SUPPLEMENTARY MATERIAL
Who escapes detection? Quantifying the causes and consequences of sampling biases in a
long-term field study
Lindall R. Kidd, Ben C. Sheldon, Emily G. Simmonds & Ella F. Cole
Appendix S1. Further method details.
Mist netting for immigrants to the study site
In addition to monitoring during the breeding season, throughout the winter, great tits are
trapped using mist nests at a seed feeder, and immigrants to the woods are ringed and fitted
with a PIT tag. Across all winters, from November to February, one 18m net was set at a
fixed centrally located site within the woods on a weekly basis. In addition to this, for the
duration of the study period, teams of people operated two-three nets simultaneously in close
proximity to one another and moved every week to cover the entire study area. In the winter
leading to the breeding season of 2009, a total of 56 days were spent catching at multiple sites
per day from late August to late February. In the winter before the 2010 breeding season, mist
netting procedures were similar to those used in the previous winter, a total of 56 days were
spent catching at multiple sites between August and early March. In the winter leading up to
the 2012 breeding season, mist netting efforts increased and were more standardised. In
addition to the centrally located feeder, teams of ringers would consecutively operate threefour nets at regularly spaced intervals throughout the study site. A total of 68 days were spent
ringing between multiple sites from mid-August to early March 2012.
Night time roosting data
Night time roost temperatures were recorded using iButton thermometers (DS1921G-F5,
accurate to +/- 1ºC, HomeChip Ltd) secured within the nest cup with blunted garden wire
twisted around the iButton and sealed within a small cotton pouch. These iButtons were
placed in half of all great tit nests during the 2014 breeding season in (n=170) throughout the
study site. Only nests in which the iButton was placed prior to clutch completion (i.e. during
the egg laying stage) were used in this study (n=63).
In order to determine if the female was present at night time, in-nest temperatures were
compared to local ambient temperature recorded using iButtons (DS1923-F5, accurate to +/0.5ºC, HomeChip Ltd) which were distributed in evenly spaced 100m grids throughout the
woodland. Data were analysed in R to determine how many of the recordings taken at night
(after 8pm but before 7am the next day) were more than 5ºC above ambient temperature. If
more than 50% of the recordings exceeded this threshold throughout the course of the night,
we considered this evidence that the female was in the nestbox. This accounts for the fact that
the female will not be actively incubating all night, whilst excluding the potential of a single
anomalously high recording for reasons other than female presence. At each nest, we showed
that the female was present in the box during each night of our recordings (133 nights).
Table S1. Roost data for 63 great tit nests installed with ibutton temperature data loggers
during the laying period in 2014. Night time presence was defined when a nest was 5ºC
above ambient temperature for more than 50% of temperature recodings taken throughtout
the evening (after 8pm but before 7am the next day).
Nestbox
Laydate
Female ID
Failed
Before
Ringed or
Night time
Catching
Unringed
Presence
B122
TP42743
20
N
R
Y
B142
TS45004
16
N
R
Y
B190
TS43230
11
N
R
Y
B201
D472069
23
N
U
Y
B212
Y163003
21
N
R
Y
B46
D472031
19
N
R
Y
B53
TS44943
6
N
R
Y
B79
Y836237
17
N
U
Y
C10
NA
3
Y
NA
Y
C108
L809323
3
N
R
Y
C113
Y035838
5
N
R
Y
C132
NA
8
Y
NA
Y
C17
Y838102
6
N
R
Y
C143
TS45597
8
N
R
N
C54
Y164653
8
N
R
Y
CP10
Y163161
16
N
R
Y
CP2
TS47111
13
N
R
Y
CP26
TS45695
25
N
R
Y
CP28
Y163166
13
N
R
Y
CP7
TS44206
14
N
R
Y
EX1
TJ24698
33
N
U
Y
EX12
Y838931
13
N
R
Y
EX29
Y839893
14
N
R
Y
EX31
TS47241
14
N
R
Y
EX40
Z031001
14
N
U
Y
EX46
Y839895
24
N
R
Y
EX51B
TS46402
15
N
R
Y
EX61
TS47469
13
N
R
Y
EX62G
D471762
19
N
R
Y
EX67
Y161935
13
N
R
Y
EX74
Z032571
15
N
U
Y
MP12
TS45689
15
N
R
Y
MP47
D472723
8
N
R
Y
MP65
L809606
5
N
R
Y
MP66
TL36088
12
N
R
Y
MP84
D472767
19
N
R
Y
O23
D471222
15
N
R
Y
O64
D472969
10
N
R
Y
O70
TP27415
19
N
R
Y
O86
TR44250
18
N
R
Y
P11
TX42522
20
N
R
Y
P14
D472011
14
N
U
Y
P5
Y160609
9
N
R
Y
SW1
TR44247
14
N
R
Y
SW107
Y160493
7
N
R
Y
SW134
D471706
19
N
R
Y
SW28
Y163532
8
N
R
Y
SW34
NA
8
Y
NA
Y
D471563
16
N
U
Y
SW73
NA
21
Y
NA
Y
SW76
D472415
16
N
U
Y
SW81
Y162433
15
N
R
Y
SW85
TR41967
10
N
R
Y
W10
TR44083
14
N
R
Y
W14
TS43349
13
N
R
Y
W19
Y160435
9
N
R
Y
W26
Y839274
27
N
R
Y
W31
TS43593
14
N
R
Y
W33
Y163075
19
N
R
Y
W38
Y838658
4
N
R
Y
W51A
X235006
7
N
R
Y
W62
NA
11
Y
NA
Y
W9
TS47253
10
N
R
Y
SW61A
Assessing PIT tag scanner accuracy
The accuracy of the scanner was assessed by comparing the identities returned by the scanner
with the identities established through trapping at the nestbox; scanner failure was inferred if
no PIT tag was read, but the female was found, on subsequent trapping, to be carrying a
functional PIT tag. During the 2009 and 2010 breeding seasons an older model of scanner
was used; this was found to have a tag-detection success rate of 93% and 91% respectively.
