Reproductive Drivers of Temporal Telomere Dynamics in Female Tree Swallows

ABSTRACT

Understanding trade‐offs between life‐history traits, particularly between reproductive investment and somatic maintenance, remains a major challenge in evolutionary ecology. Telomeres progressively shorten with age and have been suggested as good predictors of survival and longevity. Both intrinsic factors and external stressors, including high reproductive effort, accelerate telomere attrition. Thus, the dynamic nature of telomere length (TL) offers a promising avenue for assessing individual variation in the cost of reproduction. Here, we quantified telomere length (TL) in breeding female Tree swallows ( Tachycineta bicolor ) from a population breeding in southern Québec, Canada (2023–2024) to assess both cross‐sectional and longitudinal changes in TL relative to age, morphological traits, reproductive output, and habitat quality (breeding density). Cross‐sectional results showed an age‐dependent variation in TL, with evidence of a reproductive trade‐off among second year (SY) females with higher fledging success. Longitudinal analyses revealed that earlier breeding in spring was associated with short‐term telomere loss, but this trend was not maintained across breeding seasons. In contrast, habitat quality was not associated with telomere changes. Together, our findings support the use of telomere dynamics as a possible biomarker of reproductive effort at both population and individual level. Our results also underscore that the trade‐off between reproduction and telomere dynamics is contingent on the temporal scale, specific life history traits, and individual age.

1. Introduction

Studies of trade‐offs among life‐history traits are fundamental in evolutionary ecology and essential to understand how organisms optimise their fitness (Stearns 1992). For example, because reproduction is energetically costly, under limited resource availability, it is assumed to impose a cost on individuals, thereby negatively affecting their residual reproductive value or survival probability (Reznick 1985; Reznick et al. 2000; Williams 2018; Zera and Harshman 2001). However, such trade‐offs are not always detected within populations (Winder et al. 2025) because variation in resource acquisition may instead lead to positive associations between survival and fecundity (Landes et al. 2019; Laskowski et al. 2021; Roff and Fairbairn 2007; van Noordwijk and de Jong 1986).

Identifying the factors that drive the variations in the expression of trade‐offs between survival and reproduction and developing robust methods to quantify them remains a key challenge in ecological and evolutionary research. Telomere length (TL) dynamics have recently been proposed as molecular tools to help understand the factors that drive inter‐individual variation in life histories (Monaghan 2024; Monaghan et al. 2022; Monaghan and Haussmann 2006).

Telomeres are non‐coding repeated DNA sequences that serve as the end cap of eukaryotic cell chromosomes (Blackburn 2005). They prevent DNA depletion during cell division, a consequence of replication problems, making them essential for chromosome stability and inherently susceptible to shortening (Greider 1996). Thus, TL is a dynamic trait resulting from a balance between loss and repair mediated by the telomerase enzyme. Telomeres generally shorten with age, both in vitro (Harley et al. 1990) and in vivo (Bakaysa et al. 2007; Bichet et al. 2020; Kärkkäinen et al. 2022; Remot et al. 2022; Spurgin et al. 2018), until they reach a critical length that induces cellular senescence. At the interspecific level, longer telomeres are generally associated with higher survival (Bize et al. 2009; Brown et al. 2022; Chik et al. 2024; Haussmann et al. 2005; Wilbourn et al. 2018) or are a predictor of longevity (Haussmann et al. 2003; Heidinger et al. 2012; Vedder et al. 2022). Telomeres are also vulnerable to oxidative damage (Oikawa et al. 2001; von Zglinicki 2002). Stressful experiences are thought to accelerate shortening through oxidative stress, although the strength of this relationship varies across studies (Armstrong and Boonekamp 2023; Epel et al. 2004; Reichert and Stier 2017; Schultner et al. 2014; von Zglinicki 2002). As a result, TL can differ significantly among individuals of the same age class and can serve as a biomarker of the biological state and cumulative experience (Mather et al. 2011; Monaghan 2010; Pepke 2024).

TL attrition has been associated with reproduction (Bauch et al. 2016; Bichet et al. 2020; but see Sudyka 2019). Reproduction is often considered a stressful event that is physiologically costly (Bergeron et al. 2011). The elevated energy expenditure associated with breeding activities can reduce the defense against oxidative damages (Blount et al. 2016; Losdat et al. 2011), thereby negatively affecting telomere maintenance. As telomere attrition can be perceived as a direct cost of reproduction, TL for a given age may reflect the variability of trade‐offs between somatic maintenance and reproduction (Monaghan and Haussmann 2006). Based on this hypothesis, TL could serve as a biomarker of the cost of reproduction, with higher reproductive effort leading to accelerated telomere shortening and consequently reduced survival (Chatelain et al. 2020). Several studies have linked telomere loss to earlier laying dates in spring (Bauch et al. 2013; Graham et al. 2019), larger clutch size (Reichert et al. 2014; Sudyka et al. 2014), or higher reproductive success (Bauch et al. 2016; Bichet et al. 2020; Sudyka et al. 2019). Studying TL dynamics could also help uncover the variability in reproductive effort in natural settings due to individual heterogeneity, such as experience gain with age, phenotypic differences, or habitat quality. Low access to resources or harsh environments can directly accelerate telomere shortening (Criscuolo et al. 2020; Mizutani et al. 2013). For example, limited resource availability accentuates telomere shortening in Seychelles warblers ( Acrocephalus sechellensis ) having a higher number of offspring, whereas a high‐quality environment allows for better reproductive performance without a detectable cost from telomere length (Brown et al. 2022).

