Supporting online information for the paper:
Malaria infections reinforce competitive asymmetry between two Ficedula flycatchers in a recent
contact zone
Katarzyna Kulma, Matthew Low, Staffan Bensch & Anna Qvarnström
Supporting online information (Appendix S1, Appendix S2)
Appendix S1 – Multistate mark-recapture models for modelling the transition and resighting
parameters for collard and pied flycatchers
Table S1a
Candidate set of multistate mark-recapture models to determine the best covariate structure for the
resighting parameter (p) for collard flycatchers; complex structures of the survival (Ф) and transition
(ψ) parameters were held constant while modelling various effects on p. These factors were sex,
infection status, year effects (time) or a constant. Year effects were not included in survival and
transition models because of convergence problems in the full model. QAICc = AIC adjusted for
overdispersion (ĉ = 1.5) and sample size; ∆QAICc = difference in AICc relative to the best model; wi =
AICc weight of the model; K = number of parameters in the model.
Model
QAICc
∆QAICc
wi
K
Фsex, infection
psex, time
ψsex, infection
6182.2
0.0
0.67
38
Фsex, infection
ptime
ψsex, infection
6183.7
1.45
0.33
29
Фsex, infection
ptime, infection
ψsex, infection
6195.8
13.6
0
47
Фsex, infection
psex
ψsex, infection
6202.3
20.1
0
22
Фsex, infection
pconstant
ψsex, infection
6203.5
21.3
0
21
Фsex, infection
psex, infection
ψsex, infection
6204.3
22.1
0
26
Фsex, infection
psex, time, infection ψsex, infection
6219.1
36.9
0
65
Table S1b
Candidate set of multistate mark-recapture models to determine the best covariate structure for the
transition parameter (ψ) for collard flycatchers; effects were modelled for ψ while a complex structure
of the survival (Ф) parameter was held constant and the resighting parameter was fixed at psex, time (see
Table S1a and Methods for more details). These factors were sex, infection status, year effects (time)
or a constant. Year effects were not included in survival and transition models because of
convergence problems in the full model. QAICc = AIC adjusted for overdispersion (ĉ = 1.5) and
sample size; ∆QAICc = difference in AICc relative to the best model; wi = AICc weight of the model;
K = number of parameters in the model.
Model
QAICc
∆QAICc
wi
K
Фsex, infection
psex, time
ψinfection
6180.2
0.0
0.75
32
Фsex, infection
psex, time
ψinfection, sex
6182.4
2.2
0.25
38
Фsex, infection
psex, time
ψconstant
6741.3
561
0
27
Фsex, infection
psex, time
ψ sex
6743.1
563
0
28
Table S1c
Candidate set of multistate mark-recapture models to determine the best covariate structure for the
resighting parameter (p) for pied flycatchers; complex structures of the survival (Ф) and transition (ψ)
parameters were held constant while modelling various effects on p. These factors were sex, infection
status, year effects (time) or a constant. Year & sex effects were not included in the transition models
because of convergence problems in the full model (see Table S1d). QAICc = AIC adjusted for
overdispersion (ĉ = 1.3) and sample size; ∆QAICc = difference in AICc relative to the best model; wi =
AICc weight of the model; K = number of parameters in the model.
Model
QAICc
∆QAICc
wi
K
Фsex, infection
ptime
ψinfection
1679.7
0.0
0.39
23
Фsex, infection
psex,time
ψinfection
1680.7
1.01
0.24
32
Фsex, infection
pconstant
ψinfection
1682.1
2.44
0.11
15
Фsex, infection
psex
ψinfection
1682.1
2.45
0.11
16
Фsex, infection
pinfection
ψinfection
1682.2
2.54
0.11
17
Фsex, infection
psex,infection
ψinfection
1685.4
5.7
0.04
20
Фsex, infection
ptime, infection
ψinfection
1693.7
14.0
0
41
Фsex, infection
psex, time, infection ψinfection
1705.6
25.9
0
59
Table S1d
Candidate set of multistate mark-recapture models to determine the best covariate structure for the
transition parameter (ψ) for pied flycatchers; effects were modelled for ψ while a complex structure of
the survival (Ф) parameter was held constant and the resighting parameter was fixed at psex, time (see
Table S1c); this model structure for p was used despite it not being the highest-ranked model because
of equivalent model support to the highest-ranked model (see S1c) and a priori knowledge that
recapture effort was likely to vary between years and sexes; thus we wanted to ensure we generated
the most accurate survival estimates from a robust model structure. Factors included were sex,
infection status, year effects (time) or a constant. Year effects were not included in survival and
transition models because of convergence problems in the full model. QAICc = AIC adjusted for
overdispersion (ĉ = 1.3) and sample size; ∆QAICc = difference in AICc relative to the best model; wi =
AICc weight of the model; K = number of parameters in the model.
Model
QAICc
∆QAICc
wi
K
Фsex, infection
psex, time
ψinfection
1680.7
0.0
1
32
Фsex, infection
psex, time
ψconstant
1743.3
62.6
0
27
Фsex, infection
psex, time
ψ sex
1744.8
64.1
0
28
Фsex, infection
psex, time
ψinfection, sex
NA*
-
-
-
*Not available because of a failure of model convergence
Appendix S2 – Multistate mark-recapture models for modelling the survival parameters for collard
and pied flycatchers & model-averaged estimates
All survival models shown here contain a fixed resighting and transition parameter (see Appendix S1
and methods for details). For collard flycatchers the resighting parameter is fixed with a sex and year
effect (i.e. separate estimates for each sex for each year); for the transition parameter it is fixed with
separate estimates for transitions between the different infection states (unknown, infected, noninfected). Bold font indicates models within Δ QAICc < 2.
