Ignoring non‐English‐language studies may bias ecological meta‐analyses

Meta‐analysis plays a crucial role in syntheses of quantitative evidence in ecology and biodiversity conservation. The reliability of estimates in meta‐analyses strongly depends on unbiased sampling of primary studies. Although earlier studies have explored potential biases in ecological meta‐analyses, biases in reported statistical results and associated study characteristics published in different languages have never been tested in environmental sciences. We address this knowledge gap by systematically searching published meta‐analyses and comparing effect‐size estimates between English‐ and Japanese‐language studies included in existing meta‐analyses. Of the 40 published ecological meta‐analysis articles authored by those affiliated to Japanese institutions, we find that three meta‐analysis articles searched for studies in the two languages and involved sufficient numbers of English‐ and Japanese‐language studies, resulting in four eligible meta‐analyses (i.e., four meta‐analyses conducted in the three meta‐analysis articles). In two of the four, effect sizes differ significantly between the English‐ and Japanese‐language studies included in the meta‐analyses, causing considerable changes in overall mean effect sizes and even their direction when Japanese‐language studies are excluded. The observed differences in effect sizes are likely attributable to systematic differences in reported statistical results and associated study characteristics, particularly taxa and ecosystems, between English‐ and Japanese‐language studies. Despite being based on a small sample size, our findings suggest that ignoring non‐English‐language studies may bias outcomes of ecological meta‐analyses, due to systematic differences in study characteristics and effect‐size estimates between English‐ and non‐English languages. We provide a list of actions that meta‐analysts could take in the future to reduce the risk of language bias.     

Nonindependence and sensitivity analyses in ecological and evolutionary meta‐analyses

Meta‐analysis is an important tool for synthesizing research on a variety of topics in ecology and evolution, including molecular ecology, but can be susceptible to nonindependence. Nonindependence can affect two major interrelated components of a meta‐analysis: (i) the calculation of effect size statistics and (ii) the estimation of overall meta‐analytic estimates and their uncertainty. While some solutions to nonindependence exist at the statistical analysis stages, there is little advice on what to do when complex analyses are not possible, or when studies with nonindependent experimental designs exist in the data. Here we argue that exploring the effects of procedural decisions in a meta‐analysis (e.g. inclusion of different quality data, choice of effect size) and statistical assumptions (e.g. assuming no phylogenetic covariance) using sensitivity analyses are extremely important in assessing the impact of nonindependence. Sensitivity analyses can provide greater confidence in results and highlight important limitations of empirical work (e.g. impact of study design on overall effects). Despite their importance, sensitivity analyses are seldom applied to problems of nonindependence. To encourage better practice for dealing with nonindependence in meta‐analytic studies, we present accessible examples demonstrating the impact that ignoring nonindependence can have on meta‐analytic estimates. We also provide pragmatic solutions for dealing with nonindependent study designs, and for analysing dependent effect sizes. Additionally, we offer reporting guidelines that will facilitate disclosure of the sources of nonindependence in meta‐analyses, leading to greater transparency and more robust conclusions.     

Sexual selection and sperm quantity: meta‐analyses of strategic ejaculation

Multiple mating or group spawning leads to post‐copulatory sexual selection, which generally favours ejaculates that are more competitive under sperm competition. In four meta‐analyses we quantify the evidence that sperm competition (SC) favours greater sperm number using data from studies of strategic ejaculation. Differential investment into each ejaculate emerges at the individual level if males exhibit phenotypic plasticity in ejaculate properties in response to the likely risk and/or intensity of sperm competition after a given mating. Over the last twenty years, a series of theoretical models have been developed that predict how ejaculate size will be strategically adjusted in relation to: (a) the number of immediate rival males, with a distinction made between 0 versus 1 rival (‘risk’ of SC) and 1 versus several rivals (‘intensity’ of SC); (b) female mating status (virgin or previously mated); and (c) female phenotypic quality (e.g. female size or condition). Some well‐known studies have reported large adjustments in ejaculate size depending on the relevant social context and this has led to widespread acceptance of the claim that strategic sperm allocation occurs in response to each of these factors. It is necessary, however, to test each claim separately because it is easy to overlook studies with weak or negative findings. Compiling information on the variation in outcomes among species is potentially informative about the relevance of these assumptions in different taxa or mating systems. We found strong evidence that, on average, males transfer larger ejaculates to higher quality females. The effect of female mating status was less straightforward and depended on how ejaculate size was measured (i.e. use of proxy or direct measure). There is strong evidence that ejaculate size increased when males were exposed to a single rival, which is often described as a response to the risk of SC. There is, however, no evidence for the general prediction that ejaculate size decreases as the number of rivals increases from one to several males (i.e. in response to a higher intensity of SC which lowers the rate of return per sperm released). Our results highlight how meta‐analysis can reveal unintentional biases in narrative literature reviews. We note that several assumptions of theoretical models can alter an outcome's predicted direction in a given species (e.g. the effect of female mating status depends on whether there is first‐ or last‐male sperm priority). Many studies do not provide this background information and fail to make strong a priori predictions about the expected response of ejaculate size to manipulation of the mating context. Researchers should be explicit about which model they are testing to ensure that future meta‐analyses can better partition studies into different categories, or control for continuous moderator variables.     