Longer telomeres have also been associated with a better reproductive performance (Angelier et al. 2019; Le Vaillant et al. 2015) and a higher reproductive success (Bauch et al. 2020; Parolini et al. 2017). For example, female Barn swallows ( Hirundo rustica ) with longer telomeres had higher hatching success and a greater number of fledglings (Parolini et al. 2017), thus reflecting individual heterogeneity. The breeder quality hypothesis suggests that TL reflects differences between individuals in their overall condition and, consequently, their fitness (Angelier et al. 2019; Heidinger et al. 2021). Higher performing individuals would show both higher longevity and reproductive success while also mitigating the effects of stress and preventing telomere loss (Tobler et al. 2022). Intraspecific variation in TL, along with its associations with survival and reproduction, makes it a valuable tool for understanding how life‐history trade‐offs are modulated by individual differences and life stages (Monaghan et al. 2022; Monaghan and Haussmann 2006).

A review by Sudyka et al. (2019) proposed that the dichotomy in the literature regarding the relationship between TL and reproduction might be due to differences in scales. TL can serve as a biomarker of both reproductive effort at the individual level and performance at the populational level. When considering only between‐individual differences in TL, a positive association with reproductive traits may emerge, reflecting breeding capacity in favour of the breeder quality hypothesis (Sudyka 2019). On the other hand, longitudinal studies with repeated within‐individual measures could reveal the cost of reproduction, as higher reproductive effort may accelerate telomere shortening. Independent of the initial TL, telomere attrition could be a biomarker of reproductive effort at the individual level.

In this study, we explored the factors affecting TL dynamics in wild female Tree swallows ( Tachycineta bicolor ) breeding in southern Québec, Canada, over two years. More specifically, we assessed the relationships between individual traits (age, body condition, and reproductive traits), environmental quality, and TL variation between individuals and within individuals at intra‐ and inter‐seasonal levels. As recommended by Sudyka et al. (2019), we included two scales in our analyses to better understand the association between TL and reproduction. First, at the population scale, we evaluated whether TL could be used as a biomarker of reproductive performance of females, using data on their laying date, clutch size, and fledging success. We assessed the breeder quality hypothesis, whereby females with longer telomeres tend to have better reproductive performance. We predicted that longer telomeres would be found in females with earlier laying dates, larger clutches, and a higher fledging success. At the individual scale, we expected to find telomere variability related to individual differences. We thus expected that longer telomeres would be found in younger females and in females with higher mass (indicative of better body condition) and longer wings (indicative of larger individuals). Within‐individual changes in TL may also serve as a biomarker of reproductive effort, highlighting the cost of reproduction. We evaluated whether individual telomere attrition was associated with higher investment in the same reproductive traits. In addition, by analysing TL changes over short‐term (intra‐seasonal) and long‐term (inter‐seasonal) scales, we could assess whether telomere loss was temporary or persisted in the following year. We predicted that females with better reproductive performance, such as earlier laying dates, larger clutches, and higher fledging success, would experience greater telomere attrition. However, if individual differences mitigate the physiological costs of reproduction, greater telomere loss could instead occur in females with poorer reproductive performance. Finally, we predicted that females breeding in lower‐quality habitat would experience increased telomere attrition.

Tree swallows are small migrating insectivorous passerines that breed from May to July in southern Québec (Knight et al. 2018). Females generally produce one clutch per season, consisting of approximately five to six eggs but only three to four fledglings (Ghilain and Bélisle 2008; Pellerin et al. 2016; Poisson et al. 2021). Older females typically have earlier laying dates, larger clutches, and higher fledging success (Carle‐Pruneau et al. 2022; Houle et al. 2020; Poisson et al. 2021). Larger females (body mass, wings) are able to lay earlier and have more eggs (Nooker et al. 2005), as well as produce larger offspring (Pellerin et al. 2016; Poisson et al. 2021) with a higher fledging success (Gossieaux et al. 2019). We thus expected that breeding should be costlier for inexperienced younger females than for older females and for individuals in worse body condition, for similar reproductive traits. The breeding habitat quality is also an essential determinant of reproductive success in our study system, primarily because variation in insect availability is influenced by landscape heterogeneity (Baeta et al. 2012; Bellavance et al. 2018; Garrett et al. 2022).

A decline in TL with age has already been documented in Tree swallows (Belmaker 2016; Haussmann et al. 2003; Vleck et al. 2011), although considerable heterogeneity in TL is present within a given age class (Haussmann et al. 2004). Notably, shorter telomeres were found in individuals who produced larger nestlings, suggesting that better reproductive performance is costly for telomere maintenance (Ouyang et al. 2016). In contrast, another study showed that females with higher mass and earlier laying dates had lower expression of the POT1 gene, which may allow greater resilience to TL loss via telomerase activity, thus supporting the breeder quality hypothesis (Wolf et al. 2022).

