RESULTS
Temporal changes in tarsus
and telomere lengths during artificial selection
In the high population, the tarsus of fledglings with both
parents artificially selected was on average longer than the tarsus of
fledglings produced by unselected individuals (i.e. with parents not
subjected to artificial selection; model ranked 1, selected vs.
unselected: βselected =0.52,
CI=[0.04, 1.02], Fig. 3b,
Tables S2.1 and S2.2), and tended to be shorter than tarsus of
fledglings produced by unselected individuals in the lowpopulation (model ranked 2, ∆AICc=0.6, selected vs. unselected:βselected =-0.38, CI=[-0.80, 0.03], Fig. 4b,
Table S2.1). Intermediate individuals with one artificially selected
parent showed a similar weak tendency when compared to the unselected
(high : βintermediate =0.25, CI=[-0.21,
0.72]; low : βintermediate =-0.31,
CI=[-0.69, 0.08], Figs. 3b and 4b, Tables S2.1 and S2.2). Across the
study period (2002-2006), fledgling tarsus length increased linearly in
the high population (n =158, model ranked 1:βyear =0.70, CI=[0.01, 1.38], Figs. 3a, S2.1
and S2.2, Tables S2.1 and S2.2) and decreased linearly in the lowpopulation (n =408, model ranked 1:βyear =-0.63, CI=[-1.23, -0.04], Figs. 4a,
S2.1 and S2.2, Tables S2.1 and S2.2). In both populations, there was
weak evidence (i.e. marginally significant) for a curvilinear change
over the years, indicating that after the initial divergence the change
in tarsus length ceased (high , model ranked 1:βyear^2 =-0.09, CI=[-0.20, 0.03];low , model ranked 1: βyear^2 =0.10,
CI=[-0.00, 0.20], Figs. 3a and 4a, Table S2.2).
TL of fledglings with both parents artificially selected was not
different from those of unselected individuals in the highpopulation (model ranked 3, ∆AICc=0.7, selected vs. unselected:βselected =0.02, CI=[-0.18, 0.22], Fig. 3d,
Tables S2.1) nor in the low population (model ranked 3,
∆AICc=3.0, selected vs. unselected:βselected =0.03, CI=[-0.07, 0.13], Fig. 4d,
Table S2.1). However, intermediate individuals with one artificially
selected parent showed weak evidence for a tendency towards shorter
telomeres when compared to the unselected in the high population
(βintermediate =-0.13, CI=[-0.32, 0.07], Fig.
3d), and towards longer telomeres compared to unselected individuals in
the low population (βintermediate =0.05,
CI=[-0.05, 0.15], Fig. 4d). Across the study period, fledgling TL
decreased linearly in the high population (model ranked 2,
∆AICc=0.0: βyear =-0.26, CI=[-0.50, -0.01],
Figs. 3c and S2.1, Tables S2.1 and S2.2), but there was no change in thelow population (model ranked 2, ∆AICc=1.9:βyear =0.01, CI=[-0.02, 0.03], Figs. 4c and
S2.1, Table S2.1). In the high population there was weak evidence
for a curvilinear change over the years, indicating that after the
initial increase the change in TL ceased (model ranked 2:βyear^2 =0.04, CI=[-0.00, 0.08], Fig. 3c,
Table S2.2), but there was no evidence for any curvilinear change in thelow population (model ranked 4,
∆AICc=0.7:βyear^2 =0.01, CI=[-0.01, 0.03], Fig. 4c).
The model ranked 1 in the low population included only sex and
age (Table S2.1 and S2.2). Overall, our results show some evidence for
an inverse association between changes in tarsus length and TL across
cohorts in populations where body size is shifted away from its optimum.