The effect of dietary restriction on reproduction: a meta-analytic perspective

Abstract

Background: Dietary restriction (DR), a reduction in the amount of food or particular nutrients eaten, is the most consistent environmental manipulation to extend lifespan and protect against age related diseases. Current evolutionary theory explains this effect as a shift in the resolution of the trade-off between lifespan and reproduction. However, recent studies have questioned the role of reproduction in mediating the effect of DR on longevity and no study has quantitatively investigated the effect of DR on reproduction across species.

Results: Here we report a comprehensive comparative meta-analysis of the effect of DR on reproduction. In general, DR reduced reproduction across taxa, but several factors moderated this effect. The effect of DR on reproduction was greater in well-studied model species (yeast, nematode worms, fruit flies and rodents) than non-model species. This mirrors recent results for longevity and, for reproduction, seems to result from a faster rate of decline with decreasing resources in model species. Our results also suggested that not all reproductive traits are affected equally by DR. High and moderate cost reproductive traits suffered a significant reduction with DR, but low cost traits, such as ejaculate production, did not. Although the effect of DR on reproduction was stronger in females than males, this sex difference reduced to near zero when accounting for other co-factors such as the costliness of the reproductive trait. Thus, sex differences in the effect of DR on longevity may be due to a failure to expose males to as complete a range of the costs of reproduction as females.

Conclusions: We suggest that to better understand the generality of the effect of DR, future studies should attempt to address the cause of the apparent model species bias and ensure that individuals are exposed to as many of the costs of reproduction as possible. Furthermore, our meta-analytic approach reveals a general shortage of DR studies that record reproduction, particularly in males, as well as a lack of direct side-by-side comparisons of the effect of DR on males and females.

Keywords: Breeding; Life history trade-off; Meta-analysis; Nutrition; Systematic review.

Fig. 1

PRISMA flow diagram of data collection. The number of papers identified initially through key word searching is shown in the identification boxes. The number of papers excluded is shown for each stage of screening. Reasons for exclusion are given for papers that made it to final eligibility screening

Fig. 2

The effect of degree of restriction on effect size in model and non-model species. Effect sizes are Cohen’s d, the standardised mean difference in reproduction between the control and restricted groups (see Methods and Additional file 1: Dialog S1). Model species are represented by squares and the dashed line. Non-model species are represented by circles and solid line. Model species suffer a greater rate of decline in reproduction with increasing degree of restriction. Point sizes indicate the variance in the estimate of the effect size. Details of statistics are given in the main text

Fig. 3

Forest plots showing effect sizes (Cohen’s d, standardised mean difference in reproduction between the control and restricted groups (see Methods and Additional file 1: Dialog S1)) of key moderators for the effect of dietary restriction (DR) on reproduction. Each point represents the Cohen’s d value with the 95 % credible intervals (CIs). Panel a represents the outputs from univariate models, with each moderator fitted individually. Each moderator subgroup (e.g. model or non-model species) is represented by a single point. Contrasts represent the difference between effect sizes of the subgroups (e.g. the difference between model (M) and non-model (N) species). Restriction:Model, represents the interaction between degree of restriction (%) and model or non-model species. Panel b shows the output from our full model accounting for all moderators, with each point representing the effect size for that moderator