2. Materials and Methods

2.1. Study System and Data Collection

The study system, located in southern Québec (Canada), comprises 200 nest boxes equally distributed among 20 farms and covers an area of approx. 10,200 km². Breeding activity in this system was monitored since 2004. In this study, we used data collected over two reproductive seasons (2023 and 2024) from May to July, during which we visited all nest boxes every two days. We collected data on every nest present at the study site. We recorded the laying date (Julian day, when the first egg was laid), clutch size (number of eggs laid; 3 eggs or more), cumulative mass of nestlings (total mass of all nestlings within a clutch in grams at 16 days of age, used as an indicator of parental investment in nestlings), number of fledglings, and fledging success (calculated as the proportion of the number of fledglings relative to the clutch size). The age of each female was visually determined based on plumage coloration and classified into two categories: the second year (SY) for individuals with brown feathers and after the second year (ASY) for those with blue‐green feathers (Hussell 1983). SY and ASY represent different life stages that differ consistently in reproductive traits. We assigned a minimum age at first capture of 1 year if individuals were first captured as SY or 2 years if they were first captured as ASY. At each subsequent recapture, minimum age was updated by adding the number of years elapsed since the previous capture. Because some individuals enter the study population only after their early breeding years, their exact age cannot be determined with certainty.

We used two proxies to characterise environmental quality. First, we used the breeding density of Tree swallows (number of occupied nest boxes on each farm, from 0 to 10), where high occupancy rates of nest boxes on farms are indicative of higher‐quality habitats (Bourret et al. 2015). Higher farm density is associated with better reproductive performance, indicating habitat selection of more favourable locations. Second, we used the breeding density of House sparrows ( Passer domesticus ), quantified using the same approach, as indicative of lower‐quality habitats, since they are the main competitors of Tree swallows for nest boxes during reproduction (Carle‐Pruneau et al. 2022; Robillard et al. 2013). Nest boxes are rarely all used by both species on a given farm (Carle‐Pruneau et al. 2022). Therefore, the two proxies are only weakly inversely correlated (r = −0.15, p < 0.01).

2.2. Captures and Blood Sampling

All females were captured directly in their nest box during incubation and after hatching of nestlings with a plexiglass trap. Blood sampling was performed during two distinct periods of the reproductive season. The first blood sample was collected at least four days after the start of incubation. At this time, we also measured body mass (±0.01 g) and wing length (±1 mm) and identified all new individuals with an aluminium band (Canadian bird banding office). The second blood sample was collected at least 10 days after the first sample and four days after hatching of nestlings. On average, a 12‐day (SD = ±1.62, range = 10–18) interval separated the two samples (Figure 1). Body mass was measured again at that time. A total of 381 unique blood samples were obtained during the whole study. For both years, a total of 206 females were sampled during the first period, and 175 were re‐sampled during the second period. A total of 46 females were sampled for blood collection in both 2023 and 2024. The recapture proportion (35%) is in line with past estimated recapture probabilities in our study system (Lagrange et al. 2014). The average daily turnover rate of red blood cells (RBCs) DNA in birds is 3% (Glomski and Pica 2011). For example, in Zebra finches ( Taeniopygia guttata ), another passerine, RBCs' lifespan was estimated at 19 days (Bauchinger and McWilliams 2009). Thus, a substantial proportion of RBCs has likely been renewed between our two sampling periods, which may reflect changes in telomere length. At each sampling period, 70 μL of blood was collected by puncturing the brachial vein with a needle and kept in a heparinised capillary tube to prevent coagulation. Tubes were sealed with clay and kept in a cooler until returning to the laboratory at the end of the day, where the samples were stored at −80°C until DNA extraction approximately two months later (Criscuolo et al. 2009; Reichert and Stier 2017).

FIGURE 1.

Timeline of blood sampling of female Tree swallows. Blood sampling (shaded areas) during the breeding season to obtain rTL‐I (during incubation) and rTL‐H (after hatching) of female Tree swallows. Blood samples were collected at intervals of at least 10 days. Mean seasonal rTL was calculated from the two rTL measures. Tick marks represent 2‐day intervals.

