Ecological Archives E093-171-A7
Alexandre Robert, Vítor H. Paiva, Mark Bolton, Frédéric Jiguet, and Joël Bried. 2012. The interaction between reproductive cost and individual quality is mediated by oceanic conditions in a long-lived bird. Ecology 93:1944–1952. http://dx.doi.org/10.1890/11-1840.1
Appendix G. Details of comparisons among final capture–recapture models (four-state models) in the Monteiro's Storm-Petrel (Oceanodroma monteiroi) population monitored at Praia Islet (Azores, Portugal) over the period 1993–2010.
Environmental and population covariates considered in best three-state models (see Appendix D) were used in models with four states (successful breeders, unsuccessful breeders, nonbreeders, and absent). The analysis brought strong evidence that time and covariate effects on breeders survival (Table G1) and transition (Table G2) were mostly due to temporal variance in unsuccessful breeder survival and unsuccessful breeder→nonbreeder transition.
Table G1. Model selection for survival parameters of the 'successful breeder' and 'unsuccessful breeder' states, with respect to year to year variations, breeding success in the previous year (BS_{lag}) and sea surface temperatures within a 50 km radius from the colony (SST_{50}, 2nd qtr.). Model selection was performed according to Akaike's criterion corrected for small sample size (AIC_{c}). ΔAIC_{c} is the difference in AIC_{c} between each model and the best model. The AIC_{c} weight (w_{i}.) is the amount of evidence in favor of a model. Dev and n.p. are respectively the deviance and the number of estimated parameters for each model. In all presented models, survival rates of absent and nonbreeder are assumed constant, recapture rates vary additively between breeding states and years, and transition rates vary according to age and breeding state (see Table D1, Mod. A1 of Appendix D for detailed structure). ^{*} : parameters assumed equal among states.
Model name |
Successful breeder |
Unsuccessful breeder |
AIC_{c} | ΔAIC_{c} | w_{i}. | n.p. | Dev. |
Mod. 1 | constant |
BS_{lag} + SST_{50} (2nd qrt.) |
4470.00 | 0.00 | 0.639 | 38 | 2631.29 |
Mod. 2 | constant |
BS_{lag} × SST_{50} (2nd qrt.) |
4472.11 | 2.11 | 0.223 | 39 | 2631.29 |
Mod. 3 | constant | BS_{lag} | 4473.70 | 3.70 | 0.101 | 37 | 2637.10 |
Mod. 4 | constant |
SST_{50} (2nd qrt.) |
4475.75 | 5.75 | 0.036 | 37 | 2639.15 |
Mod. 5 | BS_{lag} | time | 4484.91 | 14.91 | 0 | 53 | 2614.26 |
Mod. 6 | constant | time | 4485.54 | 15.54 | 0 | 51 | 2619.19 |
Mod. 7 | constant | constant | 4485.68 | 15.68 | 0 | 35 | 2653.29 |
Mod. 8 | time | BS_{lag} | 4489.45 | 19.45 | 0 | 53 | 2618.79 |
Mod. 9 |
SST_{50} (2nd qrt.) |
constant | 4489.66 | 19.66 | 0 | 37 | 2653.06 |
Mod. 10 | constant ^{*} | constant ^{*} | 4494.77 | 24.78 | 0 | 34 | 2664.48 |
Mod. 11 |
SST_{50} (2nd qrt.) |
time | 4495.27 | 25.27 | 0 | 53 | 2624.62 |
Mod. 12 | BS_{lag} | constant | 4495.32 | 25.32 | 0 | 37 | 2658.72 |
Mod. 13 | time |
SST_{50} (2nd qrt.) |
4503.91 | 33.91 | 0 | 53 | 2633.26 |
Mod. 14 | time | constant | 4505.49 | 35.49 | 0 | 51 | 2639.13 |
Mod. 15 | time | time | 4516.70 | 46.7 | 0 | 67 | 2615.61 |
Table G2. Model selection for transition parameters among breeding states (BS: successful breeder; BU: unsuccessful breeder; NB: nonbreeder), with respect to year to year variations, breeding success in the previous year (BS_{lag}) and sea surface temperatures within a 1500 km radius from the colony (SST_{1500}, 1st quarter). Model selection was performed according to Akaike's criterion corrected for small sample size (AIC_{c}). ΔAIC_{c} is the difference in AIC_{c} between each model and the best model. The AIC_{c} weight (w_{i}.) is the amount of evidence in favor of a model. Dev and n.p. are respectively the deviance and the number of estimated parameters for each model. In all models survival rates are constrained as in Mod. 1 of Table G1. Year and covariate effects on transition parameters are only applied on transition rates of the older age class. In all models, long term transitions of individuals marked as chicks and adults are assumed equal. ^{*} : transitions assumed equal among states.
Model name | Transition rates | AIC_{c} | ΔAIC_{c} | w_{i}. | n.p. | Dev. | |||||
ψ_{UB→NB} | ψ_{SB→NB} | ψ_{SB→UB} | ψ_{UB→SB} | ψ_{NB→UB} | ψ_{NB→SB} | ||||||
Mod. 1 |
BS_{lag} × SST_{1500} (1st qrt.) |
constant | constant | constant | constant | constant | 4442.36 | 0.00 | 0.189 | 42 | 2595.20 |
Mod. 2 |
Bs_{lag} + SST_{1500} (1st qrt.) |
constant | constant | constant | constant | constant | 4443.25 | 0.89 | 0.122 | 41 | 2598.21 |
Mod. 3 |
SST_{1500} (1st qrt.) |
constant | constant | constant | constant | constant | 4447.84 | 5.48 | 0.012 | 40 | 2604.92 |
Mod. 4 | BS_{lag} | constant | constant | constant | constant | constant | 4462.23 | 19.87 | 0 | 40 | 2619.31 |
Mod. 5 | time | constant | constant | constant | constant | constant | 4464.82 | 22.46 | 0 | 54 | 2592.02 |
Mod. 6 | constant | constant | constant | constant | constant | constant | 4470.00 | 27.64 | 0 | 38 | 2631.30 |
Mod. 7 | constant | constant | constant | time | constant | constant | 4478.58 | 36.22 | 0 | 54 | 2605.77 |
Mod. 8 | constant | constant | constant | constant | time | constant | 4482.94 | 40.58 | 0 | 54 | 2610.13 |
Mod. 9 | constant ^{*} | constant ^{*} | constant | constant | constant | constant | 4487.00 | 44.64 | 0 | 37 | 2648.20 |
Mod. 10 | constant | constant | time | constant | constant | constant | 4492.40 | 50.04 | 0 | 54 | 2619.59 |
Mod. 11 | constant | constant | constant | constant | constant | time | 4497.04 | 54.68 | 0 | 54 | 2624.24 |
Mod. 12 | constant | time | constant | constant | constant | constant | 4503.14 | 60.78 | 0 | 54 | 2630.34 |
Mod. 13 | time | time | time | time | time | time | 4585.57 | 143.21 | 0 | 135 | 2527.71 |