Discussion
Our systematic map of marine complex life histories identified emphases and knowledge gaps in the literature. Most studies of life histories were done in the laboratory on the F0 adult, embryonic and larval stages in isolation. The most important issue our map identified was the lack of studies that both explore the entire life history and follow the same individuals over time. Studies that measured more than one stage almost exclusively followed cohorts rather than individuals—following individuals through the life cycle was rare (1.7%). If we are to better understand the phenotypic links across life-history stages, free of the potentially confounding influence of the Simpson’s Paradox, then individual longitudinal studies should be priority.
Experimental design and stages . The most substantial gap identified in the systematic map is the underrepresentation of individual longitudinal studies. Studies that focus on a single stage make up ~67% of studies in the map, meaning the literature has focused on characterising life histories, rather than exploring links between stages. About 31% of studies used a cohort longitudinal design—following a single cohort is useful for quantitative genetic studies or when offspring traits of the entire brood are of interest (e.g. . Our map found that just 1.7% of studies were individual longitudinal. Individual longitudinal studies have been suggested as the best approach for addressing many life-history questions including those on senescence, age-related changes in the life history, population density regulation and links between stages . An interesting next step would be to explore whether a lack of individual longitudinal studies also pervades studies of complex life histories in terrestrial taxa (e.g. insects). We predict a similar dearth of such studies in these taxa also.
Answering life-history questions often requires individual longitudinal studies that measure the entire life history—for example, estimating lifetime reproductive output between parents and offspring or over multiple cohorts is ideal for estimating selection on a trait or to project population dynamics . However, measuring individuals over the entire life history is rare—long-term studies are costly, logistically difficult to maintain and, therefore, risky to undertake, particularly for long-lived species . We found only one case in which all stages of the life cycle were measured—an individual longitudinal study that used a species with only four stages (i.e. direct development). Importantly, we know other studies exist that meet some of our criteria (and some are even well-known studies), but were not captured by our systematic map. This highlights an important limitation to systematic maps that must be acknowledged: no map will be perfectly comprehensive, so missing some important studies is unavoidable. Expanding search terms further generated an impractical number of papers to process, as the 17,000+ papers that we did screen required 100s of people-hours. Nevertheless, our map can be considered a representative and unbiased sample of the literature such that the relative abundance of different study types is unlikely to change, were broader terms used. Suffice it to say, regardless of how they are searched for, individual longitudinal studies remain rare in the literature.
We find that cohort longitudinal studies measured the F0adult, embryonic and larval stages most often (Figures 4a, 5a). However, we found the opposite trend in individual longitudinal studies which measured metamorphs, juveniles and F1 adults most often (Figure 4c). Why do studies following individuals mostly measure stages after metamorphosis? Of the studies that used an individual longitudinal approach, 71% used species with lecithotrophic larvae (i.e. planktonic, non-feeding), which also have relatively short development times. Culturing individuals with long development times hinder whole life cycle studies of those species—for example, it is much more difficult to study the whole life history of the planktotrophic (i.e. planktonic, feeding larvae) sea star Pisaster ochraceus , which matures in 5 years , compared to the lecithotrophic marine bryozoan Bugula neritina , which takes ~7 weeks to mature . Because of the challenge in executing individual longitudinal studies, our understanding of links between immature stages and F1adults is largely a black box—only 12 individual longitudinal experiments in our map measured both juveniles and F1adults, and none was done on species with long development times (i.e. planktotrophs). Ultimately, the goal in studying links between life-history stages is to understand their long-term consequences—individual longitudinal studies have important implications for conservation efforts , predicting responses to climate change , and can also better our understanding on the evolution of life histories . Our literature map highlights how limited our understanding of long-term theoretical and applied consequences of phenotypic links is—improving culturing methods so we can make measurements at maturity is an important first step in rectifying this problem.
Laboratory versus Field. Unsurprisingly, 12% of studies in the dataset were conducted under field conditions, and of the field experiments, less than 6% were individual longitudinal studies. Conducting experiments in the field is important because tests in the laboratory can sometimes yield conflicting results—for example if a trade-off only exists in nature . The first and best examples of field-based, individual longitudinal studies are in birds, primates and other mammals—these studies are disproportionately represented in the literature, reflecting both how rare and difficult generating individual-based studies in the field is, as well as how useful they are in answering ecological and evolutionary questions . Field experiments will help reveal whether patterns, or links between stages, observed in the laboratory, match patterns under natural conditions.
Taxonomic representation. For the 15 phyla in the dataset, molluscs, echinoderms and arthropods make up ~70% of species. Mollusca is the largest marine phylum with ~50,000 species, followed by arthropods and then echinoderms . We found an overrepresentation of echinoderms— Echinodermata is the third largest marine phylum but the second most abundant in our map. The overrepresentation of echinoderms may suggest a bias toward certain model species—five of the ten most common species in the dataset were echinoderms: four sea urchins (Arbacia punctulata , Paracentrotus lividus , Strongylocentrotus droebachiensis , Strongylocentrotus purpuratus ) and one sand dollar (Dendraster excentricus ). While model species, likeStrongylocentrotus purpuratus , Drosophila melanogaster orDanio rerio , are convenient and often economical, they do not necessarily represent the consequences of links between life-history stages across all taxa. The presence of links depends on species—for example, significant genetic correlations between tadpole and adult morphological traits were found for the anuran Hyla regilla , but not Hyla cinerea or Rana sylvatica . To understand whether patterns of links in the life history apply broadly across taxa, future studies should use taxonomic groups that are insufficiently represented.
Summary. Our systematic map shows that, for over 100 years, the studies of marine invertebrates have primarily collected data conducive to characterising life histories, rather than understanding links between stages. Studies following a single cohort are an important first step for answering questions about development but when trying to understand links across stages, cohort studies may lead to Simpson’s Paradox, where the pattern across cohorts does not reflect the pattern across individuals. The best strategy for avoiding Simpson’s Paradox is to conduct individual longitudinal studies—those that follow individuals through the entire life history, but these remain exceedingly rare, accounting for just 1.7% of studies in our map. Reviews in terrestrial systems also suggest that we move toward an individual longitudinal approach when studying life histories , and after systematically quantifying study types in the field using marine invertebrates as a model system, we support those reviews. We acknowledge and celebrate the tremendous progress in the field of complex life cycles, but we hope that our map encourages future studies to use an individual longitudinal approach to understand the ecological and evolutionary significance of phenotypic links across life-history stages.
Acknowledgements: We thank Annie Guillaume, Henry Wootton and Victor Shelamoff for helping to collect data. We also thank Liz Morris for editing the manuscript.