2.3. Telomere Length Analysis

Blood tissue DNA was extracted using the DNeasy Blood and Tissue Kit (Qiagen) according to the manufacturer's protocol in a uniform and systematic manner for all samples (Eastwood et al. 2018). DNA concentration and integrity were measured using a NanoDrop spectrophotometer. To minimise freeze–thaw events, all DNA samples were stored at their initial concentration at −80°C until analysis. TL was measured using the qPCR method developed by Cawthon (2002) and based on the protocol adapted for birds by Criscuolo et al. (2009). This method provides a ratio (T/S) of telomere length (T) relative to a single gene control length (S, which is fixed in length across individuals of the species). Although the measure is relative (henceforth rTL), it is highly correlated with the absolute measurement obtained using the restriction fragment method (TRF) (Criscuolo et al. 2009). In addition, this method is more time efficient, requires less blood (approximately 70 μL), and is therefore less invasive for smaller passerines (Nussey et al. 2014). The control gene, glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH), was amplified from the primers, developed by Criscuolo et al. (2009), GAPDH‐F (5′‐AAC CAG CCA AGT ACG ATG ACA T‐3′) and GAPDH‐R (5′‐CCA TCA GCA GCC TTC A‐3′). Telomeric DNA was amplified using telg (5′‐ACA CTA AGG TTT GGG TTT GGG TTT GGG TTT GGG TTA GTG T‐3′) and telc (5′‐TGT TAG GTA TCC CTA TCC CTA TCC CTA TCC CTA TCC CTA ACA‐3′) primers, as used by Wolf et al. (2022) in Tree swallows. Reactions were performed with 0.5 μL of each primer (10 μM) for each respective gene and 4 ng of DNA, diluted with double‐distilled sterilised water. The final volume of 10 μL was completed with 5 μL iTaq Universal SYBR Green Supermix (Bio‐Rad) and 2 μL H₂O. The samples were analysed in a random order in triplicate in 96‐well plates, but separately for each year. Amplification reactions were performed on two separate plates for each gene, but in the same well position (Eisenberg et al. 2015). PCR parameters for the GAPDH gene were 5 min at 95°C, followed by 40 cycles of 15 s at 95°C, 30 s at 55.5°C and 5 s at 65°C. For telomeres, the parameters were 5 min at 95°C followed by 40 cycles of 15 s at 95°C, 30 s at 50.5°C and 5 s at 65°C. Each plate included a negative control and a serial dilution (12, 6, 3, 1.5 and 0.75 ng per well) in duplicate, obtained from one of our high‐quality samples and initial concentration. Serial dilutions were prepared in advance and maintained for approximately one week in aliquots at −80°C. Serial dilution amplification generated a standardised curve that provided the amplification efficiency (E) of the plate, with acceptable ranges of 100% ± 15% and R ² ≥ 0.98. The mean GAPDH efficiency was 103.89% (±8.16), and the mean telomere efficiency was 102.81% (±7.95). The dilution series also provided the reference threshold to obtain the fractional number of PCR cycles for a sample to accumulate enough products to reach the fluorescent signal magnitude (Cq). A total of 55 GAPDH gene plates and 57 telomere plates were run. These plates also included nestlings and males samples that were part of a separate ongoing study, hence the high numbers. If the triplicate Cq of a single sample had a standard deviation (SE) > 0.2 or fell outside the range of values of the dilution series, one replicate was dropped. If the remaining values did not meet the criteria, the entire sample was excluded from analysis. Based on these criteria, we excluded 95 samples from our analysis. From the remaining samples, we obtained the intra‐assay repeatability of Cq with intra‐class correlation (ICC) using the rptR package (Eisenberg et al. 2015; Stoffel et al. 2017). The model included sample identity as a random effect with 1000 bootstrap iterations. The triplicate repeatability estimation for the GAPDH gene was 0.951 ± 0.002 (CI = 0.946, 0.955) and 0.966 ± 0.002 (CI = 0.962, 0.969) for telomeres. A Golden Sample (GS), a DNA mixture of 100 randomly selected samples, was included in triplicate on each plate as a reference sample to normalise rTL and account for inter‐plate variation. GS was prepared prior to analysis and stored in aliquots at a concentration of 2 ng/μL to prevent freeze‐thawing.

We calculated rTL using the model presented by Pfaffl (2001) (Equation 1).

$$\mathit{rTL}=\frac{{E_{\mathit{\text{Telo}}}}^{∆{\mathit{Cq}}_{\mathit{\text{Telo}}}\left({\mathit{Cq}}_{\mathit{GS}}-{\mathit{Cq}}_{\mathit{Ech}}\right)}}{{E_{\text{GAPDH}}}^{∆{\mathit{Cq}}_{\mathit{\text{GAPDH}}}\left({\mathit{Cq}}_{\mathit{GS}}-{\mathit{Cq}}_{\mathit{Ech}}\right)}}$$ (1)

The rTL equation includes the plate efficiency of both telomeres (E _Telo ) and the GAPDH gene (E _GAPDH ). An efficiency of 100% corresponded to ideal amplification, where the product doubled with each cycle (E = 100% = 2). The Cq values were standardised by the deviation between the mean Cq of the GS (Cq _GS ) and that of the unknown sample of interest (Cq _Ech ).

We assessed rTL inter‐assay repeatability, with three repeated samples across the plates, as well as the golden sample for each year (n = 85 in 2023, n = 64 in 2024). The extrapolated repeatability (Re) calculated to consider that rTL was obtained on multiple plates as explained in Nakagawa and Schielzeth (2010) was 0.899 for 2023 and 0.874 for 2024.

2.4. Statistical Analyses

A list of all variables included in the models are listed in the Table 1. First, at the population‐level we assessed the relationship between individual rTL variation obtained during the incubation period (n = 173, 81 in 2023, and 92 in 2024) and individual traits using a linear mixed effect model. We considered only data from the first clutch in all the analyses. The full model included age class (SY or ASY), mass (g), wing length (mm), laying date (Julian day), clutch size (3 eggs or more), fledging success and year. Two‐way interactions between age class and individual traits (mass, wing length) as well as reproductive traits (laying date, clutch size, fledging success) were included as fixed effects because SY and ASY females differ markedly in both morphology and breeding performance. These age differences could possibly modify the relationship between these traits and rTL. The capture time (hours of the day) was also added to account for daily temporal variability in the mass (Rioux Paquette et al. 2014). We also included female identity and qPCR plate identity (to account for possible differences in sample handling) as random effects. The variables number of fledglings and their cumulative mass at 16 days old were excluded because they were both highly correlated with the fledging success (r = 0.90 and r = 0.81, respectively). The final model was derived using backward selection based on sequential Likelihood Ratio Test comparisons, removing at each step the least significant variable (based on α < 0.05). During the two years of our study, some females categorised as ASY may have been as old as six years yet were grouped in the same category as two‐year‐old individuals. Given this potential uncertainty in age estimation of females, we performed an additional analysis using minimal age rather than age categories (see Supporting Information S1 for details). We ran a series of models to determine which age distribution best explained telomere dynamics with aging using AICc (see model list in Table S1). We found that both the quadratic cross‐sectional model (Table S2a) and the linear longitudinal model (Table S2b) had ∆AICc < 2, with a positive quadratic trend best explaining the effect of minimal age on rTL dynamics. We thus refitted a model to assess between individual rTL variation with the quadratic age distribution, and the same fixed and random effects used above.

