Individual life histories: neither slow nor fast, just diverse

1. Introduction

Species display an astonishing diversity of life-history traits [1], which are shaped by species differences in their body size or mass, Bauplan and lifestyle [2]. Organisms face time and energy allocation trade-offs because time and resources are limited in nature. Across species, the trade-off between survival and reproduction determines the pace of the life cycle [3], the so-called slow–fast continuum [4]. Along this slow–fast continuum, species rank from a combination of late reproduction, low reproductive rates, and long lifespans at the slow end to a combination of opposite characteristics at the fast end [5]. The slow–fast continuum has been consistently identified as the driving axis of life-history variation across species in most taxa studied so far, including birds, mammals, insects and plants [1,6–10].

Within a given species, individuals also show a high diversity of life-history traits [11–14]. Two decades ago, the idea that a slow–fast continuum of life-histories could also exist among populations within a given species emerged from the observation that tropical songbird populations had slower life histories than temperate ones [15,16]. Different populations experience different environmental conditions, which can affect the expression of life-history traits. A demonstration of this comes from Reznick et al.'s [17] experimental study showing a shift in population-level life histories after the introduction of predators in nature. More recently, yellow-bellied toad [18] and common lizard [19] populations affected by increased anthropization and climate warming, respectively, have been observed accelerating their life cycles through compensatory investment in reproduction. At the population level, identifying patterns of life-history variation has generated much research, but what structures such variation remains unclear. A common finding is that life-history trade-offs are not always detected [20]. Individual differences in resource acquisition can generate positive correlations between life-history traits, hence masking life-history trade-offs. Whether the slow–fast continuum, defined by a series of trade-offs between life-history traits in time units [5], structures individual variation as it does across species is therefore even less clear. The existence of a slow–fast continuum of life histories among individuals has nevertheless been repeatedly assumed in behavioural ecology in regard to the concept of pace-of-life syndrome (POLS) [21,22]. Specifically, the POLS hypothesis posits that a series of behavioural and physiological axes of variation correlates with the individual slow–fast continuum [21]. A formal test of whether the slow–fast continuum drives individual life-history variation within populations is thus needed to move forward in our understanding of how individual heterogeneity in life history is structured within populations.

Here, we empirically assessed whether a slow–fast continuum structures individual life-history variation. Our approach is based on the assessment of a slow–fast continuum both across species and among individuals within populations, using the same set of species and focal life-history traits. We relied on 17 of the world's most detailed, long-term and individual-based population monitoring of bird and mammal species (figure 1) with contrasting life-history strategies (electronic supplementary material table S1). First, using principal component analyses (PCA [23]) and correcting for phylogenetic relatedness, we identified the structuring axes of variation among four life-history traits calculated across individuals for each species. We used age at first reproduction, adult lifespan, breeding frequency and fecundity, which are life-history traits commonly used to describe the slow–fast continuum of life histories in vertebrates [1,5,9,24,25]. We deemed a slow–fast continuum as present if the first component of the PCA was selected and if both traits with a time duration dimension (i.e. age at first reproduction and adult lifespan) covaried negatively with traits with a time frequency dimension (i.e. breeding frequency and fecundity). To determine whether the first component (PC1) of the PCA was selected, we followed the broken-stick model [26], which sets the threshold above which a component explains more variation than what would be expected by chance alone. Then, we tested whether life-history variation across species and among individuals within populations aligned [27] by comparing variation in life-history traits at both levels of biological organization for all species simultaneously. Finally, we used the same procedure for each species separately to assess whether the slow–fast continuum consistently structured life-history variation within populations.

