INTRODUCTION
Telomeres are nucleoprotein structures that cap the ends of linear chromosomes in most eukaryotes (Blackburn, 1991). Understanding the causes of individual variation in telomere length (TL) is important because this trait has been shown to predict variation in survival or lifespan within and among species (Joeng, Song, Lee, & Lee, 2004; Bize, Criscuolo, Metcalfe, Nasir, & Monaghan, 2009; Monaghan, 2010; Heidinger et al., 2012; Tricola et al., 2018; Wilbourn et al., 2018; Pepke & Eisenberg, 2021) and individual fitness in wild animals (Eastwood et al., 2019). Telomeres shorten through life in many organisms (Dantzer & Fletcher, 2015) due to cell division, oxidative stress, and other factors (Jennings, Ozanne, & Hales, 2000; Reichert & Stier, 2017), which can ultimately result in telomere dysfunction, genome instability, and cell death (Nassour et al., 2019) and organismal senescence (Herbig, Ferreira, Condel, Carey, & Sedivy, 2006). Individual TLs may act as biomarkers or sensors of exposure to intrinsic and extrinsic stressors (Houben, Moonen, van Schooten, & Hageman, 2008), and hence reflect individual condition (Rollings et al., 2017), but the physiological mechanisms underlying the ontogenetic variation in TL is not well known (Monaghan, 2014; Erten & Kokko, 2020). Several studies have investigated the potential of telomere dynamics (i.e. individual differences in TL and telomere loss rate) in mediating life-history trade-offs both across (Dantzer & Fletcher, 2015; Pepke & Eisenberg, 2020) and within relatively long-lived species (Monaghan, 2010; Spurgin et al., 2018). However, despite being an ecologically important trait in many species (Wilbourn et al., 2018), knowledge about the genetic architecture of TL or its adaptive potential in wild populations remains scarce (Dugdale & Richardson, 2018).
Quantifying the additive genetic variance of a trait is required to understand mechanisms driving adaptive evolution, i.e. the response to selection on a trait (Lande, 1979; Ellegren & Sheldon, 2008; Kruuk, Slate, & Wilson, 2008). However, the magnitude of the heritability and mode of inheritance of TL is not well-known in populations of wild animals, and few general patterns have been described (Horn et al., 2011; Dugdale & Richardson, 2018; Bauch, Boonekamp, Korsten, Mulder, & Verhulst, 2019). Utilizing long-term pedigree data, individual variation in early-life TL can be decomposed into various genetic and environmental sources of variation through a type of mixed-effect model (‘animal model’), which takes all relationships from the pedigree into account (Kruuk, 2004; Wilson et al., 2010). Estimates of TL heritabilities from studies using animal models (reviewed in Dugdale & Richardson, 2018) have varied considerably across bird species from h2 =0 (n =177, in wild white-throated dippers, Cinclus cinclus , Becker et al., 2015) toh2 =0.99 (n =125, in captive zebra finches, Taeniopygia guttata , Atema et al., 2015). While most studies are characterized by relatively small sample sizes, recent long-term studies on Seychelles warblers (Acrocephalus sechellensis , n =1317, h2 =0.03-0.08, Sparks et al., 2021) and common terns (Sterna hirundo ,n =387, h2 =0.46-0.63, Vedder et al., 2021) also revealed contrasting estimates of TL heritabilities. Epidemiological studies of humans have documented consistently high TL heritabilities, ranging from h2 =0.34-0.82 (Broer et al., 2013). In humans, some studies reported strong paternal inheritance (e.g. Njajou et al., 2007) or maternal inheritance (e.g. Broer et al., 2013) or that there were no differences in parental mode of inheritance (e.g. Eisenberg, 2014). In birds, several studies have documented maternal effects on offspring telomere dynamics (Horn et al., 2011; Asghar, Bensch, Tarka, Hansson, & Hasselquist, 2015; Reichert et al., 2015; Heidinger et al., 2016), or effects of parental age at conception on offspring TL (Eisenberg & Kuzawa, 2018). Reichert et al. (2015) found a significant correlation between mother-offspring TL measured at 10 days of age in king penguins (Aptenodytes patagonicus ), but not when TL was measured at later ages (>70 days). This may be because post-natal telomere loss rate is strongly influenced by individual environmental circumstances (Wilbourn et al., 2018; Chatelain, Drobniak, & Szulkin, 2020) and does not always correlate strongly with chronological age (Boonekamp, Simons, Hemerik, & Verhulst, 2013; Boonekamp, Mulder, Salomons, Dijkstra, & Verhulst, 2014).