Bet hedging via multiple mating: A meta‐analysis

Polyandry has been hypothesized to allow females to “bet hedge” against mating only with unsuitable mates, reducing variance in offspring fitness between members of a polyandrous lineage relative to a single‐mating one. Theoretically, this reduction in fitness variance could select for polyandrous genotypes even when polyandry carries a direct cost, especially in small populations. However, this hypothesis is controversial and difficult to test empirically. Here, I apply a novel simulation model to 49 published empirical datasets, and quantify the potential selective advantage of multiple mating via reduced offspring fitness variance. For a wide range of assumptions, including those that most favor the evolution of bet hedging, I show that any fitness gains are meager. The variance in offspring quality caused by mate identity does not appear to be high enough for bet hedging to drive the evolution of polyandry.     

Formalized synthesis opportunities for ecology: systematic reviews and meta‐analyses

Narrative reviews are dead. Long live systematic reviews (and meta‐analyses). Synthesis in many forms is now a driving force in ecology. Advances in open data for ecology and new tools provide vastly improved capacity for novel, emergent knowledge synthesis in our discipline. Systematic reviews and meta‐analyses are two formal synthesis opportunities for ecologists that are now accepted as traditional publications, but the scope of validated syntheses will continue to expand. To date, systematic reviews are rarely used whilst the rate of meta‐analyses published in ecological journals is increasing exponentially. Systematic reviews provide an overview of the literature landscape for a topic, and meta‐analyses examine the strength of evidence integrated across different studies. Effective synthesis benefits from both approaches, but better data reporting and additional advances in the culture of sharing data, code, analytics, workflows, methods and also ideas will further energize these efforts. At this junction, synthetic efforts that include systematic reviews and meta‐analyses should continue as stand‐alone publications. This is a necessary step in the evolution of synthesis in our discipline. Nonetheless, they are still evolving tools, and meta‐analyses in particular are simply an extended set of statistical tests. Admittedly, understanding the statistics and assumptions influence how we conduct synthesis much as statistical choices often shape experimental design, i.e. ANOVA versus regression‐based experiments, but statistics do not make the paper. Current steps – primary research articles need to more effectively report evidence, sharing scientific products should expand, systematic reviews should be used to identify research gaps/delineate literature landscapes, and meta‐analyses should be used to examine evidence patterns to further predictive ecology.     

Bivariate random‐effects meta‐analysis models for diagnostic test accuracy studies using arcsine‐based transformations

Diagnostic or screening tests are widely used in medical fields to classify patients according to their disease status. Several statistical models for meta‐analysis of diagnostic test accuracy studies have been developed to synthesize test sensitivity and specificity of a diagnostic test of interest. Because of the correlation between test sensitivity and specificity, modeling the two measures using a bivariate model is recommended. In this paper, we extend the current standard bivariate linear mixed model (LMM) by proposing two variance‐stabilizing transformations: the arcsine square root and the Freeman–Tukey double arcsine transformation. We compared the performance of the proposed methods with the standard method through simulations using several performance measures. The simulation results showed that our proposed methods performed better than the standard LMM in terms of bias, root mean square error, and coverage probability in most of the scenarios, even when data were generated assuming the standard LMM. We also illustrated the methods using two real data sets.     

Net effects of multiple stressors in freshwater ecosystems: a meta‐analysis

The accelerating rate of global change has focused attention on the cumulative impacts of novel and extreme environmental changes (i.e. stressors), especially in marine ecosystems. As integrators of local catchment and regional processes, freshwater ecosystems are also ranked highly sensitive to the net effects of multiple stressors, yet there has not been a large‐scale quantitative synthesis. We analysed data from 88 papers including 286 responses of freshwater ecosystems to paired stressors and discovered that overall, their cumulative mean effect size was less than the sum of their single effects (i.e. an antagonistic interaction). Net effects of dual stressors on diversity and functional performance response metrics were additive and antagonistic, respectively. Across individual studies, a simple vote‐counting method revealed that the net effects of stressor pairs were frequently more antagonistic (41%) than synergistic (28%), additive (16%) or reversed (15%). Here, we define a reversal as occurring when the net impact of two stressors is in the opposite direction (negative or positive) from that of the sum of their single effects. While warming paired with nutrification resulted in additive net effects, the overall mean net effect of warming combined with a second stressor was antagonistic. Most importantly, the mean net effects across all stressor pairs and response metrics were consistently antagonistic or additive, contrasting the greater prevalence of reported synergies in marine systems. Here, a possible explanation for more antagonistic responses by freshwater biota to stressors is that the inherent greater environmental variability of smaller aquatic ecosystems fosters greater potential for acclimation and co‐adaptation to multiple stressors.     

