Telomere dynamics could underlie life-history trade-offs among growth, size, and longevity, but our ability to quantify such processes in natural, unmanipulated populations is limited. We investigated how 4 years of artificial selection for either larger or smaller tarsus length, a proxy for body size, affected early-life telomere length (TL) and several components of fitness in two insular populations of wild house sparrows over a study period of 11 years. The artificial selection was expected to shift the populations away from their optimal body size and increase the phenotypic variance in body size. Artificial selection for larger individuals caused TL to decrease, but there was little evidence that TL increased when selecting for smaller individuals. There was a negative correlation between nestling TL and tarsus length under both selection regimes. Males had longer telomeres than females and there was a negative effect of harsh weather on TL. We then investigated whether changes in TL might underpin fitness effects due to the deviation from the optimal body size. Mortality analyses indicated disruptive selection on TL because both short and long early-life telomeres tended to be associated with the lowest mortality rates. In addition, there was a tendency for a negative association between TL and annual reproductive success, but only in the population where body size was increased experimentally. Our results suggest that natural selection for optimal body size in the wild may be associated with changes in TL during growth, which is known to be linked to longevity in some bird species.

Methods are detailed in the main text and in Appendix S1.

TL: Mean relative erythrocyte telomere length (TL) was measured using the qPCR method (Cawthon, 2002).

tarsus: Tarsometatarsus ("tarsus") length was measured using calipers to the nearest 0.01 mm. Because tarsus length increases with nestling age, we estimated standardized tarsus length ("tarsus (age-corrected)") as the residuals of a multiple linear regression of tarsus length on age (in days) and age squared (i.e. a quadratic model) separately for each sex. The standardizations were made for each population separately such that we did not have to assume equal growth trajectories between treatments. "Adult tarsus" is the uncorrected tarsus length of the recaptured full-grown juvenile or adult individual.

age: Number of days since hatching.

sex: 1=male, 2=female

brood size: Number of chicks in the nest at the time of sampling.

island_ID: 60=high (Leka) or 61=low (Vega) selection regime.

selection_category: Refer to: Selected (1): Both parents had been subject to the selection regime, i.e. captured during a winter selection event and allowed to stay in the population. Intermediate (0.5): one parent had been subject to artificial selection. Unselected (0): No parents had been artificially selected (i.e. accidentally not captured during any winter selection event) and these individuals are therefore the unselected controls.

NAO_30: The daily North Atlantic Oscillation (NAO) index (retrieved from the National Oceanic and Atmospheric Administration (NOAA), 2018) was averaged over a 30-day period prior to blood sampling ("NAO_30").

LRS: Total number of recruits produced (lifetime reproductive success).

lifespan: Years between hatching and last observation.

NA: Missing values.