• Download figure
• Open in new tab
• Download PowerPoint

2. Methods

3. Results

(a) Retrieving the slow–fast continuum of life histories across species

According to the broken-stick model (electronic supplementary material, figure S1), the first principal component of the PCA correcting for phylogenetic relatedness among species (pPC1) was retained as the main structuring axis of variation in life histories across species. The pPC1 accounted for 76–84% (range across 100 phylogenetic trees; table 1) of the overall variation. Using an average phylogenetic tree, pPC1 accounted for 77% of variation. No other principal component was retained as a secondary structuring axis (electronic supplementary material, table S6). As expected, pPC1 fully aligned with the slow–fast continuum by opposing traits expressed in time units to traits expressed in time frequencies (figure 2a). The contributions of all traits to pPC1 were high and close to equi-correlation (table 1; electronic supplementary material, table S6), which indicates that all traits were equally important in defining the slow–fast continuum. Not surprisingly, a species' position on pPC1 (i.e. on the slow–fast continuum) was closely linked with its generation time (R² = 0.78; electronic supplementary material, table S8), which supports the hypothesis that generation time reliably predicts species' position along the slow–fast continuum [34]. When using a simple PCA without accounting for phylogenetic relatedness, the same pattern held among species, with even stronger support for the slow–fast continuum (electronic supplementary material, tables S3–S5 and figure S2).

• Download figure
• Open in new tab
• Download PowerPoint

Table 1. Principal component analyses (PCA) across species and among individuals within populations. The proportion of variation explained, the relative importance of each life-history trait and their loadings on PC1 are shown. The relative importance for PC1 (or pPC1) for each trait was calculated as the absolute value of the loading of PC1 (or pPC1) divided by the total loading across principal components. Because of the random direction of axis rotation in a PCA and for ease of comparison between species, the orientation (sign) of the loadings of life-history traits were reversed to keep ‘adult lifespan' on the negative side of PC1 (or pPC1), when necessary. In golden-mantled ground squirrel and blue tit, breeding frequency did not vary among individuals. Similarly, in the great tit, both breeding frequency and age at first reproduction showed no individual variation. For those three species, their respective PCAs were reduced to either two or three dimensions.

Collapse

¹Breeding frequency did not vary across individuals in golden-mantled ground squirrel and blue tit, leading to only three potential components instead of four.

²Breeding frequency and age at first reproduction did not vary across individuals in great tit, leading to only two potential components instead of four.

³Breeding frequency and fecundity are confounded in the Weddell seal, which artificially inflates the proportion of variation accounted for by PC1.

(c) Species-specific axes of variation in individual life histories

Focusing on individual variation in life histories within populations for each species independently, we tested for the presence of a slow–fast continuum in each species using species-specific PCAs (figure 3). In the great tit, there was no variation in age at first reproduction and breeding frequency among individuals. In the blue tit and the golden-mantled ground squirrel, there was also no variation in breeding frequency. For these species, the slow–fast continuum cannot be detected, hence comparison with other species was not possible. Also, breeding frequency and fecundity in the Weddell seal were confounded because of the absence of information on weaning success (electronic supplementary material, S1), which also hindered comparison with other species. Results for these four species are presented for consistency but should be interpreted with caution.

• Download figure
• Open in new tab
• Download PowerPoint

For the 13 remaining species, the proportion of variance explained by the main axis, PC1, was consistently much lower than that retrieved across species (figure 3). Indeed, PC1 accounted for between 29% of life-history variation among individuals in the Eurasian oystercatcher and 52% in wandering albatross (table 1). Interestingly, there was a positive relationship between a species' generation time and the proportion of individual variation explained by PC1 (β = 0.036, s.e. = 0.016, p = 0.048). However, following the broken-stick model (electronic supplementary material, figure S1), the only species that was close to showing a dominantly structuring axis of variation was the wandering albatross (52% of variation explained by PC1). For this species, the pattern of covariation between life-history traits appeared qualitatively in line with the slow–fast continuum (figure 4). However, contrary to the equal contribution of all time and time frequency traits expected from a slow–fast continuum, age at first reproduction had a rather independent role in explaining individual variation in this species. Indeed, age at first reproduction contributed much less to PC1 than other life-history traits (table 1) and mostly contributed to PC2 (electronic supplementary material, table S6).

