SEARCH

SEARCH BY CITATION

Keywords:

  • cost of reproduction;
  • life history;
  • meta-regression;
  • parental effects;
  • parent–offspring conflict

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

A fundamental premise of life-history theory is that organisms that increase current reproductive investment suffer increased mortality. Possibly the most studied life-history phenotypic relationship is the trade-off between parental effort and survival. However, evidence supporting this trade-off is equivocal. Here, we conducted a meta-analysis to test the generality of this tenet. Using experimental studies that manipulated parental effort in birds, we show that (i) the effect of parental effort on survival was similar across species regardless of phylogeny; (ii) individuals that experienced reduced parental effort had similar survival probabilities than control individuals, regardless of sex; and (iii) males that experienced increased parental effort were less likely to survive than control males, whereas females that experienced increased effort were just as likely to survive as control females. Our results suggest that the trade-off between parental effort and survival is more complex than previously assumed. Finally, our study provides recommendations of unexplored avenues of future research into life-history trade-offs.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Fitness trade-offs have a central role in evolutionary thinking because they provide a framework with which to explain the diversity of life histories observed at many taxonomic levels (Williams, 1966; Stearns, 1992; Reznick et al., 2000; Roff & Fairbairn, 2007). Life-history theory is based on the notion that individuals have a limited amount of resources available to allocate to competing functions or structures, thus resulting in trade-offs (van Noordwijk & de Jong, 1986; Martin, 1987; Lindén & Møller, 1989; Stearns, 1992; Roff, 2002). Therefore, an individual's fitness is determined by the outcome of greater resource allocation to some traits and a concomitant decrease in resource allocation to other traits. Possibly, the most studied life-history relationship is the trade-off between current and future reproduction, also known as the cost of reproduction (Williams, 1966; Martin, 1987; Lindén & Møller, 1989; Stearns, 1992; Roff, 2002). The cost of reproduction has commonly been investigated in birds, by studying correlations between life-history traits and also by experimentally manipulating traits related to current reproduction (such as brood size or clutch size manipulations that affect levels of parental effort) with an intent to induce changes in survival or future reproduction (see Bryant, 1979 and Reid, 1987 for an example of an observational and an experimental study, respectively).

The trade-off between parental effort and survival remains generally controversial because some observational studies have failed to find the predicted relationship and others have even found positive relationships, which is contrary to what is expected for phenotypic trade-offs (see Reznick et al., 2000; Roff & Fairbairn, 2007). Some of these unexpected trade-off relationships could result from, for example, environmental fluctuations in resource availability, or individual intrinsic quality and also individual age (Reznick et al., 2000; Pettifor et al., 2001; Creighton et al., 2009). van Noordwijk & de Jong, 1986 proposed a theoretical solution for the problem of unexpected trade-off relationships, suggesting that positive correlations between life-history traits can result from some individuals allocating greater amounts of resources to several life-history traits (i.e. high quality individual) and others allocating little (i.e. low quality individual) (but see Wilson & Nussey, 2010 for a recent discussion of individual quality). Other unexpected results could result from not accounting for the sex of the individual providing parental care, because females and males have different parental strategies. It is commonly accepted that females, in a wide variety of taxonomic groups, provide more care, whereas males invest further in obtaining additional mating opportunities (Queller, 1997). In addition to the controversy in observational studies, even some experimental research designed to control confounding effects of the relationship between parental effort and survival has failed to demonstrate the predicted trade-off (e.g. Korpimäki, 1988; Orell et al., 1996; Doligez et al., 2002). For these inconsistent results, Graves (1991) pointed out a simple explanation that statistical power is often too low to detect the effects of parental effort manipulations on survival in most studies.

Here, we used a powerful statistical approach, a meta-analysis, to synthesize studies that investigated the trade-off between parental effort and adult survival in birds to resolve the controversies regarding this trade-off. We tested one main prediction of life-history theory. That is, individuals that experience reduced parental effort have higher survival rates than individuals that experience natural levels of effort, and also, individuals that experience increased parental effort have lower survival rates than individuals that experience natural levels of effort. We used only experimental studies (i.e. altering clutch size or brood size) to avoid the confounding effects of individual quality and individual optimization on the trade-off between parental effort and survival (see 'Inclusion criteria' for details). We also restricted our analyses to those experimental studies that estimated the effect of experimental manipulation on the survival of each sex separately, as it is likely that the effects of parental effort are sex-specific. In addition, we examined the heterogeneity of this trade-off across species (i.e. phylogenetic signal) using a phylogenetically controlled meta-analysis. Interestingly, we found a sex-specific effect of parental effort on survival; males that experienced increased parental effort were less likely to survive, whereas females were little affected by change in parental effort. These results are an important addition to our understanding of life-history trade-offs.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Data collection

