Implications of Demographic Uncertainty for Harvest Management of North American Sea Ducks

Author(s): 
Koneff, Mark. D.
Dwyer, Chris
Zimmerman, Guthrie
Fleming, Kathy
Padding, Paul
Devers, Patrick
USFWS -Migratory Bird Management - Orono Maine
USFWS -Migratory Bird Management - Laurel MD.
USFWS -Migratory Bird Management - Hadley MA
USFWS -Migratory Bird Management - Sacremento CA
Publication Date: 
2016

In 2010, the Sea Duck Joint Venture (SDJV) identified the need for improved science support for harvest and habitat management of North American sea ducks.
In order to prioritize monitoring and research needs in support of harvest management, we applied a Prescribed Take Level (PTL) framework to assess the influence of uncertainty about sea duck demographic
parameters on comparisons of observed and allowable harvest estimates. We focused on 7 populations of North American sea ducks: the American subspecies of common eider (Somateria mollissima dresseri
), the continental populations of long-tailed duck (Clangula hyemalis) and white-winged scoter (Melanitta fusca), and eastern and western populations of black (M. americana) and surf scoter (M. perspicillata).
Prescribed Take Level (PTL) is an estimate of the allowable harvest of a population.Formulated as total harvest, calculation of PTL requires estimates of population size (Nt) and maximum growth rate (rmax), while formulation of PTL as a harvest rate requires only an estimate of rmax . We used a total harvest formulation of PTL for all populations, except common eider where banding data were sufficient to formulate PTL based on harvest rate. We defined rmax as the maximum growth rate achievable by a population in the absence of harvest under average environmental conditions. We derived rmax from the maximum finite growth rate (λmax) using an age-structured population projection matrix. In implementing the PTL framework we: (1) combined information from empirical studies and the opinions of experts to create probability distributions reflecting uncertainty in the individual demographic parameters needed to conduct the PTL; (2) used simulation to propagate that uncertainty into probability distributions of
allowable harvest for each species; (3) compared estimates of allowable harvest to observed harvest; and (4) evaluated the sensitivity of the comparison of allowable to observed harvest estimates to uncertainty in the parameters used to derive those estimates
.
We relied on a combination of published and unpublished data and estimates as well as the results of a formal expert elicitation to specify probability distributions for the parameters used in this asse
ssment: age-specific survival, fecundity (calculated from reproductive rates such as nest success, clutch survival, and breeding propensity, as well as harvest age ratios), fall population size, observed harvest (sport and subsistence), and for common eiders, observed harvest rate. The probability distributions reflected uncertainty about the true mean value of each demographic parameter for each population. We used Monte Carlo simulations to estimate rmax, allowable harvest, and observed harvest for each population. We then used linear regression to assess the sensitivity of the difference between allowable and observed harvest estimates to uncertainty in the component parameters of rmax, fall population size, and observed harvest. We identified populations at risk of overharvest by the proportion of simulations where
observed harvest exceeded allowable, and categorized demographic information needs into three levels of priority based on their uncertainty and their influence on the comparison of allowable
and observed harvest. Our literature search revealed a dearth of empirical data for most of the populations, and our effort to augment the empirical data by eliciting opinions from subject-
matter experts met with limited success. Accurate quantification of uncertainty was a crucial component of the assessment, and our results and conclusions below are conditional on adequate descriptions of
ii uncertainty for each parameter. In general, our allowable harvest (or harvest rate) estimates were very uncertain, much more so than the estimates of observed harvest. American Common Eider.