The efficacy and safety of nutrient supplements in the treatment of mental disorders: a meta‐review of meta‐analyses of randomized controlled trials

The role of nutrition in mental health is becoming increasingly acknowledged. Along with dietary intake, nutrition can also be obtained from “nutrient supplements”, such as polyunsaturated fatty acids (PUFAs), vitamins, minerals, antioxidants, amino acids and pre/probiotic supplements. Recently, a large number of meta‐analyses have emerged examining nutrient supplements in the treatment of mental disorders. To produce a meta‐review of this top‐tier evidence, we identified, synthesized and appraised all meta‐analyses of randomized controlled trials (RCTs) reporting on the efficacy and safety of nutrient supplements in common and severe mental disorders. Our systematic search identified 33 meta‐analyses of placebo‐controlled RCTs, with primary analyses including outcome data from 10,951 individuals. The strongest evidence was found for PUFAs (particularly as eicosapentaenoic acid) as an adjunctive treatment for depression. More nascent evidence suggested that PUFAs may also be beneficial for attention‐deficit/hyperactivity disorder, whereas there was no evidence for schizophrenia. Folate‐based supplements were widely researched as adjunctive treatments for depression and schizophrenia, with positive effects from RCTs of high‐dose methylfolate in major depressive disorder. There was emergent evidence for N‐acetylcysteine as a useful adjunctive treatment in mood disorders and schizophrenia. All nutrient supplements had good safety profiles, with no evidence of serious adverse effects or contraindications with psychiatric medications. In conclusion, clinicians should be informed of the nutrient supplements with established efficacy for certain conditions (such as eicosapentaenoic acid in depression), but also made aware of those currently lacking evidentiary support. Future research should aim to determine which individuals may benefit most from evidence‐based supplements, to further elucidate the underlying mechanisms.     

The impact of pharmacological and non‐pharmacological interventions to improve physical health outcomes in people with schizophrenia: a meta‐review of meta‐analyses of randomized controlled trials

We summarized and compared meta‐analyses of pharmacological and non‐pharmacological interventions targeting physical health outcomes among people with schizophrenia spectrum disorders. Major databases were searched until June 1, 2018. Of 3,709 search engine hits, 27 meta‐analyses were included, representing 128 meta‐analyzed trials and 47,231 study participants. While meta‐analyses were generally of adequate or high quality, meta‐analyzed studies were less so. The most effective weight reduction interventions were individual lifestyle counseling (standardized mean difference, SMD=–0.98) and exercise interventions (SMD=–0.96), followed by psychoeducation (SMD=–0.77), aripiprazole augmentation (SMD=–0.73), topiramate (SMD=–0.72), d‐fenfluramine (SMD=–0.54) and metformin (SMD=–0.53). Regarding waist circumference reduction, aripiprazole augmentation (SMD=–1.10) and topiramate (SMD=–0.69) demonstrated the best evidence, followed by dietary interventions (SMD=–0.39). Dietary interventions were the only to significantly improve (diastolic) blood pressure (SMD=–0.39). Switching from olanzapine to quetiapine or aripiprazole (SMD=–0.71) and metformin (SMD=–0.65) demonstrated best efficacy for reducing glucose levels, followed by glucagon‐like peptide‐1 receptor agonists (SMD=–0.39), dietary interventions (SMD=–0.37) and aripiprazole augmentation (SMD=–0.34), whereas insulin resistance improved the most with metformin (SMD=–0.75) and rosiglitazone (SMD=–0.44). Topiramate had the greatest efficacy for triglycerides (SMD=–0.68) and low‐density lipoprotein (LDL)‐cholesterol (SMD=–0.80), whereas metformin had the greatest beneficial effects on total cholesterol (SMD=–0.51) and high‐density lipoprotein (HDL)‐cholesterol (SMD=0.45). Lifestyle interventions yielded small effects for triglycerides, total cholesterol and LDL‐cholesterol (SMD=–0.35 to –0.37). Only exercise interventions increased exercise capacity (SMD=1.81). Despite frequent physical comorbidities and premature mortality mainly due to these increased physical health risks, the current evidence for pharmacological and non‐pharmacological interventions in people with schizophrenia to prevent and treat these conditions is still limited and more larger trials are urgently needed.     