We performed an extensive literature search using the online database Scopus (subject area: Life Sciences) using all available years up to and including April 2012 (when the search was last updated). We used the following terms and their combinations in our search: ‘cost*’, ‘trade-of*’, ‘effort’, ‘mortality’, ‘survival’, ‘fitness’, ‘reproduction’, ‘life-history’. Our search yielded a large number of references (3040 papers). Thus, to be more efficient in our retrieval, we sorted the search results by publication date (i.e. having the most recently published study on the top of the list). We then proceeded to find the most recent study that fitted our criteria, which was Erikstad et al. (2009). Using Erikstad et al. (2009), we performed a backward search of all the studies that were in their reference list that could potentially be included in our analysis. Furthermore, we applied the backward search procedure to the cited references of all relevant publications that we encountered. This recursive procedure was used until we stopped finding new studies. We also checked for papers that cited Erikstad et al. (2009). In total, we found 59 articles that were appropriate for our meta-analysis. After applying this procedure, we returned to the online search to look for any studies that we could have missed. We inspected the titles and abstracts of all references from our search, but did not find any new studies, thus confirming the efficiency of our procedure.

Inclusion criteria

We selected studies according to four specific criteria. To be included in our analysis, the study must have (i) been on wild bird populations (as was mentioned previously, we chose birds because of the amount of experimental work that has been conducted with this taxon); (ii) examined the relationship between parental effort and parental survival by experimentally increasing or decreasing effort (e.g. manipulating clutch or brood size by adding or removing eggs/nestlings); (iii) fully reported the number of individuals per experimental group that returned during subsequent breeding seasons, or model estimated survival rates for the groups, from which we could recover the sample sizes; and (iv) investigated the effect of parental effort on survival for each sex (male/female) separately. Studies that did not fulfil our inclusion criteria are listed, along with the reason for exclusion, in the Supporting information (Table S2). We used graphclick (http://www.arizona-software.ch/graphclick/) to retrieve data reported in figures (survival rates or number of individuals in each experimental group). After applying our inclusion criteria, 29 studies performed on 19 species and yielding 125 effect size estimates remained in the data set (Table S3). The species included in our analyses were distributed among 13 families representing six orders: Anseriformes (n = 1 species), Charadriiformes (n = 2), Strigiformes (n = 1), Falconiformes (n = 2), Piciformes (n = 1) and Passeriformes (n = 12).

Data coding and calculation of effect sizes

We coded the sex of the individuals (either male or female) for each effect size in our data set because we expected sex-specific effects in the cost of reproduction as stated above. We then included the sex variable as a fixed effect (effect size moderator) in all our meta-analytic models. We also included the type of parental effort manipulation (clutch size or brood size) as a fixed effect in a set of models (results shown in Table S4 in Supporting information, as there was no evidence of a distinct effect of the different types of manipulation). We did not include interactions between sex and type of manipulation in our models because we did not have enough sample size to evaluate these relationships. We chose to use the odds ratio (OR) as the measure of effect size, that is, as the estimate of the magnitude of the impact of parental effort on adult survival. Our OR effect sizes express the ratio between the survival of birds in the treatment group (the numerator) and that of birds in the control group (the denominator). When OR equals 1, there is no difference in survival between the two groups, OR > 1 indicates an increase in survival probability of treatment birds vs. control, and OR < 1 a decrease (see Fleiss & Berlin, 2009 for more information). We conducted all of the analyses on the logarithms of OR (lnOR) along with its measurement error variance. We used standard equations to calculate OR (or lnOR) and associated variances for lnOR values from the number of dead and alive birds in each group (control, reduced/increased parental effort) (Nakagawa & Cuthill, 2007; Fleiss & Berlin, 2009).