TABLE 1.

List of variables included in full models to test telomere dynamic variability at the population and individual levels in female Tree swallows.

Response variables	Fixed effects		Random effects
rTL‐I	Age (SY or ASY)		Female ID
	Year (2023 or 2024)		qPCR plate ID
	Body condition	Mass
		Hour of capture
		Wing length
		Mass × Age
		Wing length × Age
	Reproductive traits	Laying date
		Clutch size
		Fledging success
		Laying date × Age
		Clutch size × Age
		Fledging success × age
rTL‐H	Age (SY or ASY)
	Year (2023 or 2024)
	Days between samples
	Body condition	∆Mass
		Wing length
		∆Mass × Age
		Wing length × Age
	Reproductive traits	Laying date
		Clutch size
		Fledging success
		Laying date × Age
		Clutch size × Age
		Fledging success × age
	Habitat quality	Swallow breeding density
		Sparrow breeding density
		Swallow breeding density × Age
	rTL‐H
Mean rTL 2024	Minimal age
	∆Capture date
	Body condition	∆Mass
		Wing length (2023)
	Reproductive traits Mean of both years	Laying date
		Clutch size
		Fledging success
	Habitat quality Mean of both years	Swallow breeding density
	Mean rTL 2023

Note: rTL at incubation (rTL‐I) was used as the response variable for population‐level analyses. For individual‐level analyses, rTL at hatching (rTL‐H) was used as the response variable for intra‐seasonal changes, whereas mean rTL in 2024 was used for inter‐seasonal changes.

Second, we used linear models to assess the relationship between relative change in rTL within individuals and both individual traits and environmental quality at the intra‐ and inter‐seasonal levels. First, the intra‐seasonal difference in rTL was assessed using the rTL measured at post‐hatching (rTL‐H) as the response variable in the model, while also including initial rTL, obtained during the incubation period (rTL‐I), as a covariable (n = 110, 48 in 2023 and 64 in 2024). Other variables included in the full model were age class (SY or ASY), ∆mass (difference in mass between captures, g), wing length, laying date, clutch size, and fledging success as individual variables. Mass usually changes within an individual throughout the season by an average of 2.6 g in our study (±1.3). The model also included the breeding densities of Tree swallows, House sparrows, and year as environmental variables. All two‐way interactions with age class were also included to account for age differences in reproductive performances. To account for temporal differences between the two capture periods, the number of days between sampling events was also included as an explanatory variable. The final model was obtained through backward selection, as explained above. Finally, the relative inter‐seasonal differences in rTL were assessed using the mean rTL measured in 2024 as the response variable in the model, while also including the initial rTL (mean rTL in 2023) as a covariable. We obtained the rTL of 33 females in both years. We used minimal age in 2023, as all females captured subsequently in 2024 were classified as ASY. Other variables included in the full model were ∆mass (differences in mass at first capture between years), wing length (mm, in 2023), mean reproductive trait values (laying date, clutch size, and fledging success), and mean breeding densities of Tree swallows in both years. A significant decline in mass between years was previously found for females in this study system (Rioux Paquette et al. 2014). We could not include House sparrows breeding density because there was no variability in this variable (all females were captured on farms without sparrows). The ∆capture date (difference in Julian days of the first capture between years) was also added to account for temporal variability. The final models were also obtained by backward selection. All full models can be found in Tables S3–S6. All statistical analyses were conducted in the software R (version 4.4.0) with the lme4 and lmerTest packages (Bates et al. 2015; Kuznetsova et al. 2017; R Core Team 2024).

3. Results

3.1. Between‐Individual (Populational) Telomere Variability

The final models explaining rTL variation among females are presented in Table 2. From the categorical age distribution (Table 2A), we found an interaction between wing length and age class, showing that the relationship between wing length and rTL was steeper and more positive for SY than for ASY females (Figure 2A). We also found an interaction between age and fledging success associated with rTL: the decline of fledging success with rTL was steeper in SY than in ASY females; ASY females showed no decline (Figure 2B). Clutch size was kept in the final model as a statistical control for fledging success. Although non‐significant, we found a negative relationship between clutch size and rTL. Laying date was not associated with rTL. The full model yielded similar estimates, although the interaction between age class and wing length was only marginally non‐significant (Table S3). From the quadratic age distribution model (Table 2B), we found that rTL tended to be longer in younger and older females and shorter at an intermediate age (Figure 2C).

TABLE 2.

Estimates with the standard errors (SE) from final linear mixed effect model analysis of inter‐individual rTL variability for female Tree swallows (N = 173) over two years (2023–2024) with (A) categorical age and (B) quadratic age distribution.