• Download figure
• Open in new tab
• Download PowerPoint

For all other species, the proportion of life-history variation explained by PC1 was (often far) below the 52% threshold necessary for it to be interpreted as a structuring axis based on the broken-stick model (table 1). Individual variation could thus not be distinguished from a random process and therefore did not seem to be structured by any of the life-history traits considered. When looking at species-specific PCAs (figure 4), we nevertheless noted that across species, adult lifespan and breeding frequency covaried negatively, which might suggest a trade-off between those two life-history traits. Indeed, the covariance between adult lifespan and breeding frequency was negative and strong (less than or equal to −0.5) in six species (electronic supplementary material, table S10). Overall, age at first reproduction was the life-history trait that covaried the least with other life-history traits, suggesting that individual variation in this trait is independent from the other traits we analysed. Although species-specific PC1s could not be interpreted as structuring axes of individual life-history variation, we assessed the relationship between PC1 scores and individual-level generation time in each species to provide a direct comparison with the pattern of life-history variation retrieved across species. The strength of species-specific relationships was much weaker (average R² = 0.32, with R² values ranging from 0.003 in southern fulmar to 0.69 in yellow-bellied marmot) than that observed across species (0.78 and 0.92 with and without phylogeny included, electronic supplementary material, table S8). Detailed PCAs and correlation plots for each species can be found in electronic supplementary material, figures S3–S6. Our analyses thus provide clear evidence that the slow–fast continuum is not a structuring axis of life-history variation among individuals within a population.

4. Discussion

Based on high-quality and long-term demographic data from a substantial number of bird and mammal species with contrasting life-history strategies, we assessed whether the slow–fast continuum of life histories is the main factor structuring individual life-history variation within a given population. Using key demographic traits that shape the speed of a life cycle, namely age at first reproduction, adult lifespan, fecundity and breeding frequency, we successfully retrieved the expected slow–fast continuum across species. The percentage of variance in life-history traits explained by the slow–fast continuum we provided is in line with previous studies (70–80% in vertebrates [5,10,24]). Moreover, as expected from previous work, all time and time frequency traits equally contributed to the slow–fast continuum and species-specific scores on the slow–fast continuum were closely associated with species' generation times. However, individual variation in those same life-history traits within a given population did not align with the slow–fast continuum. The wandering albatross was the only species for which PC1 almost supported expectations from the slow–fast continuum. The apparent breeding frequency–adult lifespan trade-off in the wandering albatross could be explained by their extreme long life and their quasi-biennial breeding tactic. Indeed, most successful breeders, but not failed breeders, skip breeding the following year to replenish body reserves [37], which may exacerbate the trade-off between breeding and survival processes and increase individual variation. Nevertheless, for all species including the wandering albatross, the broken-stick model indicated that no structure could be statistically identified beyond a random variation. As a result, our study showed that individual life histories within a population are not structured by a slow–fast continuum.

Our approach was based on comparing life-history variation across species and among individuals within populations using the same set of species, life-history traits and methodological tools. The approach we used has been previously suggested [27,29], but not yet applied, potentially because of the challenges of monitoring a large number of individuals over their entire lifetime within a population in the wild. In our collaborative effort, we could access the required individual life-history records for 17 bird and mammal species. With this set of species, we first confirmed the existence of the slow–fast continuum as the major structuring axis of life-history variation across species before moving on to the individual level. Based on the same data, our approach would have allowed for the detection of the slow–fast continuum of life histories among individuals within a population, if present in any of the species considered. Many studies have previously acknowledged that the challenges associated with the detection of a slow–fast continuum within populations are great and numerous [24,29]. Below we discuss the generalization of a slow–fast continuum of life histories across levels of biological organization, the various factors that can mask its expression within a population and the implications of our findings for the POLS hypothesis.

We relied on individual data to reconstruct the slow–fast continuum across species. By taking the average of individual life-history traits for each species, we assumed the data to be normally distributed, but this assumption was not always met (electronic supplementary material, S9; figures S7 and S8). We confirmed the robustness of our results by repeating the analysis using the median as well as the first and third quartiles of the data distribution (electronic supplementary material, figure S9). We nevertheless acknowledge that there may be uncertainties in the individual data due to, e.g. imperfect detection, variation in recapture rates and selective disappearance, and that the importance of such uncertainties is likely to differ among species.