Telomeres shorten during growth and a negative phenotypic correlation between TL and body size has been documented within several species (Monaghan & Ozanne, 2018). This may indicate that there is a negative genetic correlation between TL and size, which could act as an evolutionary constraint on the response of TL to selection on body size and contribute to the trade-off between growth and lifespan (Metcalfe & Monaghan, 2003; Roff & Fairbairn, 2012). Thus, quantifying the genetic correlation between TL and size enables us to determine whether TL can evolve independently of body size. Pepke et al. (2021, submitted ) showed that artificial directional selection on body size affected TL in the opposite direction. However, it is not known if there is a genetic correlation between the two traits, in which case selection acting on TL will affect body size. It is also possible that the negative phenotypic correlation between TL and size has no genetic basis but is shaped by environmental (co)variances (Hadfield, 2008; Kruuk et al., 2008).
TL is a complex phenotypic trait (Aviv, 2012; Hansen et al., 2016) expect to be polygenic, i.e. affected by small effects of many genes (Hill, 2010; Dugdale & Richardson, 2018). Accordingly, numerous genome-wide association studies (GWAS), which tests associations of single-nucleotide polymorphisms (SNPs) with specific traits, have identified several loci correlated with TL in humans that map to genes involved in telomere and telomerase maintenance, DNA damage repair, cancer biology, and several nucleotide metabolism pathways (e.g. Vasa-Nicotera et al., 2005; Andrew et al., 2006; Codd et al., 2010; Levy et al., 2010; Mirabello et al., 2010; Jones et al., 2012; Mangino et al., 2012; Soerensen et al., 2012; Codd et al., 2013; Deelen et al., 2013; Liu et al., 2014; Mangino et al., 2015; Ojha et al., 2016; Delgado et al., 2018; Zeiger et al., 2018; Coutts et al., 2019; Nersisyan et al., 2019; Li et al., 2020). None of the GWA studies in humans specifically tested the marker associations of early-life TL, which pose a challenge to the interpretation of the results, as TL shortens through life in humans (Blackburn, Epel, & Lin, 2015) and genes may have different impacts at various life stages. Furthermore, large sample sizes and dense sampling of genetic loci is needed to ensure high power in GWA studies (Mackay, Stone, & Ayroles, 2009) and resolve any pleiotropic effects (Prescott et al., 2011). The genes influencing TL in humans that were identified through GWAS only explain a small proportion of the inter-individual variation in TL (<2 %, Aviv, 2012; Codd et al., 2013; Fyhrquist, Saijonmaa, & Strandberg, 2013). One GWAS on TL of a non-human species (dairy cattle, Bos taurus ) was recently performed (Ilska-Warner et al., 2019) supporting the polygenic nature of early-life TL. However, domesticated species in captivity may display TL dynamics that are not representative for natural populations (Eisenberg, 2011; Pepke & Eisenberg, 2021). There is a paucity of GWAS on TL performed in natural populations.
In this study, we aim to provide novel insights into the genetic architecture of TL and the evolutionary mechanisms by which natural selection can alter telomere ecology using data from a passerine bird. We sampled TL of most individuals (n =2746) born within 20 cohorts in two natural insular populations of wild house sparrows (Passer domesticus ) at about the same age (11 days), in addition to individuals at the same age in two insular populations that underwent artificial selection on body size for 4 consecutive years (n =569, Kvalnes et al., 2017; Pepke et al., 2021, submitted ). First, we estimate the phenotypic correlations between TL and tarsus length (as a proxy for body size, Araya-Ajoy et al., 2019) in house sparrow nestlings. Second, we test for effects of parental age on offspring TL. Third, we estimate heritability, environmental variances, and parental effects on early-life TL, and test for genetic correlations between TL, body size, and body condition in the natural populations (primary analyses). We then use similar analyses in the artificially selected populations to validate our results from the primary analyses. Finally, we use high-density genome-wide Single Nucleotide Polymorphism (SNP) genotype data (Lundregan et al., 2018) in a GWAS to identify genetic regions and potential candidate genes underlying variation in early-life TL within wild house sparrows (up ton =383).