Models	Variables	Estimates	SE	t‐values	CI	p
(A) Categoric	Intercept	−2.644	2.152	1.23	[−6.89; 1.61]	0.22
	Age (ASY)	5.100	2.512	2.03	[0.14; 10.06]	0.044
	Wing length	0.037	0.019	1.93	[−0.00; 0.07]	0.06
	Clutch size	−0.046	0.028	1.63	[−0.10; 0.01]	0.11
	Fledging success	−0.213	0.128	1.66	[−0.47; 0.04]	0.10
	Wing length × Age (ASY)	−0.047	0.022	2.12	[−0.09; −0.00]	0.035
	Fledging success × Age (ASY)	0.309	0.122	2.15	[0.02; 0.59]	0.033
	Random Effects	Variance
	Female ID	0.04
	qPCR plate ID	0.04
	Residual	0.06
Marginal/conditional R 2: 0.061/0.596
(B) Quadratic	Intercept	1.333	0.115	11.61	[1.11; 1.56]	< 0.001
	Minimal age	−0.192	0.084	2.28	[−0.36; −0.03]	0.024
	Minimal age 2	0.030	0.013	2.29	[0.00; 0.06]	0.023
	Random Effects	Variance
	Female ID	0.05
	qPCR plate ID	0.05
	Residual	0.05
Marginal/conditional R 2: 0.023/0.671

Note: Bold indicates a p‐value < 0.05 or CIs not overlapping zero. Full model estimates can be found in Tables S3 and S4.

FIGURE 2.

Between‐individual rTL relationships with individual characteristics in female Tree swallows. Model predictions from the linear mixed effect models of variation in rTL in female Tree swallows. Prediction of rTL as a function of the interaction between age class and (A) wing length (mm) and (B) fledging success. (C) Nonlinear relationship between rTL and minimal estimated age in female Tree swallows. Points represent individual data, and regression lines are shown separately for SY (open circles, dotted lines) and ASY (black circles, dashed lines) individuals. Shaded areas represent 95% confidence intervals. The dashed line at rTL = 1.0 serves as a reference.

3.2. Within‐Individual Telomere Dynamic (Intra‐Season)

We found no overall trend of telomere shortening or elongation at the intra‐seasonal level between the incubation and the post‐hatching period (Figure S1). Our analysis of relative changes in telomeres revealed that laying date was positively significantly related to rTL during the post‐hatching period (Table 3 and Figure 3). A trend toward greater telomere shortening with increasing clutch size was observed in the full models but was not retained in the final model. Fledging success was not associated with intra‐seasonal telomere dynamics. No other individual or environmental variables were found to be significant in the final model or the full model.

TABLE 3.

Final linear model analyses of relative rTL change variability, while accounting for baseline telomere length, for female Tree swallows, at both the (a) intra‐seasonal level (rTL _H ; N = 110) and (b) inter‐seasonal level (mean 2024 rTL; N = 33).

Model	Variables	Estimates	SE	t‐value	CI	p
(a) Intra‐season	Intercept	−0.402	0.742	0.54	[−1.29; 1.55]	0.59
	rTL incubation	0.095	0.106	0.89	[−1.12; 0.31]	0.37
	Days between samples	−0.019	0.022	0.85	[−0.07; 0.01]	0.40
	Laying date	0.011	0.005	2.36	[0.00; 0.02]	0.020
R 2 adjusted: 0.036
(b) Inter‐season	Intercept	0.757	0.159	4.76	[0.43; 1.08]	< 0.001
	Mean 2023 rTL	0.337	0.128	2.64	[0.08; −0.60]	0.013
	Days between samples	0.009	0.007	1.20	[−0.01; 0.02]	0.24
R 2 adjusted: 0.203

Note: Bold indicates a p‐value < 0.05 or CIs not overlapping zero. Full model estimates can be found in Tables S5 and S6.

FIGURE 3.

Relative intra‐seasonal changes in rTL depending on laying date in female Tree swallows. Predictions from the linear models of variation in post‐hatching rTL in female Tree swallows, while accounting for baseline rTL during incubation, as a function of laying date (Julian date). Shaded areas represent 95% confidence intervals. The dashed line at rTL = 1 serves as a reference.

3.3. Within‐Individual Telomere Dynamic (Inter‐Season)

Again, we found no consistent trend in telomere shortening or elongation at the inter‐seasonal level (Figure S2). The final model revealed that the baseline initial mean rTL from the previous year (2023) was the main predictor of mean rTL in the following year (2024), with no other variables being significant (Table 3 and Figure 4). Neither laying date, clutch size, or fledging success was associated with telomere dynamics between the two reproductive seasons.

FIGURE 4.

Relative inter‐seasonal rTL changes in female Tree swallows. Predictions from the linear models of variation in mean rTL in 2024 in female Tree swallows as a function of mean rTL in 2023. Points represent individual data, and shaded areas represent 95% confidence intervals. The red dashed line serves as a reference, indicating no change in telomere length between years.

4. Discussion

We assessed the relationship between rTL and multiple reproductive traits across two breeding seasons and at both the between‐ and within‐individual levels in female Tree swallows. In younger (SY) females, shorter telomeres were observed in individuals with a higher fledging success. We also found that temporal scales influenced individual rTL variability. At the intra‐seasonal scale, shorter telomeres were associated with earlier laying dates. At the inter‐seasonal scale, variation in telomere length was related to initial baseline telomere length: individuals with longer rTL in 2023 also had longer telomere length the following year. We thus found no evidence for the breeder quality hypothesis, as differences in rTL were not associated with improved reproductive performance. Instead, our results partly support the role of telomeres as a biomarker of reproductive costs at both the between and within individual levels. Surprisingly, habitat quality didn't impact telomere changes.

