No evidence for a trade-off between reproduction and survival in a meta-analysis across birds
- Ecology & Evolutionary Biology, School of Biosciences, The University of Sheffield, United Kingdom
Figures
Figure 1 with 2 supplements
The effects size (log odds of survival given an increase in clutch size) for three different measures of clutch size.
(A) Raw, (B) standardised, and (C) proportional clutch size. Coloured points are the combined effect sizes of the odds ratios with their 95% confidence intervals (n = 58 experimental and 20 observational). Points are coloured by whether they represent brood manipulation experiments (costs of reproduction) or they are observational (quality). Grey points are the odds ratios of each study, with their sizes weighted by the sampling variance.
Figure 1—figure supplement 1
Survival effects for increasing clutch size for female, male, and mixed-sex studies.
The clutch size was measured in three ways; (A) raw clutch size, (B) standardised, and (C) proportional. Separate meta-analyses were run for observational (n = 11 [female], 7 [male], and 2 [mixed sex]) and brood manipulation (n = 30 [female], 20 [male], and 8 [mixed sex]) studies. Points are the combined effect size and whiskers are the 95% confidence intervals.
Figure 1—figure supplement 2
Funnel plot of meta-analysis residuals against standard error.
Brood manipulation and observational data are combined. Red points are for experimental studies and blue points are observational studies.
Figure 2 with 1 supplement
The meta-analytic linear regression (Appendix 1—table 1) of the effect size for increasing clutch size (per egg) on parental survival, given the average clutch size for the species for (A) brood manipulation and (B) observational studies.
Species with small clutch sizes showed stronger costs of reproduction and a stronger relationship with quality (Interaction between treatment and species clutch size effect size = –0.036, p=0.015). The points are the survival effect sizes (log odds ratio) per egg (as in Figure 1A) on parental survival in each study, with the point size reflecting the meta-analytic weight of that study (n = 58 experimental and 20 observational).
Figure 2—figure supplement 1
Brood size comparisons across studies.
(A) The percentage change in brood size of the maximum manipulation (regardless of whether the manipulation was an increase or reduction) in experimental studies given the species’ average brood size. (B) The brood size of each experimental study. Points indicate the species’ average brood size (i.e., before manipulation occurred), shaded bars indicate the natural standard deviation in clutch size observed in the species (i.e., without manipulation), and the whiskers represent the lower and upper clutch sizes after a manipulation occurred.
Figure 3
Phylogenetic tree of species included in our meta-analysis.
Numeric labels denote the node support (as a percentage). Coloured points indicate whether the species were included in the experimental or observational studies.
Figure 4
Isoclines of selection differentials among hypothetical control populations (in which individuals reproduce at the species’ mean rate) and hypothetical brood-manipulated populations (where individuals reproduce at an increased rate compared to control) for their whole lives.
Selection differentials (i.e., the relative difference in lifetime reproductive output between hypothetical control and brood-manipulated populations) above 1 represents high lifetime fitness. Survival rates, clutch sizes, the magnitude of the manipulation (chicks added) and effect sizes represent the range of these variables present in the studies used in our meta-analysis. For each clutch size, we used a predicted survival rate and effect size to give isoclines that are biologically meaningful (exemplar birds shown in red). Arrows indicate the relative size and direction of selection in life-history space (on the reproduction axis). The costs of reproduction we estimated within species are predicted to result in a fast–slow life-history continuum across species, and the exemplar species we used as examples fit on this diagonal of survival rate/ clutch size combinations. We suggest that individual species show limited costs of reproduction as they operate within relatively wide constraints imposed by the cost of reproduction that is responsible for the strong life-history trade-off observed across species.
Figure 5
Decision tree representing the logical steps from our original hypothesis to our overall interpretation of our findings.
Tables
Table 1
Effect size estimates for the odds of survival with increasing clutch size (raw, standardised, and proportional clutch size).