A great challenge in transposing the slow–fast continuum from the across-species to the among-individual levels is that the overall variation in all life-history traits is much larger across species than among individuals. Indeed, the coefficient of variation in the four life-history traits we analysed was consistently larger across species compared to among individuals within populations, especially for adult lifespan, fecundity and age at first reproduction (electronic supplementary material, table S11). As hypothesized by Stearns in 1983 [4], as we descend along the taxonomic scale, the partitioning of the variation should change between taxa. The covariation between life-history traits would persist, but among fewer and fewer traits as some traits become fixed. Based on the set of species we analysed, our results also support this claim because we found an absence of individual variation in one or two life-history traits in three species (great tit, blue tit, and golden-mantled ground squirrel). Because the potential for variation in each life-history trait likely depends on the species, not all life-history traits may be equally relevant for the detection of individual patterns of variation across species.

Generalizing the concept of a slow–fast continuum of life histories across levels of biological organization (i.e. species, populations, and individuals) may be challenging, potentially because evolutionary and environmental constraints (e.g. body size, habitat quality) differ at each level and among species [15]. Our analyses revealed that the main axis of life-history variation among individuals within populations might differ depending on the species' position on the slow–fast continuum. The magnitude of individual variation within a population is likely not equal for all life-history traits and should differ between slow and fast species. For example, in slow species such as humans (with a generation time of ca 30 years), the potential to observe differences in lifespan between individuals is greater than in fast species such as white-throated dipper (with a generation time of less than 2 years). When focusing on the effect of individual heterogeneity alone, however, the opposite pattern is found: the variance is lower in lifespan for slow compared with fast species [38,39]. Therefore, finding a pattern of life-history variation that would fit all species is unlikely. General patterns may be easier to detect across species of similar generation times and their detection may require redefining, for a restricted range of generation times, the life-history traits that would capture the most individual variation.

There is still a possibility that a slow–fast continuum of life-histories exists among individuals within a population but is difficult to detect. Indeed, despite studies showing genetic correlations between life-history traits [40,41], a plethora of factors influence the phenotypic expression of genotypes in natural populations [15], which may prevent the detection of a slow–fast continuum. The slow–fast continuum consistently retrieved across species is generally interpreted as being brought about by species-specific differences in resource allocation to biological functions [42]. Evolutionary constraints in body plan, size, lifestyle and habitat requirements are much weaker among individuals within a population than across species. As a result, individuals within a population only differ slightly in energy allocation patterns but differ strongly in resource acquisition. Since van Noordwijk & de Jong's pioneering work [20], we know that higher variance in acquisition than allocation generates positive covariation among performance traits (see e.g. [13,43–45] in some of the species included in our analyses). These individual differences in resource acquisition explain why theoretical predictions seldom match empirical evidence in the POLS literature [29,46]. This widespread situation indicates that individual heterogeneity in resource acquisition confounds trait variation within a population. In addition, density dependence, which often occurs in vertebrate populations monitored for a long period, can modulate selective pressures on the speed of life at the population level [11,47], leading individuals to be faster at lower than higher population densities. Individual differences in resource acquisition and density-dependence are two processes illustrating the importance of considering the ecological and demographic context under which a population occurs when attempting to retrieve life-history trade-offs and the slow–fast continuum [27,48].

Two other important factors that might explain why there is no cross-level consistency, despite marked individual variation in the considered traits, are individual stochasticity and phenotypic plasticity. First, individual stochasticity corresponds to differences in life courses that are generated by the occurrence of random events in the life cycle of an individual (e.g. surviving or not, reproducing or not, etc.), and is thus the result of chance alone [49]. For instance, an individual allocating a lot of resources to reproduction may still live longer than an individual prioritizing allocation to maintenance because mortality may be due to random events whose occurrence is independent from the individual allocation strategies. Individual stochasticity is known to increase among-individual variation in life-history traits and to weaken correlation among traits [29]. Thus, stochasticity is likely an important source of variation in life-history traits generating unstructured individual variation in populations in the wild. Second, at the individual level, plasticity in the expression of genotypes may allow individuals to adjust their phenotypes according to environmental conditions. However, there is no reason to expect plastic responses to necessarily match evolutionary responses at the levels of populations or species and mismatches are likely to occur [24]. A careful examination of what shapes life-history variation at each level separately, from individuals to populations to species is thus needed.

