Supplementary material
Table S1: Matrix patterns for dispersal, survival, initial states and events used in the model developed in E-SURGE (by convention, “from states” are in rows and “to states” are in columns). At the first occasion, which corresponds to ringing, encounter rates are fixed to 1.
Appendix S2: Description of Goodness Of Fit tests
In multistate modelling, a Goodness-of-fit (GOF) test exists for the Jolly Move Model (JMV) but no GOF test is currently available when using a mixture of recaptures and recoveries (Pradel et al. 2005). We therefore separately examined GOF tests of the live recaptures and of the dead recoveries as explained in (Duriez et al. 2009). The tests allow to detect an excess of immediate recoveries (i.e. in the year following ringing) and an excess of recoveries concentrated in some later years. All GOF tests were performed with the program U-CARE v2.5 (Choquet et al. 2005).
For live recaptures only, an excess of transients, i.e. individuals never re-encountered after ringing, was detected (TEST3G.SR: p<0.001 for nestlings and p<0.01 for adults). The tests for memory and trap dependence were not significant. For recoveries only, the tests detected an excess of immediate recoveries for birds ringed as nestlings (p=0.001) but not for adults (p=0.09). There was no excess of recoveries concentrated in any other years for either of the groups (p=0.24 for nestlings and p=0.46 for adults). Because the presence of transients leads to the underestimation of survival rates, a model where parameters in the first year after ringing are estimated separately from parameters in the following years has to be considered (Pradel et al. 2005).
References
Choquet R, Reboulet A-M, Lebreton JD, Gimenez O, Pradel R (2005) U-CARE 2·2 User's Manual. In. CEFE, UMR 5175, Montpellier, France
Duriez O, Sæther SA, Ens BJ, Choquet R, Pradel R, Lambeck RHD, Klaassen M (2009) Estimating survival and movements using both live and dead recoveries: a case study of oystercatchers confronted with habitat change. J. Appl. Ecol. 46:144-153
Pradel R, Gimenez O, Lebreton JD (2005) Principles and interest of GOF tests for multistate capture-recapture models. Anim. Biodivers. Conserv. 28:189-204
Table S3: List of models without testing external or periodical covariables.
i: constant, g: group of age at ringing, f: departure state, to: arrival state, t: time, current: current state, juv: first year of birds ringed as nestlings, ad: parameter equal for birds ringed as adults and after first year for birds ringed as nestlings.
Other abbreviations: H: alive in High and sandy area, R: alive in Riverside area, Dr: death recovered by ringer within monitored areas, Dp: death recovered by public within unmonitored areas, best: best model selected from previous step, tlin: testing linear trend in survival variation, trend: three survival rates according to trends period (see Fig. 1).
Model n° / Survival / Dispersal / Probability of reencountered or recoveries / Initial state / Nr par. / Deviance / Δ QAICc1 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current.t.g / t.g / 1001 / 65016 / 384.34
2 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current.t.g / t / 966 / 65714.7 / 581.73
3 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current.t.g / g / 933 / 67778 / 1547.00
4 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current.t.g / i / 932 / 67972.8 / 1649.02
5 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current.t / best / 855 / 66350.8 / 390.98
6 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current.g / best / 604 / 68124.16 / 372.15
7 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current / best / 596 / 68346.53 / 466.90
8 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / i / best / 589 / 71499.41 / 2175.42
9 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current(H:R).t.g+ current(Dr).t.g+ current(Dp).t.g / best / 906 / 66011.67 / 443.75
10 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current(H).t.g+ current(R).t.g+ current(Dr:Dp).t.g / best / 808 / 68461.87 / 1343.03
11 / to.t.juv+to.t.ad / f.to.t.juv+f.to.t.ad / current(H:R).t.g+ current(Dr:Dp).t.g / best / 808 / 68461.87 / 1343.03
12 / to.t.juv+to.t.ad / f.to.juv+f.to.ad / best / best / 795 / 66592.55 / 260.41
13 / to.t.juv+to.t.ad / f.to.t.juv+f.to.ad / best / best / 908 / 65916.2 / 400.97
