Analyses of TL and life history traits
TROC was significantly but moderately negatively associated withAdult TL and Chick TL (respectively; r = -0.349, P = 0.04, n = 34; r = 0.514, P < 0.01, n = 24). However, the association of TROC and Adult TL was strongly influenced by two extreme points, Larus Andouiin i and Parus major , which presented long telomere lengths and high rates of annual erosion (ESM 1). If we exclude these two extremes, the association changed in sign and became non-significant (r = 0.127, P = 0.48, n = 32). Adult TL and Chick TL showed a strong positive association (r = 0.883. P < 0.0001, n = 28).
Using the scores of the different avian species on the principle components, we investigated the associations of TL with the three statistically independent life history axes. The three telomere variables (Adult TL , TROC and Chick TL ) showed small to medium and non-significant associations with phylogeny (Figure 2). Body size and the slow-fast continuum of pace of life (PC1, PC2) were strongly and significantly associated with the phylogenetic pattern. Parental care (PC3) exhibited a small-to-medium association with phylogeny, but a significant one.
We then investigated the associations of bird TL with the three principle component axes (that displayed phylogenetic patterns) using three different approaches. First, we explored direct Pearson’s correlations (without considering any phylogenetic effect) between TL variables and the principle components. This suggested a small-to-medium and non-significant correlation of Adult TL with body size (PC1; Table 2A, Figure 3A). TROC exhibited a medium positive association with body size (PC1) and a large negative association with the slow-fast pace of life (PC2; species greater reproduction and shorter life exhibited more strongly negative TROC ; Table 2B, Figure 4A, B). Correlations were at best small and non-significant between Chick TL and the life-history axes (Table 2C, Figure 5).
Next, univariate analyses that statistically adjusted both telomere variables and life-history data for the phylogenetic pattern confirmed only some of the ahistorical correlational analyses. First, it revealed that the relatively weak pattern of association of Adult TL and body size was likely due to the phylogenetic pattern (the residual “phylogeny-adjusted” correlation was -0.153; Table 2A). The same was evident for the moderate positive association of TROC with body size (the residual correlation was -0.108; Table 2B). The negative association of TROC and the slow-fast pace of life was confirmed as a strong pattern. Association between TROC and life-history axes exhibited especially strong phylogenetic patterns.
We next examined these potential associations while considering the effect of phylogeny on both variance and co-variance of TL and PC variables, using a multivariate phylogenetic framework. The correlation between PCs and TL variables due to phylogeny were marginal, since their posterior distributions mostly were broadly centred on or largely overlapped zero (Table 2, Figure 6). Only the negative association of TROC and slow-fast pace of life approached significance, and it again exhibited a strong phylogenetic pattern (Table 2B; Figure 6, center-panels).
Finally, we investigated whether relationships among PC axis and TL variables might differ among avian orders. We found that the association of Adult TL and the slow-fast pace of life (PC2) was significantly positive in the Passeriiformes (r = 0.690, t = 3.321, p = 0.006, n = 14). However, in the Procellariiformes the association was negative (r = -0.754, t = -2.812, p = 0.031, n = 8), i.e.opposite to the general non-significant inter-specific trend (Figure 3B). TROC and PC1 and PC2 axes were significantly and negatively correlated in the Procellariiformes (Figure 4; PC1: r = -0.951, t = -6.872, p < 0.001; PC2: r = -0.890, t = -4.368, p = 0.007; all n = 7). No significant correlations were found within orders for PC axes and Chick TL (Figure 5, all p > 0.25).