Whether the lack of structure in individual life-history traits we found here corresponds to random individual variation or is the result of the coexistence of different processes or types of trait covariation cannot be teased apart in our analyses. Using a different analysis, Jenouvrier et al. [14] found that in addition to individual stochasticity, complex life-history patterns could be detected in the southern fulmar. Indeed, individuals were found to exhibit patterns of trait covariation in line with both the slow–fast continuum and a continuum of individual performance [50]. Specifically, a first group of individuals had late, but successful, recruitment and extended reproductive lifespan, a second group were less likely to recruit, had higher adult survival, and skipped breeding often, and a third group recruited early, attempted to breed often, and had a short lifespan. Individuals in the first and third groups performed better than individuals in the second group as they produced, on average, more offspring over their lifetime. The interplay between these two structuring axes of variation (slow–fast continuum and performance continuum; sensu Nussey et al. [51]) may interfere and prevent the detection of a unique main structuring axis of variation within populations.

The POLS hypothesis was first formulated to suggest covariations between physiological and life-history trade-offs among species and populations [15], and several studies have repeatedly supported its predictions (e.g. [16,52,53]). The POLS was later extended to incorporate expected covariations between behavioural and life-history traits, relying on the assumption that slow and fast phenotypes co-occur in a single population, opening a new branch of research on life-history variation within populations. However, this integrated POLS has so far received only limited empirical support [54], especially for vertebrates [55]. Indeed, out of 141 studies on vertebrates, support of the POLS was mixed [55]. This may be because the basic assumption of the POLS, i.e. the existence of a slow–fast continuum, does not hold. Considering that an individual slow–fast continuum should arise from a series of individual life-history trade-offs and that the latter have been proven difficult to detect in many taxonomic groups [48], this is not totally surprising. In our study, we go beyond single life-history trade-offs and specifically tested for the underlying assumption of the POLS, i.e. the existence of a slow–fast continuum. We showed little to no evidence of an individual slow–fast continuum within 17 populations of 17 different bird and mammal species. Our results show decisive evidence that the slow–fast continuum may not be the most appropriate framework on which to graft individual behavioural and physiological suites of traits, as the POLS hypothesis originally formulated [21]. It may remain possible to detect a generalizable slow–fast continuum among individuals in the wild in other taxonomic groups or once confounding factors such as demographic stochasticity, density dependence and individual resource acquisition will have been adequately accounted for. However, this task is far from easy because the relative importance of these confounding factors will vary across species and contexts [29].

In conclusion, our study revealed that the slow–fast continuum does not structure variation in individual life histories within bird and mammal populations as it does across species. Different species have different life cycles [1], but those differences may get blurred as we descend the taxonomic scale. For example, evidence of populations from the same species living in very contrasted environmental conditions showing both similar to as well as dissimilar life cycle speeds can be found [56,57]. Within populations, the variation among individuals appears much more diversified, involving multiple causes. Individual differences in resource acquisition and demographic stochasticity are likely to be dominant causes, but other biological processes may also come into play. In spite of this, individual heterogeneity in life histories is often detected and its impact on demography can be large (e.g. [14]). What governs individual differences in life-history traits is likely idiosyncratic across species, with the consequence that identifying a ‘one size fits all’ pattern of variation is unlikely. Quantifying and comparing the relative roles of chance and resource acquisition across species would help move forward in our understanding of what shapes individual variation in life histories within populations in the wild.