Meta-analytic procedures

First, we tested the assumption that the experimental manipulations effectively altered parental effort (see Supporting information for further details). Data available from the studies in our data set support this assumption. There was a significantly positive association between parental effort of both sexes and manipulation of clutch/brood size, which indicates that parents of reduced clutches/broods provided less care than parents of increased clutches/broods. The posterior mean estimate of this association could be considered to be of moderate magnitude (r = 0.1, 0.3 and 0.5 are considered to be small, medium and large effects according to Cohen (1988); mean meta-analytic correlation between experimental manipulation and parental effort: r = 0.345; 95% CI: 0.138 to 0.523; n = 11). We then used a model to estimate the effect size of increased parental effort vs. control and a different model to estimate the effect size of reduced parental effort vs. control. To estimate meta-analytic means, we used Bayesian phylogenetic mixed-effects meta-analysis (BPMM; see Hadfield & Nakagawa, 2010 for details), implemented in the mcmcglmm package (Hadfield, 2010) for r (version 2.13.0; R Development Core Team, 2011). The squared standard error of OR was used as the weighting value in the BPMM models, such that more reliable estimates (i.e. lower variance, either from larger sample sizes or more consistent results) contributed more to the model. We used a mixed-effects approach to deal with nonindependence among effect sizes that originated from the same species, and also to account for the potential lack of independence due to evolutionary history among the species (see Nakagawa & Santos, 2012). Hence, for all BPMMs, we included the species identity and phylogeny as random effects. Phylogeny is included in the form of a correlation matrix of distances from the root of the tree to the most recent common ancestor between two species. In a preliminary analysis, we also included the study identity as a random effect, but removed it from the final model because the deviance information criteria (DIC; Lunn et al., 2000) indicated no improvement in model fit (reduced effort model: ΔDIC = 0.3; increased effort model: ΔDIC = 0.8; see Supporting information for mcmcglmm parameter details). We used a phylogenetic tree that was adapted from trees available at bird supertree Project (http://linnaeus.zoology.gla.ac.uk/~rpage/birdsupertree/) (Supporting information; Fig. S1). As model results, we present posterior means with 95% credible intervals (CIs) for point estimates of fixed effects and posterior modes for point estimates of random effects throughout this paper. For inference purposes, all model parameters were back-transformed to the original metric (OR). We deem parameters with 95% CIs not spanning zero to be statistically significant (in case of OR, not spanning 1).

Heterogeneity and publication bias

Following recent advances in meta-analytic technique (Nakagawa & Santos, 2012), we conducted two types of analyses to assess the reliability of our meta-analysis. First, we estimated heterogeneity using a modified version of the heterogeneity statistic I2 (originally defined by Higgins & Thompson, 2002; see Supporting information for details). Heterogeneity estimates the consistency of the results obtained from published studies used in our meta-analysis. A low level of heterogeneity means that discrepancies among the effects reported by the original studies are satisfactorily explained solely by sampling errors and are not the result of different methodologies or species characteristics (Higgins & Thompson, 2002; Higgins et al., 2003; Cooper & Hedges, 2009). Second, we assessed our data set for signs of publication bias by visually inspecting funnel plots (Sterne et al., 2005) and also by conducting Egger's regressions (Egger et al., 1997; Peters et al., 2006). We applied an Egger's regression design, modified for lnOR to data points consisting of the residuals of the effect sizes plus their measurement errors (Peters et al., 2006; see Supporting information for details). If the intercept of the Egger's regression is significantly different from zero, it is evidence of publication bias.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We observed small levels of heterogeneity in both BPMMs (reduced effort model: total I2 = 2.972; 95% CI: 0.284 to 40.519, species identity I2 = 0.320, 95% CI: 0.065 to 19.895, phylogeny I2 = 0.365, 95% CI: 0.081 to 32.479; increased effort model: total I2 = 3.189, 95% CI: 0.379 to 32.975, species identity I2 = 0.297, 95% CI: 0.042 to 11.912, phylogeny I2 = 0.374, 95% CI: 0.076 to 29.413), which suggests that the effect of parental effort on adult survival is uniform across the range of species investigated. Furthermore, there was little sign of publication bias in our data set (see funnel plots in Fig. S2 in Supporting information), which was supported by the intercepts of the Egger's regressions not being significantly different from zero (reduced effort model: β[intercept] = −0.837, 95% CI: −4.427 to 3.482; increased effort model: β[intercept] = −1.440, 95% CI: −5.305 to 1.994).

Regardless of sex, the probability of survival of individuals in the reduced parental effort group was not significantly different from that of control individuals (reduced effort model: β[male mean OR] = 0.850, 95% CI: 0.611 to 1.224; β[female mean OR] = 1.054, 95% CI: 0.770 to 1.491; Fig. 1). However, there was a sex effect on the probability of survival of individuals in the increased parental effort group. Males in the increased effort group suffered a reduction of approximately 26% in the probability of survival until the next breeding season when compared with control males, although this estimate had a large level of uncertainty (increased effort model: β[male mean OR] = 0.731, 95% CI: 0.534 to 0.983; Fig. 1). Conversely, females in the increased effort group survived just as well as females of the control group (increased effort model: β[female mean OR] = 0.950, 95% CI: 0.689 to 1.246; Fig. 1).