Linking soil food web structure to above‐ and belowground ecosystem processes: a meta‐analysis

Soil fauna can be an important regulator of community parameters and ecosystem processes, but there have been few quantitative syntheses of the role of soil fauna in terrestrial soil communities and ecosystems. Here, we conducted a meta‐analysis to investigate the impacts of invertebrate soil micro‐ and mesofauna (grazers and predators) on plant productivity and microbial biomass. Overall our results indicate that an increase in the biomass of soil fauna increased aboveground plant productivity across ecosystems by 35% and decreased microbial biomass by 8%. In addition, we found no evidence for trophic cascades in terrestrial soil food webs, but the bacterivorous component of soil fauna influenced plant productivity and microbial biomass more than did the fungivorous component. Furthermore, changes in the biomass of soil fauna differentially affected plant productivity among plant functional groups: a higher biomass of soil fauna increased aboveground productivity by 70% in coniferous systems. However, in ecosystems dominated by legumes, a functional group with lower inorganic nitrogen requirements, there was no response of aboveground productivity to increases in the biomass of soil fauna. In sum, the results of this meta‐analysis indicate that soil fauna help to regulate ecosystem production, especially in nutrient‐limited ecosystems.     

Causes and consequences of variation in offspring body mass: meta‐analyses in birds and mammals

Early survival is highly variable and strongly influences observed population growth rates in most vertebrate populations. One of the major potential drivers of survival variation among juveniles is body mass. Heavy juveniles are better fed and have greater body reserves, and are thus assumed to survive better than light individuals. In spite of this, some studies have failed to detect an influence of body mass on offspring survival, questioning whether offspring body mass does indeed consistently influence juvenile survival, or whether this occurs in particular species/environments. Furthermore, the causes for variation in offspring mass are poorly understood, although maternal mass has often been reported to play a crucial role. To understand why offspring differ in body mass, and how this influences juvenile survival, we performed phylogenetically corrected meta‐analyses of both the relationship between offspring body mass and offspring survival in birds and mammals and the relationship between maternal mass and offspring mass in mammals. We found strong support for an overall positive effect of offspring body mass on survival, with a more pronounced influence in mammals than in birds. An increase of one standard deviation of body mass increased the odds of offspring survival by 71% in mammals and by 44% in birds. A cost of being too fat in birds in terms of flight performance might explain why body mass is a less reliable predictor of offspring survival in birds. We then looked for moderators explaining the among‐study differences reported in the intensity of this relationship. Surprisingly, sex did not influence the intensity of the offspring mass–survival relationship and phylogeny only accounted for a small proportion of observed variation in the intensity of that relationship. Among the potential factors that might affect the relationship between mass and survival in juveniles, only environmental conditions was influential in mammals. Offspring survival was most strongly influenced by body mass in captive populations and wild populations in the absence of predation. We also found support for the expected positive effect of maternal mass on offspring mass in mammals (r_pearson = 0.387). As body mass is a strong predictor of early survival, we expected heavier mothers to allocate more to their offspring, leading them to be heavier and so to have a higher survival. However, none of the potential factors we tested for variation in the maternal mass–offspring mass relationship had a detectable influence. Further studies should focus on linking these two relationships to determine whether a strong effect of offspring size on early survival is associated with a high correlation coefficient between maternal mass and offspring mass.     

Differential responses of size‐based functional groups to bottom–up and top–down perturbations in pelagic food webs: a meta‐analysis

We performed a meta‐analysis of 31 lake mesocosm experiments to investigate differences in the responses of pelagic food chains and food webs to nutrient enrichment and fish presence. Trophic levels were divided into size‐based functional groups (phytoplankton into highly edible and poorly edible algae, and zooplankton into small herbivores, large herbivores and omnivorous zooplankton) in the food webs. Our meta‐analysis shows that 1) nutrient enrichment has a positive effect on phytoplankton and zooplankton, while fish presence has a positive effect on phytoplankton and a negative effect on zooplankton in the food chains; 2) nutrient enrichment has a positive effect on highly edible algae and small herbivores, but no effect on poorly edible algae, large herbivores and omnivorous zooplankton in the food webs. Planktivorous fish have a positive effect on highly edible algae and small herbivores, a negative effect on large herbivores and omnivorous zooplankton, and no effect on poorly edible algae. Our meta‐analysis confirms that nutrient enrichment and planktivorous fish affect functional groups differentially within trophic levels, revealing important changes in the functioning of food webs. The analysis of fish effects shows the well‐described trophic cascade in the food chain and reveals two trophic cascades in the food web: one transmitted by large herbivores that benefit highly edible phytoplankton, and one transmitted by omnivorous zooplankton that benefit small herbivores. Comparison between the responses of food webs and simple food chains also shows consistent biomass compensation between functional groups within trophic levels.