Subpopulation contributions to a breeding metapopulation of migratory Arctic herbivores: Survival, fecundity, and asymmetric dispersal

Estimates of demographic parameters for lesser snow geese (Anser caerulescens caerulescens) have become critical to understand ecosystem change in northern Canada. Exponential increase in abundance has produced hyperdensities of these herbivores that can affect Arctic ecosystem stability through intense foraging. Increased and sustained marking of individually-identifiable lesser snow geese over their breeding distribution now permits joint estimation of local vital rates and movement probabilities among widely scattered subpopulations. We used multi-state models, including an unobservable state, with live captures from 5 subpopulations and dead recoveries to estimate annual probabilities of (i) survival, (ii) capture, (iii) reported mortality and (iv) movement to other subpopulations, as well as derived estimates for probabilities of site fidelity and harvest. Our dataset included 144,754 captures of 139,177 lesser snow geese marked with metal legbands, from 2006 to 2015, of which 5,542 were recaptured near breeding sites and 9,709 were recovered dead in North America. The best model supported variation in survival by subpopulation and age, with additive effects of subpopulation, age and sex on movement probability. Male breeding dispersal was greater than by females, and juvenile geese were more likely to move than adults. Strong northeastward geographic asymmetry in the probability of breeding movement was consistent with an eastward shift in wintering distribution observed in hunter recoveries. Mean annual survival ranged from 0.79-0.94 for adults, and 0.16-0.47 for juvenile geese, with a strong negative relationship between regional adult and juvenile survival. Harvest probabilities were all ≤ 0.03 for adult and ≤ 0.06 for juvenile geese, suggesting little influence from direct anthropogenic exploitation. Metrics for subpopulation persistence and contributions of each to the metapopulation suggested declines in all but one subpopulation, and a declining midcontinent population overall. Our study highlights the importance of all subpopulation demographic parameters as modulators of persistence at both local and range-wide population dynamics.In addition to the multi-state capture-recovery data set as input data for Program MARK (file 1 "enchist2.inp"), consult the survival portion of the Parameter Index Matrix (PIM) (file "Parameter Index Matrix (PIM) from Global Model (survival parameters only).JPG"), as well as further constraints imposed by manipulating the Design Matrix (DM) (file "Design Matrix (DM) Top Model.xlsx") for additional guidance.Geese were captured en masse either as goslings (Age: HY, Hatch Year) close to fledging or as geese with adult plumage (Age: AHY, After Hatch Year) from 2006 to 2015 in Canada's Arctic and Subarctic. Captures were with the use of a helicopter and ground crews to locate and guide target groups into portable corral traps. Geese were marked with metal legbands only, no neckbands, at each of 5 subpopulations. Subpopulations are QMG (State Q, Queen Maud Gulf), SOU (State S, Southampton Island), BAF (State B, Baffin Island), LPB (State L, La Perouse Bay), and JAM (State J, James Bay). Number of birds banded in each year for each subpopulation, as well as year of recovery of dead birds (2006-2016), were supplied by the Bird Banding Laboratory, Laurel, Maryland, USA. Recaptures of birds already marked with legbands at each subpopulation were recorded. Recaptures occurred only between years, so all recaptures were of adult (AHY) geese only, whether marked initially as goslings (HY) or adults (AHY). Encounter histories of 139,177 were summarized in the Live-Dead format, with a pair of columns for each year. The first column in a pair denotes captures and the second column denoting dead recoveries within a year of captures. Captures in the first column of each pair were coded with the letter corresponding to the geographic state of capture (Q,S,B,L,J) , otherwise zero in not captured. Recoveries of dead geese in the second column of each pair were coded with a 1, otherwise zero if not recovered. The last four columns in "enchist2.inp" represent whether each capture hisotry was of an adult males (AM), adult female (AF), juvenile male (JM), or juvenile female (JF), respectively. An "unobservable state" was set as "U" when importing the encounter history as input data and selecting "Multi-state for Live and Dead Encounters" for Program MARK. An input data summary of the information in "enhist2.inp" can be found in the file "m-ARRAY.txt". Age structure in the global and other models was imposed in Program MARK with the Parameter Index Matrix (PIM) so that recoveries of birds marked as HY geese informed juvenile parameters, but also adult (AHY) parameters if recaptured or recovered as adults; sex was structured as a group effect (see file 3 "Parameter_Index_Matrix_(PIM)_from_ Global_Model_(survival_parameters_only).JPG"). Additional constraints were imposed to model additive effects between age, sex, state using the Design Matrix (DM) of Program Mark (see file 4 "Design_Matrix_(DM)_Top_Model.csv" as an example). Nonidentifiable paramaters resulted form situations where there were no captures of geese from which to estimate parameters. Nonidentifiable parameters were set to zero; these included male and female HY survival in states S and U, capture probabilies for all 4 age-sex cohorts in state U, transition of male and female HY geese from state S, transition of male and female HY geese from state U, and reported mortality for male and female HY geese in States S and U