image

Figure 1. Results from the Bayesian phylogenetic mixed-effect meta-analyses. Posterior means and 95% CIs are shown for females and males. Effect sizes (log odds ratio; lnOR) were back-transformed to the original scale (odds ratio; OR) to facilitate interpretation. Labels on the y-axis represent the models from which the estimates were calculated (increased: increased parental effort model; reduced: reduced parental effort model). An OR smaller than 1 means that individuals from the treatment group had a lower probability of survival than control individuals; the converse (OR > 1) implies that treatment individuals had a higher probability of survival than control individuals. Vertical dashed line indicates OR = 1.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Over the past 50 years, studies of the trade-off between parental effort and survival have provided equivocal evidence for the cost of parental care (Stearns, 1992), with many studies failing to find any support or even some providing evidence against it (e.g. Reznick, 1985; Graves, 1991; Roff & Fairbairn, 2007). By combining estimates from a variety of species, the results of our meta-analysis represent a powerful quantitative assessment of whether birds indeed pay a cost of parental care in terms of survival. Our meta-analysis yielded three important results that will improve the understanding of this fundamental life-history trade-off. First, we found that bird species respond similarly to manipulations of parental effort regardless of their phylogenetic relationships (i.e. we found little evidence phylogenetic signal). Second, there is little evidence of survival benefits for either sex when parental effort was experimentally reduced (Fig. 1). Lastly and perhaps most importantly, the life-history trade-off between parental effort and parental survival differs between sexes. On one hand, males of the experimentally increased effort group suffered a significant reduction in the probability of survival when compared with control males. On the other hand, females did not appear to suffer decreased survival when subjected to greater levels of parental effort.

Despite a limited number of species included in our analyses (19 species), we covered taxa from a wide taxonomic range (13 families), including representatives from distantly related clades, such as Anatidae to Muscicapidae (see Fig. S1). The lack of phylogenetic signal in this fundamental trade-off between parental effort and survival may indicate that the process of this trade-off operates homogeneously across species because it evolved in a shared ancestor (see Losos, 2011 for an extended discussion on causes for the lack of phylogenetic effect). This observation, in turn, could imply that energy allocation to survival is similar irrespective of the organism's life-history strategy (i.e. slow or fast life histories). However, it is worth noting that the limited number of available species could be responsible for our analysis not being able to detect phylogenetic signal. Under this scenario, phylogeny might potentially play a role in the trade-off between parental effort and survival (for a discussion of phylogenetic impact on meta-analysis, see Chamberlain et al., 2012).

We found little evidence of survival benefits from a reduction in parental effort. One must bear in mind that, prior to experimental manipulation, birds are probably allocating an optimum amount of resources to survival. Thus, it is possible that a reduction in parental effort (either experimentally or naturally) creates an energy surplus that will not be allocated to survival, but instead will be allocated to other life-history traits. For example, male collared flycatchers, Ficedula albicollis, subjected to an experimental decrease in brood size had larger forehead patches in the next breeding season (Gustaffson et al., 1995). Male eastern bluebirds, Sialia sialis, with reduced reproductive effort experience an increase in plumage brightness (Siefferman & Hill, 2005). Finally, blue tits, Cyanistes caeruleus, exposed to lower reproductive effort, have a higher probability of undertaking a second reproductive attempt during the breeding season (Parejo & Danchin, 2006). These examples and our results support the notion that individuals that suffer a reduction in parental effort can potentially gain future benefits in terms of additional mating opportunities.

A sex-specific survival cost of reproduction indicates that males and females may have evolved distinct parental allocation strategies to cope with perceived brood value (brood size). Recent studies have shown that males increase their parental investment when they perceive an opportunity to enhance their reproductive success (Harrison et al., 2009). Further studies have shown that females do not seem as responsive as males to increases in brood value (Mock et al., 2005; Ardia, 2007; Nakagawa et al., 2007). A potential explanation for this sex-specific difference in the sensitivity of parental demand lies in the baseline amount of variation in parental provisioning behaviour. Females are perhaps working close to their maximum provisioning capacity at natural levels of parental demand (MacGregor & Cockburn, 2002; Low et al., 2011). Consequently, an experimental increase in demand will only have a small impact on this workload, as there is no room to intensify the provisioning rates. Conversely, males often provision at lower rates than females and display greater between-individual variation. Thus, it is conceivable that an increase in brood value and, subsequently, potential reproductive success could cause males to shift their allocation strategies to supply more parental care. With increased parental effort, these males will suffer greater survival costs. Given that multiple males often sire a brood, these distinct parental allocation strategies are likely the result of the difference in the degree of certainty with which male and female parents are related to their offspring (Westneat & Sherman, 1993; Queller, 1997; but see Sheldon, 2002), or the probability of remating of each sex if they desert the brood (Olson et al., 2008).

