranked Brownie model, adjusted with bandreporting probabilities for waterfowl from Arnold et al (2020;2000-2010 estimates and log-linear predictions of these data for 2011-2013).…”

“…To estimate harvest probabilities ( h ), we expanded recovery probabilities ( f ) by annual band‐reporting probabilities (λ) for waterfowl in Arnold et al (2020) and used the delta method to estimate variances (Powell 2007). Arnold et al (2020) provided band‐reporting probabilities through 2010, so we used predicted mean values for 2011–2013 from a log‐linear regression of the 2000–2010 estimates with annual variances calculated from the average coefficient of variation during 2000–2010. We wanted to learn whether the addition of live recaptures would allow us to fit more parameterized models, principally a fully time‐specific effect (t), on the survival and recovery parameters.…”

Effect of Pre‐Harvest Mortality on Harvest Rates and Derived Population Estimates

“…We estimated abundance of midcontinent lesser snow goose goslings and adults alive at the time of marking using Lincoln's (1930) method (Alisauskas et al 2009, 2011, 2014). Estimates of band reporting () were from a metanalysis initially reported by Alisauskas et al (2014) and recently updated by Arnold et al (2020). We used the same general approach as Alisauskas et al (2011) for Lincoln's , except that we used the abundance estimator provided in Alisauskas et al (2014), which addresses bias stemming from low numbers of marked or recovered animals.…”