The p-values indicate the difference between brood manipulations and observational data, with the individual effect p-values (from zero) in parentheses. Values in bold show statistical significance of p < 0.05.
| Parameter | Effect size | 95% CI lower bounds | 95% CI upper bounds | p Value | p Value (Individual) | ||
|---|---|---|---|---|---|---|---|
| Raw | Clutch size | Brood manipulation | –0.0522 | –0.1406 | 0.0363 | 0.0007 | (0.2477) |
| Observational | 0.0747 | 0.1571 | –0.0288 | (0.1571) | |||
| Standardised | Clutch size | Brood manipulation | –0.0651 | –0.1478 | 0.0177 | 0.0065 | (0.1232) |
| Observational | 0.1143 | –0.0046 | 0.2333 | (0.0595) | |||
| Proportional | Clutch size | Brood manipulation | –0.2703 | –0.4984 | –0.0423 | 0.0005 | (0.0202) |
| Observational | 0.3850 | 0.0583 | 0.7116 | (0.0209) |
-
Model = ~obervational_or_experimental, random = (species, phylogeny, study reference).
Appendix 1—table 1
Model outputs for meta-analyses estimating the effect size of the odds of survival for increasing clutch size given the species’ average clutch size.
Treatment was coded as a categorical variable indicating whether studies were either experimental or observational. The species’ average clutch size was centred to the average clutch size of all species used in the meta-analysis. The increase in clutch size was modelled as the raw increase in clutch size and the standardised increase in clutch size. We have not presented the proportional increase in clutch size as this represents the change from the species average and so is null in this model. Values in bold show statistical significance of p < 0.05.
| Model | Parameter | Effect size | 95% CI lower bounds | 95% CI upper bounds | p Value |
|---|---|---|---|---|---|
| Raw | Intercept | –0.047 | –0.147 | 0.054 | 0.363 |
| Treatment: observational | 0.150 | 0.079 | 0.222 | <0.0001 | |
| Centred species clutch size | 0.011 | –0.012 | 0.034 | 0.337 | |
| Treatment: observational × species clutch size | –0.036 | –0.066 | –0.007 | 0.015 | |
| Standardised | Intercept | –0.065 | –0.222 | 0.092 | 0.418 |
| Treatment: observational | 0.202 | 0.074 | 0.330 | 0.002 | |
| Species clutch size | 0.015 | –0.026 | 0.055 | 0.482 | |
| Treatment: observational × species clutch size | –0.057 | –0.117 | 0.002 | 0.057 |
-
Model = ~observational_or_experimental * mean_adjusted_clutchsize, random = (species, phylogeny, study reference).
Appendix 1—table 2
Model outputs for survival given increasing clutch size for brood manipulation (n=30 [female], 20 [male], and 8 [mixed sex]) and observational studies for the different sexes (n=11 [female], 7 [male], and 2 [mixed sex]).
Mixed-sex studies were found to be at the extremes of the trend, a reflection of species who lay smaller clutch sizes rather than an effect of the mixed sex itself. Values in bold show statistical significance of p < 0.05.