Contrary to life-history theory, our results suggest that female birds do not seem to suffer a survival cost of increased parental effort. One potential explanation for our results is that females indeed do not bear the costs of parental effort, but instead transfer the costs to their offspring, a possibility that has been suggested repeatedly in qualitative reviews (see Lindén & Møller, 1989; Martin, 2004). Interestingly, it has been recently shown in mammals that, when resources are limited, breeding females adopt a conservative energy allocation strategy by transferring the costs of reproduction to their offspring (i.e. females ensure their own mass gain over their offspring's; Hamel et al., 2010; Martin & Festa-Bianchet, 2010). Studies that report negative effects of increased parental effort (either through manipulations of clutch or brood size) on offspring recruitment provide further evidence of the transfer of costs from mothers to offspring (e.g. Smith et al., 1989; Knowles et al., 2010). Stearns (1992) briefly discussed in his seminal book that most research on life-history trade-offs focused on intra- rather than intergenerational costs (e.g. the trade-off between parental effort and survival). Based on our results and the fact that studies that investigated the transfer of costs to offspring are scarce, we suggest that future research should follow the recommendation of investigating both intra- and intergenerational costs. More formally, when there is reason to expect little sensitivity to changes in one life-history trait, as seems to be the case with female parental effort, we predict that the costs of parental effort will be directed towards current and future progeny (i.e. intergenerational trade-offs). However, when there are reasons to expect high sensitivity (e.g. male parental effort in response to shifts in perceive brood value), we predict that the classic intragenerational trade-off (i.e. negative relationship between effort and survival) will occur. For instance, in some biparental species, we do not expect to observe negative effects of the cost transferred to offspring by the females, because males could offset this effect by increasing their efforts considerably. When mothers are the sole providers of care, as is the case in mammals (e.g. Hamel et al., 2010; Martin & Festa-Bianchet, 2010), however, we should observe the negative effects (in terms of decreased survival and recruitment) of the transfer of costs to offspring.

In conclusion, this study provides a quantitative test of the trade-off between parental effort and survival in a wide range of bird species. Our results shed new light on life-history relationships, demonstrating that increases in parental effort do not necessarily translate in reductions in survival (Fig. 1). Hence, as has been advocated for over 20 years (Lindén & Møller, 1989; Martin, 2004), future experiments of the life-history trade-off between effort (either reproductive or parental) and survival should not only measure its fitness effects in the manipulated individuals (intragenerational trade-off), but also measure the fitness consequences on current and future progeny (intergenerational trade-off). This integrative approach to the study of life-history trade-offs will improve our understanding of other evolutionary processes, such as developmental plasticity and parental effect (Qvarnström & Price, 2001; Uller, 2008). Furthermore, comparative analyses and meta-analyses could provide important evidence for the relationship between sex-specific sensitivities (e.g. variation in parental behaviour in birds) and how trade-offs are expressed (intra- or intergenerationally). A comparison of the expression of parental effort trade-offs between taxa in which parental care is predominantly maternal (e.g. mammals) and taxa in which care is predominantly paternal (e.g. some fish, frogs or arthropods) should provide some clues to the issues regarding sex specificity and the transgenerational effects of the trade-off between parental effort and survival.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We thank Glauco Machado, Julia Schroeder, Alistair Senior, Bruce Lyon and three anonymous reviewers for their most valuable suggestions and comments on an earlier version of this paper. This work was supported by a University of Otago postgraduate scholarship to ESAS and a Marsden grant (UOO-0812) and a Humboldt Fellowship to SN.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Data deposited at Dryad: doi: 10.5061/dryad.mt6m4

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
jeb2569-sup-AppendixS1-FigS1-S2-TableS1-S4.docWord document608KAppendix S1 Supplementary methods. Figure S1 Topology of the phylogenetic tree used in the Bayesian phylogenetic mixed-effects meta-analysis models. Figure S2 Funnel plots of original effect sizes (log Odds Ratio; lnOR) plotted against their corresponding precision (the square root of sample size). Table S1 Studies that reported data on the effectiveness of experimental manipulation in changing parental effort. Table S2 Studies that were found in our literature search but were excluded from our meta-analysis because they did not comply with our inclusion criteria. Table S3 Studies used in our meta-analyses of the trade-off between parental effort and survival in birds. Table S4 Results of the Bayesian mixed-effect models including the variable ‘Type of Manipulation’ as a fixed effect.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.