In 2012 we used an improved scanner, which had a success rate of 100%.
Table S2. Results of ordinary bootstrap analyses assessing whether fecundity selection
coefficients for laying date for female great tits 2009, 2010 and 2012 across three datasets
were influenced by differing sample sizes. Datasets consisted of (a) all breeding females,
including those whose breeding attempts failed early (unbiased dataset), (b) only breeding
female whose ring number was known and (c) only females that were identified using the
standard protocol of trapping when nestling are 8-10 days old. The errors are 97.5th
percentile ranges of resampled parameter estimates. Selection coefficients in bold are
significantly different from zero (p<0.05).
Laying date
Random 1 (n dataset b)
β
97.5% CI
Year
N
2009
169
-0.071
-0.175-0.031
Random 2 (n dataset c)
β
N
97.5% CI
157 -0.058
-0.160-0.043
2010
296
0.051
-0.025-0.131
271
0.044
-0.036-0.136
2012
300
0.025
-0.078-0.146
246
0.092
-0.033-0.228
Table S3. Annual non-linear fecundity selection coefficients for laying date for female great
tits 2009, 2010 and 2012. Analyses were carried out on three datasets: (a) all breeding
females, including those whose breeding attempts failed early (unbiased dataset), (b) only
breeding female whose ring number was known and (c) only females that were identified
using the standard protocol of trapping when nestling are 8-10 days old. Selection
coefficients in bold are significantly different from zero (p<0.05).
(a)
Laying date
Year
N
Fecundity
( β ± SE)
(b)
(c)
Fecundity
N
( β ± SE)
Fecundity
N
( β ± SE)
2009
179 -0.555 ± 0.470
169
-0.586 ± 0.421
157
-0.2606 ± 0.386
2010
313 -0.131 ± 0.288
296
-0.252 ± 0.266
271
-0.213 ± 0.237
2012
318 -1.374 ± 0.321
300
-1.307 ± 0.307
246
-1.364 ± 0.257
Table S4. Annual directional recruitment selection coefficients for laying date for female
great tits 2009, 2010 and 2012. Analyses were carried out on three datasets: (a) all breeding
females, including those whose breeding attempts failed early, (b) only breeding female
whose ring number was known and (c) only females that were identified using the standard
protocol of trapping when nestling are 8-10 days old. Selection coefficients in bold are
significantly different from zero (p<0.05).
(a)
Laying date
Year
N
Fecundity (β± SE)
N
2009
2010
2012
179
313
318
-0.294 ± 0.118
-0.005 ± 0.103
-0.037 ± 0.196
169
296
300
(b)
Fecundity
(β± SE)
-0.300 ± 0.116
0.024 ± 0.103
-0.058 ± 0.194
(c)
Fecundity
N
(β± SE)
157 -0.336 ± 0.113
271 -0.016 ± 0.099
246 -0.077 ± 0.192
Table S5. Annual non-linear recruitment selection coefficients for laying date for female
great tits 2009, 2010 and 2012. Analyses were carried out on three datasets: (a) all breeding
females, including those whose breeding attempts failed early, (b) only breeding female
whose ring number was known and (c) only females that were identified using the standard
protocol of trapping when nestling are 8-10 days old. Selection coefficients in bold are
significantly different from zero (p<0.05).
(a)
Laying date
Year
N
Fecundity
( β ± SE)
(b)
N
Fecundity
( β ± SE)
(c)
N
Fecundity
( β ± SE)
2009
179 0.836 ± 1.155
169
0.798 ± 1.126
157
1.529 ± 1.182
2010
313 -0.912 ± 0.739
296
-1.001 ± 0.728
271
-0.884 ± 0.698
2012
318 -1.394 ± 1.168
300
-1.326 ± 1.161
246
-1.371 ± 1.146