The median allowable harvest rate for American common eiders was –0.0009 (95% credible interval -0.0812; 0.0692). The percent of simulations where observed harvest rate
was less than allowable harvest rate was 20%. The comparison of observed and allowable harvest rates was most influenced by uncertainty in adult survival, as well as several components of fecundity including duckling survival, the ratio of juvenile to adult female wings in samples submitted by hunters (i.e., harvest age ratio), hatching success, and clutch size. Highest priorities for research and monitoring were estimates of age ratios and duckling survival.Eastern/Western Black Scoter.For eastern black scoters allowable harvest was 29,940 (807; 93,753), and the percent of simulations where observed harvest was less than allowable harvest was 52%. For western black scoter allowable harvest was 10,854 (-11,058; 37,219), and observed harvest was less than allowable harvest in 30% of the simulations. Adult survival was
highly influential for both populations but due to its low uncertainty was only a medium priority for research and monitoring. For eastern black scoters, the highest priority information needs were population size and duckling survival, while moderate priority needs, in addition to adultsurvival, included age ratio, and the proportion of hens first breeding at age 2. For western black scoters, 3 fecundity parameters were the highest priorities for research or monitoring: nest success, duckling survival, and harvest age ratio. Observed harvest was also categorized as a high priority information need though it was less influential on comparisons between allowable and observed harvest than the fecundity parameters.Eastern/Western Surf Scoter. For eastern surf scoters, the median allowable harvest of 23,149 (-9,308; 78,894) was less than the median observed harvest by approximately 15,000 birds, The percent of simulations in which observed harvest was less than allowable harvest was 25%. High priority information needs based on the sensitivity analysis were harvest age ratios,
nest success, and population size. Adult survival and differential vulnerability were classified as moderate information needs. For western surf scoters, the median allowable harvest was 14,354 (-61,985; 82,110). Observed harvest was less than allowable harvest in 59% of the simulationsAdult survival was most influential on comparisons of observed and allowable harvest and was the highest priority information need. Other high priority information needs included population size, clutch size, juvenile survival, and differential vulnerability.White-winged Scoter.Median allowable harvest was 13,054 (-68,824; 61,072). The
percent of simulations in which observed harvest was less than allowable harvest was 36%. Observed harvest was a high priority information need, although its influence on the harvest comparison (based on absolute slope) was less than the 4 parameters that were ranked as moderate priority information needs (differential vulnerability, nest success, hatching success, and adult survival) as a result of a larger relative uncertainty surrounding observed harvest.Long-tailed Duck. Median allowable harvest for long-tailed ducks was -48,966 (-202,663; 60,561). The percent of simulations in which observed harvest, 43,044 (32,151;
57,589), was less than allowable harvest was only 5%. Reproductive rate estimates for long-tailed ducks from the literature were very low compared to all populations other than common eiders. Population size was the only high-priority information need identified according to our criteria. Four parameters were categorized as moderate priority information needs: adult iii survival, nest success, proportion of first time breeders breeding at age 2, and survival of second-year birds.In general, this assessment highlights the high degree of uncertainty associated with simulated values of allowable harvest for all populations. We have particularly low confidence in the assessment for long-tailed ducks, and the assessment for American common eider may apply only to the segment of this population breeding in Maine and the Maritimes
. Comparisons of our simulated median values of intrinsic growth rates were lower than theoretical maximum values indicating that these populations were experiencing sub-optimal environmental conditions
, input parameter values were not consistent with growth unconstrained by density or harvest, or input parameter values were representative of only a subpopulation with lower growth potential than the entire population. Conclusions from this assessment include: (1) reductions in uncertainty in the high and moderate priority parameters could most significantly improve harvest inferences and decision
making; (2) uncertainty about overall fecundity had more influence on comparisons of allowable and observed harvest than adult survival or observed harvest, however, individual components of fecundity can be difficult to study at a population scale; (3) adult survival, though characterized by less uncertainty than individual components of fecundity, is a high priority information need given the sensitivity of growth rate and allowable take to this parameter, and (4) uncertainty about population size was a high priority information need for four of the six populations where it factored into the assessment. We recommend that the SDJV (1) prioritize research and monitoring efforts on the long-tailed duck and American common eider; (2) prioritize research and monitoring on high priority parameters identified for each population; (3) continue efforts to integrate the operating procedures and analysis of presently disparate breeding population surveys for sea ducks; and (4) conduct PTL assessments periodically, incorporating new information in order to revise priority information needs