Global Meta-Analysis Reveals Science’s Quiet Skew Toward Charismatic Species

For decades, ecologists have suspected that their science favors the familiar: birds over beetles, mammals over mosses, pandas over fungi. A sweeping new study has now put that intuition to the most rigorous test yet, and the results are both reassuring and quietly alarming. In the largest synthesis of its kind ever attempted, an international team led by Stefano Mammola of Italy’s National Research Council and the University of Helsinki pooled evidence from 78 studies spanning nearly three decades of research, distilling them into 905 standardized effect sizes that measure how unevenly scientific and public attention is distributed across the tree of life. The verdict: taxonomic bias is real, pervasive, and structural, but far more modest than many researchers assumed. More troubling, the analysis found that being threatened with extinction—the single factor that should matter most—does not reliably earn a species more attention.

The study, published in the journal Ambio, represents the first attempt to quantify the magnitude of taxonomic bias globally using consistent statistical methods. Previous work had documented bias in piecemeal fashion: entomologists describing ‘institutional vertebratism’ in funding decisions, botanists diagnosing ‘plant blindness’ in public perception, and conservation scientists coining ‘taxonomic chauvinism’ to explain why amphibians and reptiles attracted so little research. At the microbial scale, the problem is most extreme of all—one recent analysis found that half of all bacteriology papers focus on just ten species. But no one had systematically converted these scattered findings into a common currency to answer the fundamental question: how large is the distortion, really?

To build the dataset, the researchers searched the Web of Science database using an elaborate string of terms targeting studies of research bias, awareness disparity, and favoritism across all branches of life. An initial search in November 2023 returned 1,541 articles, which the team screened against strict eligibility criteria requiring quantitative measures of bias at the species level. Following the PRISMA framework for systematic reviews, they whittled the list down to 78 studies published between 1998 and 2025 across 54 journals, supplemented by a further 11 papers captured through automated alerts through the end of 2025. Two authors independently scored a sample of 100 studies, achieving near-perfect agreement—Cohen’s kappa of 0.89—a standard test of reproducibility in evidence synthesis.

The methodological challenge was formidable. Because the source studies used wildly different approaches—public surveys, bibliometric counts, Wikipedia page views, funding databases—the team converted every reported result into Pearson’s correlation coefficient, a universal metric of association ranging from −1 to 1. Chi-square statistics without standardized effect sizes were transformed using the phi coefficient or Cramér’s V, allowing every comparison to be placed on the same scale. The analysis then employed multilevel meta-analytic mixed-effects models in the R statistical environment, with publication identity included as a random effect to account for the non-independence of multiple estimates drawn from the same study. The researchers grouped the drivers of bias into five conceptually distinct categories: appearance, commonness, conservation priority, utilitarian value, and general taxonomic contrasts.

The headline finding is that positive biases—greater research attention, knowledge, funding, or interest toward particular species—were widespread across every study design, geographic region, and taxonomic scope examined. Yet their magnitude was generally small, with most effect sizes falling below r = 0.5. General taxonomic contrasts produced the strongest average effect (r = 0.40), confirming that plants and invertebrates systematically receive less attention than vertebrates. Aesthetic appearance (r = 0.18), commonness (r = 0.17), and utilitarian value (r = 0.15) all showed smaller but significant positive effects. Notably, the strongest single contrasts emerged precisely where neglect is most commonly alleged: studies comparing plants (r = 0.55) and invertebrates (r = 0.42) against animal controls reported the largest disparities.

Then came the result the authors describe as the most concerning of all. Conservation priority—whether a species is threatened, endangered, or data-deficient—was the only dimension of bias not consistently associated with positive effects across study designs, yielding an effect size indistinguishable from zero (r = 0.02). In plain terms, research effort, funding, and public interest are not reliably directed toward the species that need them most. Charisma, visibility, and familiarity continue to shape the allocation of attention more powerfully than vulnerability. Curiously, Oceania was the lone exception, but across the rest of the world the misalignment held firm. The implication is stark: emotional resonance and symbolism, not extinction risk, determine which organisms enter the political arena and which are left to disappear unnoticed.

The analysis also turned its statistical lens on the literature about bias itself, and found it wanting in familiar ways. Funnel plot asymmetry and a multilevel Egger regression revealed strong evidence of small-study effects: smaller investigations, often based on online surveys or classroom experiments, reported systematically larger effect sizes than their data-intensive counterparts (β = 1.88 ± 0.09, p < 0.001). Crucially, however, effect sizes showed no temporal trend over nearly three decades, and extrapolations for larger studies remained positive—suggesting the underlying signal is robust even if its magnitude is inflated in portions of the literature. There was, tellingly, a further irony: the meta-literature itself is taxonomically biased, with 72 percent of all estimates concerning animals, 68 percent involving vertebrates, and studies focused exclusively on fungi or microorganisms exceedingly rare. The same groups neglected by primary research are neglected by the research that studies neglect.

Geography offered one of the study’s few genuinely surprising insights. Despite enormous differences in regional biodiversity, research traditions, and conservation priorities, the magnitude of detected bias was remarkably uniform across continents—the same groups are overlooked everywhere. The authors note, however, that the Global South remains largely underrepresented in studies of taxonomic bias itself, meaning the current picture may still be incomplete. Within vertebrates, the team found that disparities tied to appearance were particularly pronounced: the gap between a resplendent quetzal and a drab sparrow appears wider than comparable divides among invertebrates or plants, hinting that charisma concentrates attention even within already-favored lineages.

The findings come with candid limitations. Effect sizes are distributed unevenly across taxa and regions—a classic example of the ‘streetlight effect,’ where researchers search where it is easiest to look—and birds and mammals benefit from richer distribution, genetic, and trait data that make studying their biases far more tractable. Categorizing heterogeneous studies into broad predictor groups required interpretive judgment despite explicit coding criteria, and the standard toolkit for detecting publication bias cannot capture the ‘file drawer problem,’ in which null results simply go unpublished. The authors also flag a recent large-scale analysis of roughly 14,600 conservation projects over 25 years, which reported stronger funding biases than the meta-analysis could capture but was too descriptive to yield standardized effect sizes—a reminder that the true picture for money may be darker than the statistics suggest.

So what should change? The authors argue that because the system is skewed rather than broken, no systemic revolution is needed—but deliberate, targeted action is. They propose redirecting conservation focus from individual species toward habitats and species assemblages, adopting positive discrimination mechanisms or dedicated funding schemes for neglected taxa, and encouraging journals to adopt ‘taxonomic inclusivity statements’ akin to existing diversity commitments. Above all, they contend, structural change depends on communication: mild but persistent distortions will remain socially inert if documented only in specialist journals. Just as research teams employ statisticians and data managers, they argue, institutions and funders should support trained science communicators as integral members of the research enterprise—because a biodiversity crisis we cannot see, in the species we ignore, is a crisis we are unlikely to solve.

Subject of Research: Global meta-analysis of taxonomic bias in biodiversity research, funding, and public attention across the tree of life

Article Title: The biodiversity we ignore: A global meta-analysis on taxonomic bias

Article References: Mammola, S., Fontaneto, D., Bellvert, A., Chialva, M., & Adamo, M. (2026). The biodiversity we ignore: A global meta-analysis on taxonomic bias. Ambio. https://doi.org/10.1007/s13280-026-02479-8

Image Credits: AI Generated

DOI: 10.1007/s13280-026-02479-8

Keywords: taxonomic bias, biodiversity, conservation, meta-analysis, charismatic species, plant blindness, institutional vertebratism, speciesism, extinction risk, research funding, invertebrates, fungi