Differences in the temporal scale of reproductive investment across the slow-fast continuum in a passerine

Abstract

Life-history strategies differ with respect to investment in current versus 'future' reproduction, but when is this future? Under the novel 'temporality in reproductive investment hypothesis', we postulate variation should exist in the time frame over which reproductive costs are paid. Slow-paced individuals should pay reproductive costs over short (e.g. inter-annual) time scales to prevent reproductive costs accumulating, whereas fast-paced individuals should allow costs to accumulate (i.e. senescence). Using Fourier transforms, we quantify adjustments in clutch size with age, across four populations of blue tits (Cyanistes caeruleus). Fast populations had more prevalent and stronger long-term changes in reproductive investment, whereas slower populations had more prevalent short-term adjustments. Inter-annual environmental variation partly accounted for short-, but not long-term changes in reproductive investment. Our study reveals individuals differ in when they pay the cost of reproduction and that failure to partition this variation across different temporal scales and environments could underestimate reproductive trade-offs.

Keywords: blue tits; carry-over effects; life-history; pace-of-life; senescence.

FIGURE 1

A schematic illustrating two simulated reproductive investment curves (RICs) occurring over different temporal scales. An RIC can vary in the temporal scale of changes in investment and in the magnitude of these changes (amplitude). RIC 1 (Panel a, c, e): An example where clutch size increases gradually over time until mid‐life when it begins to decline and clutch size ranges from 6–10. RIC 2 (Panel b, d, f): An example where clutch size oscillates inter‐annually and ranges from 7 to 9. RIC 1 shows a long‐term RIC with a high amplitude, whereas RIC 2 shows a short‐term RIC with a lower amplitude. Here, we compare two widely used methods for analysing reproductive investment with a novel Fourier mode analysis. (a) A Fourier mode of wavelength 14 explains 99% of the variation in clutch size (note only part of this wave is shown in the figure), whereas (c) a quadratic (senescence) model explains 97% and (e) a state‐dependent model with a one‐year lag explains 17% of the variation. (b) A Fourier mode with a wavelength of two explains 83% of the variation, whereas (d) a quadratic model explains 17% and (f) a state‐dependent model explains 70%. This figure shows a simplified example, which is likely not realistic in the wild. First, individuals may not have the same lifespan, and hence time series length, and this may covary with the temporal scale of RICs. Second, multiple RICs over different temporal scales (i.e. multiple Fourier modes) can, and are likely to, occur within an individual

FIGURE 2

Reproductive investment strategies (RICs) among four blue tit populations, living in different habitats (D = deciduous oak; E = evergreen oak). Proportion of birds with a dominant (i.e. the wavelength with the highest amplitude) long‐term reproductive investment curve (wavelength of >2 years). (a) Using RICs calculated on absolute clutch size, E‐Pirio had the smallest proportion of birds with long‐term RICs and D‐Rouviere the largest. (b) Using RICs calculated on mean‐centred clutch size, E‐Pirio remained the population with the highest proportion of short‐term RICs, but D‐Muro was observed to have the lowest proportion of individuals with short‐term RICs, largely because the proportion of long‐term RICs in D‐Rouviere decreased. Plots shows estimated marginal means from a model including all fixed effects maintained in the best fitting models and associated ±1 standard errors on the original scale

FIGURE 3

The wavelength amplitude associated with reproductive investment curves (RICs) among four blue tit populations, living in different habitats (D = deciduous oak; E = evergreen oak). (a) Results using absolute clutch size (b) Results using mean‐centred clutch size. Red =short‐term RICs; blue =long‐term RICs. Using absolute clutch size (a), deciduous populations had short‐term RICs with greater amplitude, but these differences between populations did not persist using mean‐centred clutch size (b) suggesting inter‐annual variation in environmental conditions is an important component of short‐term RICs. The amplitude of long‐term RICs was higher for birds in deciduous compared to evergreen woodlands using absolute clutch size (a), and these results were very similar when using mean‐centred clutch size (b), suggesting little effect of inter‐annual variation in environmental conditions among populations. Plots shows estimated marginal means from a model including all fixed effects maintained in the best‐fitting models and associated ±1 standard errors on the original scale. When population was maintained in the best‐fitting model solid lines are shown, and when it was dropped these effects are shown with dashed lines