Data collection
Complete data and references list are presented in the Electronic Supplementary Material (ESM) 1.
We first identified published articles in peer reviewed journals that reported telomere length or rate of telomere erosion in birds. We (i) screened our own reference lists, (ii) backward searched references sections of selected articles from the list, and (iii) collected publications using Google Scholar with the following terms: “telomere” OR “telomere shortening” OR “telomere erosion” AND “birds” OR “avian.” For each retrieved paper, methods were individually appraised by one author (FC) to restrict the selection to studies using the telomere restriction fragment (TRF) method to assess telomere length or erosion. Our meta-analysis therefore only relies on published telomere length data expressed in absolute length (kilobases, kb). Using absolute values of telomere length (TL) allowed us to do interspecific comparisons of telomere length relationships with other life history traits, based on the calculation for each species of a mean adult telomere length (hereafter Adult TL ), mean chick telomere length (hereafter Chick TL ) and mean telomere length rate of change over the lifecycle (hereafter TROC ), allowing inter-specific comparisons. We primarily used studies that sampled erythrocytes, with one exception of a study that used tissue attached to feathers. Data extraction is detailed in ESM 2.
Data on life history traits were compiled for species for which TRF measurements of Adult TL were available. ESM 1 references the principal sources used to collect TRF values and 12 life-history variables for each species. We included publications in peer reviewed journals, on-line peer-reviewed data bases, and books. The protocol for collecting life history data is described in ESM 2. In particular, we examined (i) species body mass and size, (ii) longevity (age at maturity and lifespan), (iii) reproductive patterns (e.g. using egg or clutch sizes), (iv) growth patterns (pre- and post-hatching growth rates), and (v) post-fledging parental care.
We adjusted statistical analyses for sample sizes using the standard procedure for meta-analyses (Liberati et al., 2009)). We could not control for the mean age at which telomeres were measured in adult individuals of each species. However, age is likely to be related to life history variables included in our analysis (e.g., body size, longevity) and likely to scale with body mass. As such, risks of multicollinearity of variables were high. Thus, we used principle components analyses as a variable reduction method (see details below). We conducted a separate analysis of Chick TL to check for specific relationships between life history traits and telomere length early in the species lifecycles. In addition, we analysed associations of TROC and life-history traits. As in previous studies, TROC was estimated as the slope of the linear regression between TL and age over life (i.e. including chick and adults life stages, (Haussmann et al., 2003; Tricola et al., 2018).