We also found that while telomeres were shorter in older females, there was important heterogeneity within age classes.

4.1. Between‐Individual Telomere Variability

Contrary to previous suggestions, we found no indication that rTL is an adequate biomarker of inter‐individual variation in reproductive performance, and thus no support for the breeder quality hypothesis (Le Vaillant et al. 2015; Parolini et al. 2017). Between‐individual rTL variability instead reflected a possible trade‐off with reproduction, but only in SY females, where a higher fledging success was found for individuals with shorter telomeres. Our results are thus in line with previous findings, such as in the Common terns ( Sterna hirundo ), where telomeres were shorter when the number of fledglings produced in the previous year was higher (Bichet et al. 2020). The specificity of this trade‐off for SY females highlights possible age‐related differences in reproductive performance. SY individuals prioritizing self‐maintenance may have done so at the expense of investment in current reproductive success (see Ardia and Clotfelter 2007). Younger females are less experienced, possibly explaining the more apparent trade‐off between reproductive performance and telomere length (Forslund and Pärt 1995; Robertson and Rendell 2001).

We also found a weak, albeit non‐significant, negative effect of clutch size on rTL for both age classes. This suggests that the trade‐off between fledging success and rTL may reflect an intrinsic limitation in parental capacity, rather than simply being driven by brood size.

Telomere length tended to be longer in younger, but also in older females, with shorter telomeres being found at intermediate ages. This result suggests that telomeres tend to shorten during the first years of life and that individuals with longer telomeres are more likely to survive to older ages (Haussmann and Mauck 2007). Only a small number of females in our population reached advanced ages. Because longer telomeres are widely accepted as predictors of increased longevity (Bize et al. 2009; Haussmann et al. 2005), our result supports the contention that individuals with longer telomeres follow a life‐history trajectory characterised by better somatic maintenance and a higher probability of survival.

Large variability in rTL was found within age classes. As reported in previous studies on birds (Angelier et al. 2019; Benowitz‐Fredericks et al. 2022; Hall et al. 2004; Pauliny et al. 2012), this result suggests that telomere length is not solely explained by age. Specifically, variation in rTL was substantial among SY females and approximately one‐tenth of this variation was explained by morphological differences. Among younger females, individuals with longer wings tended to have longer telomeres. Longer wings enhance flight efficiency, reducing energetic costs of migration (Norberg 1995) and shortening the duration of transit between wintering and breeding grounds (Hahn et al. 2016). In contrast, individuals with shorter wings likely migrate more slowly and require more frequent stopovers due to the higher energy demands (Nilsson et al. 2013). Such constraints may limit energy allocation for somatic maintenance, potentially leading to accelerated telomere attrition. Indeed, migration was previously found to be costly on telomeres in migratory birds (Bauer et al. 2016). An association between rTL and wing length was not observed in the older ASY age class. This suggests that telomere length captures individual heterogeneity in body size during early life stages, a signal that may be weaker later in life due to accumulating external influences. A differential mortality selection could also occur between the SY and ASY age classes. SY females with longer telomeres may be more likely to survive and reproduce again, hence explaining the reduced variability in the ASY age class.

We also note that a small portion of the variation in rTL was explained by fixed effects in our model (marginal R ² = 0.061). Female identity accounted for a substantial proportion of the remaining variability (28.6%), suggesting that additional maternal environmental or genetic effects may also play a role in determining rTL variations (Bauch et al. 2022; Chik et al. 2022).

4.2. Intra‐Seasonal Within‐Individual Telomere Variability

At the intra‐seasonal level, earlier laying dates were associated with shorter telomeres at the end of the season, when controlling for initial rTL at incubation. Early initiation of reproduction was associated with short‐term loss of telomeres, suggesting that laying earlier led to greater reproductive effort. A similar effect of early breeding on telomere dynamics has been reported in Dark‐eyed juncos ( Junco hyemalis ) (Graham et al. 2019). Additionally, Common terns that arrived and bred earlier had shorter telomeres (Bauch et al. 2013). Although early breeding often can increase annual reproductive success (Berzins et al. 2020; Tremblay et al. 2025), it is often constrained by harsher early‐season environmental conditions, which may explain the observed negative effects on telomere maintenance (Dunn et al. 2011). In our study system, Diptera, the main prey of swallows, are less abundant early in the season, especially if temperatures are low (Bellavance et al. 2018). As a result, females that lay earlier may have reduced energy availability to allocate toward telomere maintenance (Benowitz‐Fredericks et al. 2022; Criscuolo et al. 2020). A previous study has also shown that colder temperatures during early spring are associated with increased telomere shortening (Graham et al. 2019), likely driven by elevated oxidative stress (Stier et al. 2014). Telomere dynamics at the intra‐season scale in our study system therefore reflect the increased reproductive effort associated with laying eggs earlier to maximise the likelihood of reproductive success.