14 / to.t.juv+to.t.ad / f.to.juv+f.to.t.ad / best / best / 923 / 65911.67 / 471.86
15 / to.t.juv+to.t.ad / f.to.t / best / best / 980 / 65463.7 / 517.40
16 / to.t.juv+to.t.ad / f.to / best / best / 791 / 66770.28 / 341.14
17 / to.t.juv+to.t.ad / f / best / best / 788 / 66824.56 / 358.33
18 / to.t.juv+to.t.ad / to / best / best / 788 / 66720.2 / 300.98
19 / to.t.juv+to.t.ad / t.juv+ t.ad / best / best / 850 / 66331.37 / 357.64
20 / to.t.juv+to.t.ad / t.juv+ ad / best / best / 817 / 66530.27 / 320.64
21 / to.t.juv+to.t.ad / juv+ t.ad / best / best / 819 / 66515.32 / 321.13
22 / to.t.juv+to.t.ad / juv+ ad / best / best / 786 / 66704.27 / 283.83
23 / to.t.juv+to.t.ad / t / best / best / 817 / 66694.26 / 410.74
24 / to.t.juv+to.t.ad / i / best / best / 785 / 66867.68 / 369.43
25 / to.juv+to.t.ad / best / best / best / 708 / 67043.97 / 157.14
26 / to.t.juv+to.ad / best / best / best / 710 / 67004.18 / 142.98
27 / to.juv+to.ad / best / best / best / 618 / 67406.35 / 26.32
28 / to.t / best / best / best / 710 / 67022.76 / 153.19
29 / to / best / best / best / 616 / 67710.33 / 186.36
30 / t.juv+ t.ad / best / best / best / 676 / 66978.1 / 0.0
31 / t.juv+ ad / best / best / best / 645 / 67353.16 / 92.79
32 / juv+ t.ad / best / best / best / 645 / 67389.22 / 112.59
33 / juv+ ad / best / best / best / 614 / 67583.55 / 109.73
34 / t / best / best / best / 644 / 67389.38 / 109.09
35 / i / best / best / best / 613 / 67594.89 / 112.49
36 / juv+ ad+t / best / best / best / 645 / 67290.11 / 58.14
37 / to.juv+to.ad+t / best / best / best / 650 / 67204.43 / 29.09
38 / to.t.juv+t.ad / best / best / best / 737 / 66847.13 / 162.32
39 / t.juv+to.t.ad / best / best / best / 735 / 66892.39 / 179.26
40 / to.t.juv+ ad / best / best / best / 708 / 67048.78 / 159.79
41 / juv+to.t.ad / best / best / best / 706 / 67145.66 / 205.34
42 / juv*tlin + ad*tlin / best / best / best / 616 / 67516.0 / 79.9
43 / juv*tlin + ad.t / best / best / best / 644 / 67194.28 / 1.90
44 / juv.t + ad*tlin / best / best / best / 646 / 67340.5 / 94.9
45 / to.t.juv+to.t.ad / f.to.juv.trend+f.to.ad.trend / best / best / 684 / 67103.56 / 98.78
46 / to.t.juv+to.t.ad / f.to.juv.trend+f.to.ad / best / best / 684 / 67117.19 / 106.27
47 / to.t.juv+to.t.ad / f.to.juv+f.to.ad.trend / best / best / 676 / 67113.85 / 74.58
Table S4: List of models to test external or periodical covariables on survival rates (n=45).
pp: average seasonal precipitation (mlm), tmin: average seasonal minimum temperature (°C), snowcov: number of days with >1cm of snow on the ground, barnowlprod: ratio of number of ringed nestlings on ringed adults, vole: two survival rates according to years with high density and low density of voles, swinter: two survival rates according to years with severe winter or not, trend: three survival rates according to trends period (see Fig. 1), road: intensity of road traffic i.e. the number of car kilometres per year, pp/t: different survival rates according to Dry/Cold –Dry/Warm – Wet/Cold – Wet/Warm years.
Survival / model n° / Nr Par. / Deviance / Δ QAICc / Vjuv / ad
Continous covariates
ppwinter / ppwinter / 48 / 614 / 67867.71 / 265.86 / 0.0%
ppwinter / 49 / 646 / 67350.22 / 94.76 / 56.6%
ppwinter / 50 / 647 / 67327.19 / 85.71 / 62.2%
ppspring / ppspring / 51 / 614 / 88961.94 / 11856.10 / 0.0%
ppspring / 52 / 670 / 67316.59 / 163.75 / 64.7%
ppspring / 53 / 647 / 67334.73 / 89.85 / 60.3%
ppsummer / ppsummer / 54 / 614 / 67901.05 / 284.18 / 0.0%
ppsummer / 55 / 646 / 67346.99 / 92.98 / 57.4%
ppsummer / 56 / 647 / 67332.43 / 88.59 / 60.9%
ppfall / ppfall / 57 / 614 / 73791.17 / 3520.51 / 0.0%
ppfall / 58 / 646 / 67346.22 / 92.56 / 57.6%
ppfall / 59 / 647 / 67334.7 / 89.83 / 60.3%
tminwinter / tminwinter / 60 / 614 / 67954.52 / 313.56 / 0.0%
tminwinter / 61 / 646 / 67349.87 / 94.57 / 56.7%
tminwinter / 62 / 647 / 67332.57 / 88.66 / 60.9%
tminspring / tminspring / 63 / 616 / 67564.66 / 106.32 / 4.6%
tminspring / 64 / 646 / 67350.03 / 94.66 / 56.6%
tminspring / 65 / 647 / 67322.19 / 82.96 / 63.4%
tminsummer / tminsummer / 66 / 616 / 67575.4 / 112.22 / 2.0%
tminsummer / 67 / 646 / 67347.5 / 93.27 / 57.2%
tminsummer / 68 / 647 / 67334.65 / 89.81 / 60.4%
minfall / minfall / 69 / 616 / 67571.55 / 110.10 / 2.9%
minfall / 70 / 646 / 67347.45 / 93.24 / 57.3%
minfall / 71 / 647 / 67331.61 / 88.13 / 61.1%