| Clutch size measure | Sex | Estimate | SE | p Value | CI.lb | CI.ub | |
|---|---|---|---|---|---|---|---|
| Raw | Brood manipulation | Female | –0.0361 | 0.0407 | 0.3754 | –0.1158 | 0.0437 |
| Male | 0.0186 | 0.0452 | 0.6807 | –0.07 | 0.1072 | ||
| Mixed | –0.2079 | 0.0617 | 0.0007 | –0.3287 | –0.087 | ||
| Observational | Female | 0.1279 | 0.104 | 0.2189 | –0.076 | 0.3317 | |
| Male | 0.0232 | 0.1067 | 0.8282 | –0.186 | 0.2324 | ||
| Mixed | 0.49 | 0.2638 | 0.0632 | –0.027 | 1.007 | ||
| Standardised | Brood manipulation | Female | –0.082 | 0.0665 | 0.2179 | –0.2124 | 0.0484 |
| Male | 0.0318 | 0.0764 | 0.6778 | –0.1181 | 0.1816 | ||
| Mixed | –0.2686 | 0.1264 | 0.0336 | –0.5163 | –0.0209 | ||
| Observational | Female | 0.1732 | 0.1491 | 0.2452 | –0.119 | 0.4654 | |
| Male | 0.0076 | 0.156 | 0.9614 | –0.2982 | 0.3133 | ||
| Mixed | 0.5216 | 0.3214 | 0.1047 | –0.1084 | 1.1516 | ||
| Mean adjusted | Brood manipulation | Female | –0.2776 | 0.1909 | 0.1458 | –0.6517 | 0.0965 |
| Male | 0.1135 | 0.2368 | 0.6318 | –0.3506 | 0.5776 | ||
| Mixed | –0.6169 | 0.2773 | 0.0261 | –1.1604 | –0.0734 | ||
| Observational | Female | 0.5721 | 0.3148 | 0.0692 | –0.0449 | 1.1892 | |
| Male | 0.0639 | 0.3287 | 0.8459 | –0.5803 | 0.7081 | ||
| Mixed | 0.9363 | 0.5614 | 0.0953 | –0.164 | 2.0366 |
-
Model = observational_or_experimental + sex, random = (species, phylogeny, study reference).
Appendix 1—table 3
I2 values for each model showing the proportion of variation accounted for by the random effects of the model.
The phylogenetic signal was included as a correlation matrix within the model.
| Model | I2 | ||||
|---|---|---|---|---|---|
| Total | Species | Phylogenetic | Reference | Total species effect (species + phylogenetic) | |
| Raw | 0.494 | 0.000000003 | 0.287 | 0.208 | 0.287 |
| Standardised | 0.542 | 0.080 | 0.00000002 | 0.463 | 0.080 |
| Proportional | 0.428 | 0.137 | 0.00000001 | 0.291 | 0.137 |
Appendix 1—table 4
Excluded studies and the rationale for exclusion.
| Reference | Species | Reason for exclusion |
|---|---|---|
| Ashcroft, 1979 | Puffinus puffinus | No parental survival values given clutch/brood size. Also no clutch/brood size variation in focal species. |
| Erikstad et al., 2009 | P. puffinus | No clutch/brood size manipulation. Manipulation is age of offspring. |
| Wernham and Bryant, 1998 | P. puffinus | No clutch/brood size variation in study. |
| Wiebe, 2005 | Colaptes auratus | Mate removal, not clutch/brood manipulation. |
| Askenmo, 1979 | Ficedula hypoleuca | Doesn't state manipulation size. |
| Tinbergen and Both, 1999 | Parus major | Manipulation is to equalise brood size throughout population. |
| Annett and Pierotti, 1999 | Larus occidentalis | Breeding lifespan not survival. |
| Murphy, 2007 | Tyrannus tyrannus | No survival values given. |
| Lessells, 1986 | Branta canadensis | No parental survival values given clutch/brood size. |
| Schaub and von Hirschheydt, 2009 | Hirundo rustica | Clutch sizes are pooled (0 offspring,1–6 offspring and 6+ offspring) with large variation in each group meaning it is not informative for our study to gain reasonably accurate survival given clutch size raised. |
| Milonoff and Paananen, 1993 | Bucephala clangula | Clutch size before manipulation varies significantly. |
| Blondel et al., 1998 | Parus caeruleus | No parental survival values given clutch/brood size. |
| Knowles et al., 2010 | P. caeruleus | No parental survival values given clutch/brood size. |
| Kluijver, 1971 | P. major | Combined first and second broods. |
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
No evidence for a trade-off between reproduction and survival in a meta-analysis across birds
eLife 12:RP87018.
https://doi.org/10.7554/eLife.87018.5