The absence of a consistent change in rTL for females during reproduction may be partly explained by the short time interval between our measurements. The intra‐seasonal change in rTL captured telomere dynamics prior to fledging of juveniles and after incubation, reflecting only the cost of early reproductive activities such as laying initiation, part of the incubation, and some parental care. In birds, the increased reproductive effort associated with a larger clutch is primarily driven by provisioning activities by parents (Boonekamp et al. 2020). Additional rTL measurements taken at later breeding stages might have more consistently captured the cost of reproduction on telomere shortening. Furthermore, our initial rTL measurements taken during incubation could not represent true baseline values, as several potentially stressful events occurring beforehand could not be captured. Considering these limitations, intra‐seasonal changes in rTL do not fully capture the overall seasonal reproductive effort.

4.3. Inter‐Seasonal Within‐Individual Telomere Variability

We found no clear pattern of inter‐seasonal rTL shortening. This may reflect a selective disappearance of frail individuals, which could bias our sample, as the longitudinal analysis included only individuals that returned to the study area between 2023 and 2024. As females with higher reproductive success have a greater probability of recapture the following year (Lagrange et al. 2017), we may have captured the telomere dynamics of individuals with higher fitness, who could also experience reduced telomere attrition.

When considering relative inter‐seasonal changes in telomere length, mean initial rTL in 2023 emerged as the main predictor of the subsequent year mean rTL. It was previously found that baseline rTL was an important predictor of telomere shortening (Pepke et al. 2023; Salomons et al. 2009; Verhulst et al. 2013). The relationship presented on Figure 4 suggests that females with shorter telomeres in 2023 were less likely to experience attrition than females with longer telomeres. Telomere maintenance may depend on the initial length, with individuals possessing shorter telomeres potentially investing more in their repair, partly by the higher activity of telomerase (Nordfjäll et al. 2009). Conversely, longer telomeres may be more susceptible to oxidative damage (Aviv et al. 2008).

We found no effect of reproductive traits on mean rTL in 2024 at the inter‐season level, when accounting for baseline mean rTL in 2023. Therefore, the cost of reproduction of early laying on telomere dynamics appears to be temporary and does not persist into the following year. Between‐individual variability in baseline rTL thus explained a substantial proportion of the variation in female telomere dynamics at the inter‐seasonal level, independently of past reproductive experience or of habitat quality.

The number of reproductive events has been linked to faster annual telomere shortening in other passerines (Heidinger et al. 2012). However, due to our study design, where we captured only breeding individuals, we were also unable to compare telomere dynamics between breeding and non‐breeding individuals, which might have limited our capacity to detect a cost of reproduction using rTL.

4.4. Conclusion

Our findings on telomere dynamics in female Tree Swallows partly support the cost of reproduction hypothesis at both the between and intra‐individual levels. Between‐individual variation in rTL revealed trade‐offs with fledging success, specifically for first‐time breeders. Within individuals, relative changes in rTL from incubation to the post‐hatching period may serve as biomarkers of reproductive effort in relation to laying date.

Our study provides new insights into the variation in telomere dynamics in a wild short‐lived bird both at the population and individual levels. We observed a general trend of telomere shortening with age, particularly marked early in life, suggesting that this period is critical for telomere attrition. However, rTL remained highly variable within life stage, indicating considerable individual heterogeneity associated with reproductive traits and wing length. Using telomeres as biomarkers may help clarify how changing reproductive conditions influence the magnitude of the trade‐off with somatic maintenance (Monaghan et al. 2022).

Author Contributions

Juliette Barbeau conceived the study, collected the data, performed the analyses, interpreted the results, drafted the manuscript, and contributed to its revision. Dany Garant contributed to the study conception, revised the analyses, participated in the interpretation of the results, contributed to the manuscript revision, secured funding, and is co‐responsible with Fanie Pelletier for the field work and long‐term monitoring of tree swallows. Fanie Pelletier contributed to the manuscript revision, secured funding, and is co‐responsible with Dany Garant for the field work and long‐term monitoring of tree swallow.

Funding

This work was supported by the Natural Sciences and Engineering Research Council of Canada (549146 and 2024‐05209), Canada Research Chairs (CRC‐2022‐00486), Fonds de recherche du Québec—Nature et technologies (329093).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Model comparison based on Akaike's Information Criterion (AIC) of the relationship between individual rTL variability and minimal estimated age. All models had female ID and qPCR plate ID as random effects.

Table S2: Details of two top models (from Table S1) explaining rTL dynamics in relation to minimal age in female Tree swallows (N = 173).

Table S3: Estimates with the standard errors (SE) from full linear mixed effect model analysis of inter‐individual rTL variability for female Tree swallows (N = 173) over two years (2023–2024).

Table S4: Estimates with the standard errors (SE) from full linear mixed effect model analysis of inter‐individual rTL variability with quadratic age distribution for female Tree swallows (N = 173) over two years (2023–2024).

Table S5: Estimates with their standard errors (SE) from full linear model analyses of relative rTL change variability, while accounting for baseline telomere length, for female Tree swallows at the intra‐seasonal level (N = 110).

Table S6: Estimates with their standard errors (SE) from full linear model analyses, without interactions, of relative rTL change, while accounting for baseline telomere length, for female Tree swallows at the inter‐seasonal level (N = 33).

Figure S1: Individual trends of intra‐seasonal rTL changes.

Figure S2: Individual inter‐seasonal rTL changes in female Tree swallows.

Data Availability Statement

The data supporting the findings of this study are openly available in Dryad at https://doi.org/10.5061/dryad.9s4mw6mzb.