snowcov / snowcov / 72 / 614 / 82502.51 / 8306.96 / 0.0%
snowcov / 73 / 646 / 67350.2 / 94.75 / 56.6%
snowcov / 74 / 616 / 67544.27 / 95.12 / 9.5%
barnowlprod / barnowlprod / 75 / 616 / 67546.43 / 96.30 / 9.0%
barnowlprod / 76 / 645 / 67742.05 / 306.46 / 0.0%
barnowlprod / 77 / 647 / 67327.86 / 86.08 / 62.0%
road / road / 78 / 614 / 67747.75 / 199.95 / 0.0%
road / 79 / 645 / 67618.61 / 238.63 / 0.0%
road / 80 / 645 / 67293.42 / 59.96 / 70.4%
Periodical variables
vole / vole / 81 / 616 / 67537.65 / 91.48
vole / 82 / 646 / 67338.37 / 88.25
vole / 83 / 647 / 67329.43 / 86.94
swinter / swinter / 84 / 616 / 67544.27 / 95.11
swinter / 85 / 646 / 67313.2 / 74.42
swinter / 86 / 647 / 67334.44 / 89.69
trend / trend / 87 / 618 / 67465.76 / 58.96
trend / 88 / 647 / 67260.89 / 49.28
trend / 89 / 648 / 67320.56 / 85.67
pp/t / pp/t / 90 / 620 / 67531.18 / 101.90
pp/t / 91 / 648 / 67246.68 / 45.08
pp/t / 92 / 649 / 67319.81 / 88.87
Appendix S5: Analyses of small scale data set.
Data and Methods
We used the data from two best monitored populations in order to test if the survival estimates at the local scale were in accordance with the dynamics observed. The two best monitored study areas are located in different geographical regions: “Midden Betuwe” (MB) in the riverside region and “Achterhoek” (At) in the high and sandy plateau region (Fig. S6). The study in Midden Betuwe was conducted in a 100km² study area to study the impact of intensification of land use on little owl populations (Fuchs 1986, Fuchs and van de Laar 2008). We used data from 1977 to 1994 (2,187 ringed individuals; 1,643 nestlings and 544 adults). During this period, the number of breeding territories declined, the mean growth rate of the population was 0.97 (SD = 0.21) and the mean number of juvenile produced per brood was estimated to 2.08 (SE = 0.05, Piet Fuchs pers. comment). The study in Achterhoek covered 120 km² (reference from RH/PS). We used data from 1988 to 2007 (1,775 ringed individuals; 1,519 nestlings and 254 adults). The number of successful breeding pairs was used as an index of the population size. For this population, the mean growth rate was 1.03 (SD=0.18) and the mean number of juvenile produced per brood was estimated to 1.98 (SE = 0.07, Ronald van Harxen pers. comment). Using the information of the number of offspring per brood recorded at ringing, we estimated the number of successful broods per year divided by the area of occupied territories in order to create a density index for both areas. Average pair density per km² was higher in MB (1.41 [1.27-1.56]95%CI) than in At (1.11 [1.08-1.13]95%CI).
We estimated the survival rate of juveniles and adults in MB and At using the same model structure as for the large scale analysis. We assumed that probabilities of reencounter and recovery varied among time and that dispersal rates differed between juveniles and adults and between regions. However, since the data were scarcer, we fixed the dispersal rates to those estimated for each region with the large scale model.
We correlated the survival rates estimates with growth rate of the local population.
We used a single sex projection matrix with two age classes (Caswell 2001) to estimate the proportion of breeders knowing the other demographic parameters and the growth rate within each local populations (MB and At):
Where ε was the sex ratio at birth (we assumed it was balanced i.e. equal to ½),
r was the estimate of the number of young produced per breeding within area (Willems et al. 2004),
δ was the unknown proportion of breeders,
Sjuv and Sad were the estimated mean survival rate for juvenile and adult respectively
Assuming that the observed growth rate equalled the asymptotic λ for each population, we could calculate the proportion of breeders according to the equation:
Results
For the MB study site, the best model (QAICc=4 703.2, np=92) included a constant survival rate for juveniles (0.285 [0.234 -0.342] 95%CI) and a time dependent survival rate for adults (mean over all years: 0.727 [0.681 -0.773] 95%CI). As at the large scale, low survival rates were found in 1977, 1981 and 1985 for adults, but not in 1989 and 1990. In the second best model (QAICc=4 706.6, np=76) survival rate was constant for both juveniles (0.287 [0.238 -0.341] 95%CI) and adults (0.697 [0.668 -0.724] 95%CI).