a) Collared flycatchers – general (pooled) infection
i) Set of candidate models
Model
Infection
Infection + Sex
Intercept
Infection * Sex
Sex
k
30
31
29
32
30
QAICc
6177.02
6178.96
6179.45
6180.17
6181.42
Δ QAICc
0.00
1.93
2.42
3.14
4.43
QAICc Weight
0.5018
0.1902
0.1492
0.1042
0.0546
ii) Model-averaged survival probability estimates for CF pooled estimates
Parameter
Estimate
Uninfected Female
Uninfected Male
Infected Female
Infected Male
0.5219
0.5233
0.6078
0.6192
Standard
Error
0.0419
0.0419
0.0499
0.0555
Lower CI
95%
0.4399
0.4413
0.5071
0.5062
Upper CI
95%
0.6027
0.6042
0.7002
0.7206
b) Collared flycatchers – shared infections
i) Set of candidate models
Model
infection
intercept
infection + sex
infection*sex
sex
k
34
33
35
36
34
QAICc
6176.77
6177.16
6177.95
6178.19
6178.25
Δ QAICc
0.0000
0.3942
1.1831
1.4236
1.4793
QAICc Weight
0.2992
0.2457
0.1655
0.1468
0.1428
ii) Model-averaged survival probability estimates (CF shared infections)
Parameter
Estimate
Uninfected Female
Uninfected Male
Infected Female
Infected Male
0.5219
0.5233
0.5963
0.6325
Standard
Error
0.0419
0.0419
0.0537
0.0619
Lower CI
95%
0.4399
0.4413
0.4883
0.5052
Upper CI
95%
0.6027
0.6042
0.6957
0.7436
c) Collared flycatchers – hPHSIB1 infections
i) Set of candidate models
Model
infection
intercept
infection*sex
infection + sex
sex
k
34
33
36
35
34
QAICc
6179.855
6179.872
6180.024
6180.345
6181.791
Δ QAICc
0.0000
0.0162
0.1681
0.4895
1.9354
QAICc Weight
0.2455
0.2435
0.2257
0.1921
0.0933
ii) Model-averaged survival probability estimates (CF hPHSIB1 infections)
Parameter
Estimate
Uninfected Female
Uninfected Male
Infected Female
Infected Male
0.5219
0.5233
0.6021
0.6351
Standard
Error
0.0419
0.0419
0.0617
0.0683
Lower CI
95%
0.4399
0.4413
0.4774
0.4941
Upper CI
95%
0.6027
0.6042
0.7148
0.7562
d) Pied flycatchers – general infection
i) Set of candidate models
Model
infection + sex
infection * sex
sex
infection
intercept
k
31
32
30
30
29
QAICc
1680.597
1680.759
1683.817
1690.350
1692.820
Δ QAICc
0.0000
0.1619
3.2198
9.7531
12.2236
QAICc Weight
0.4691
0.4326
0.0938
0.0036
0.0010
ii) Model-averaged survival probability estimates (PF pooled)
Parameter
Estimate
Uninfected Female
Uninfected Male
Infected Female
Infected Male
0.4413
0.6651
0.2486
0.5516
Standard
Error
0.0989
0.0769
0.0662
0.0674
Lower CI
95%
0.2646
0.5024
0.1418
0.4190
Upper CI
95%
0.6342
0.7963
0.3985
0.6773
e) Pied flycatchers – shared infections
i) Set of candidate models
Model
infection + sex
infection * sex
sex
infection
intercept
k
31
32
30
30
29
QAICc
1680.983
1681.177
1684.006
1685.794
1688.156
Δ QAICc
0.0000
0.1937
3.0227
4.8114
7.1728
QAICc Weight
0.4451
0.4041
0.0982
0.0401
0.0123
ii) Model-averaged survival probability estimates (PF shared)
Parameter
Estimate
Uninfected Female
Uninfected Male
Infected Female
Infected Male
0.4413
0.6651
0.2996
0.5740
Standard
Error
0.0989
0.0769
0.1001
0.0753
Lower CI
95%
0.2646
0.5024
0.1438
0.4242
Upper CI
95%
0.6342
0.7963
0.5215
0.7114
f) Pied flycatchers – hPHSIB1 infection
i) Set of candidate models
Model
infection * sex
infection + sex
sex
infection
intercept
k
40
39
38
38
37
QAICc
1681.253
1682.111
1684.417
1688.156
1693.010
Δ QAICc
0.0000
0.8579
3.1640
6.9027
11.7569
QAICc Weight
0.5287
0.3443
0.1087
0.0168
0.0015
ii) Model-averaged survival probability estimates (PF hPHSIB1 infection)
Parameter
Estimate
Uninfected Female
Uninfected Male
Infected Female
Infected Male
0.4413
0.6651
0.2426
0.7381
Standard
Error
0.0989
0.0769
0.0991
0.1026
Lower CI
95%
0.2646
0.5024
0.1001
0.4991
Upper CI
95%
0.6342
0.7963
0.4797
0.8885
g) Pied flycatchers – hPFC1 infections
i) Set of candidate models
Model
infection + sex
infection * sex
sex
infection
intercept
k
31
32
30
30
29
QAICc
1762.852
1762.944
1764.521
1770.104
1771.023
Δ QAICc
0.0000
0.0921
1.6691
7.2514
8.1709
QAICc Weight
0.4111
0.3926
0.1785
0.0110
0.0069
ii) Model-averaged survival probability estimates (PF hPFC1 infections)
Parameter
Estimate
Uninfected Female
Uninfected Male
Infected Female
Infected Male
0.4413
0.6651
0.3099
0.5852
Standard
Error
0.0989
0.0769
0.0786
0.0632
Lower CI
95%
0.2646
0.5024
0.1794
0.4585
Upper CI
95%
0.6342
0.7963
0.4799
0.7015