Ethics

Animal captures and handling were approved by or were conducted in compliance with the Université de Sherbrooke ethical committee (Protocol MFB01), the Canadian Council for Animal Care committee of the Université Laval, the institutional animal care and use committees of the University of California at Davis, the Rocky Mountain Biological Laboratory, the Ethic Committee of Institut Polaire Francais (IPEV), the Préfet of Terres australes et antarctiques francaises (TAAF) after advice from the Comité de l'Environnement Polaire (CEP), laws and regulations for animals used in research in Norway (FOR-2003-03-18-349, LOV-2009-06-19-100, LOV-1981-05-29- 38§26), the Norwegian Bird Ringing Centre on behalf of the Norwegian Environment Agency, the Director of Food, Agriculture and Forest (Prefectoral order 2009-14 from Paris), the Office National des Forêts (ONF) through Partnership Convention ONCFS-ONF, guidelines and regulations of the Ethical Committee of Lyon 1 University (project DR2014-09, 5 June 2014), the National Marine Fisheries Service permits for work on marine mammals, Antarctic Conservation Act permits, the Institutional Animal Care and Use Committee Protocols, the New Zealand Department of Conservation, the New Zealand national bird banding scheme (permit to band birds no. 2008/077), the ethics committee for animal experimentation of Languedoc Roussillon (305-CEEA-LR-12066, APAFIS#8608-2017012011062214), the Regional Institutions (bylaws issued by the Prefecture of Haute-Corse on 17 April 2018 no 2B-2018-04-17-001 and on 7 April 2022 no 2B-2022-04-07-00002), personal ringing permits delivered by the CRBPO (Centre de Recherches sur la Biologie des Populations d'Oiseaux, e.g. ringing permit number 1907 for A.C.) for CRBPO ringing programme 369, the Norwegian Animal Research Authority, the Ringing Centre at Stavanger Museum, Norway, the animal ethics license AVD250002015200, the Law of Animal Experiments (WoD Nederland), the IUCUC protocols approved by UCLA (2001-191), the permits approved by the Colorado Department of Parks and Wildlife (TR917), the license AVD801002017831 of the Centrale Commissie Dierexperimenten (CCD) in The Netherlands.

Data accessibility

Data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.3bk3j9kpm [58].

Additional information is provided in the electronic supplementary material [59].

Authors' contributions

J.V.d.W.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, visualization, writing—original draft, writing—review and editing; R.F.: conceptualization, data curation, methodology, resources, writing—review and editing; J.-M.G., F.P., S.H. and M.G.: conceptualization, data curation, funding acquisition, methodology, resources, writing—review and editing; C.B., D.T.B., A.C., J.A.M., E.M., F.M.M., J.R., B.-E.S., M.v.d.P., D.H.V.V. and M.E.V.: data curation, funding acquisition, writing—review and editing; F.G.B.: methodology, validation, visualization, writing—review and editing; K.D., J.M., C.T. and C.P.W.: data curation, writing—review and editing; B.L.: data curation; J.Y.: data curation; S.J.: conceptualization, data curation, funding acquisition, methodology, resources, supervision, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

J.V.d.W. was funded by Fonds de Recherche du Québec – Nature et Technologies and the National Science Foundation, NSF (GEO-NERC 1951500). S.J. was funded by the NSF (OPP 1840058, and GEO-NERC 1951500). F.P. was funded by the Natural Sciences and Engineering Research Council of Canada, NSERC (Discovery 2018-05405 and E.W.R Steacie fellowship no. 549146-2020). S.H. and data collection on mountain goats were mostly supported by NSERC and the Alberta Conservation Association, ACA. F.G.B. was funded by NSERC (Discovery 2021-03943). D.T.B. was supported by the National Geographic Society, UCLA (Faculty Senate and the Division of Life Sciences), a Rocky Mountain Biological Laboratory research fellowship, and by the NSF (I.D.B.R.-0754247, and D.E.B.-1119660 and 1557130 to D.T.B., as well as D.B.I. 0242960, 0731346, and 1226713 to the R.M.B.L.). J.A.M. received a Marsden Grant from the Royal Society of New Zealand 1996–1999. J.R. and data collection for Weddell seals were supported by the NSF, Division of Polar Programs (ANT 1640481 and 2147553) and prior NSF grants to R.A. Garrott, J.R., D. B. Siniff and J.W. Testa. J.-M.G. was supported by the ANR program ‘DivInT' and data collection for roe deer was supported by the Office Français de la Biodiversité. The long-term demographic studies at Crozet (wandering albatross), Kerguelen (black-browed albatross) and Pointe Géologie (southern fulmar and snow petrel) were supported by the French Polar Institute IPF (project 109 ‘Seabirds and marine mammals as sentinels of global changes in the Southern Ocean', PI C.B.) and by Zone Atelier Antarctique et Terres Australes (LTSER France). This work was also supported by the Research Council of Norway through its Centres of Excellence funding scheme, project number 223257. Extraction and formatting of human life-history data from the île aux Coudres Population Register were supported by NSERC (Discovery 2015-03683) to E.M., Université de Montréal, the Fonds pour la Formation de Chercheurs et l'Aide à la Recherche du Québec, and the Social Sciences and Humanities Research Council of Canada.