Meta-analytic approaches and new effect sizes to account for treatment differences across studies in comparative physiology
Daniel W.A. Noble, Patrice Potter, Sammy Burke, Szymon M. Drobniak, Malgorzata Lagisz, Rose O’Dea, Shinichi Nakagawa
Meta-analysis is a tool that provides comparative physiologists with powerful, unbiased and quantitatively informed answers to topical research questions by synthesising effect sizes across disparate studies. Distilling published research results into standardised effect sizes that can be weighted by their sample size (i.e., inverse sampling variance) and compared across broad sets of research designs, study systems and species is its core objective and strength. Estimating overall effect sizes, and understanding what drives effect variability, provides opportunities to model how organisms will respond to pressing global challenges. Despite this ambition, research designs in comparative physiology can appear, at the outset, as being vastly different to each other (e.g., using different temperatures or treatment dosages). Differences in treatments across studies has led many to believe that meta-analysis is an exercise in comparing “apples with oranges”. Here, we dispel this myth by showing how standardised effects sizes can be used in conjunction with powerful multi-level meta-analytic models to both account for factors driving differences across studies and make them more comparable. In addition, we derive new effect size measures that provide comparative physiologists with a means to directly make effect sizes comparable without the need to resort to more complex statistical models that may be harder to interpret. Our ‘new’ effect sizes and corresponding sampling variances help physiologists deal with temperature and dosage differences across studies; allowing researchers to compare both mean differences (e.g., Q10 to compare mean differences in physiological rates over 10 degrees) and associated differences in variance (e.g, SQ10 for comparing variability in in physiological rates over 10 degrees) for physiological traits. The new effect sizes we propose, in combination with existing meta-analytic models, will pave the way for comparative physiologists to explore exciting new questions by making results from large-scale data sets from the literature more accessible and widely interpretable.
We searched for meta-analyses tackling comparative physiological questions that relied on data compiled from studies using animal model systems with more than a single species. We excluded plant meta-analyses because they were often focused on ecological (e.g., quantified changes in abundance and distribution of species) or agricultural questions (e.g., studies on soil microbial function, nitrogen and phosphorus capture, methane output and greenhouse gas emissions) that did not focus on organism physiological function across species. Meta-analyses needed to have collected physiological trait data from the literature including traits, such as hormones, thermal physiology (CTmax, CTmin, thermal preferences, thermal breadth, LT50, lethal thresholds), metabolic rates (routine, maximum, and minimum metabolic rate, aerobic scope), enzyme reaction rates and concentrations, oxidative stress and consequences (e.g., DNA damage, teleomere attrition), performance measurements (heart rates, sprint speed, swim speed, growth rates) and immune traits (cell function, antibodies, ect). Studies were also included if they measured reproductive allocation, survival / mortality or quantified scaling relationships with body size (e.g., metabolic scaling) across taxa. Studies on behavior were only included if they were explicitly interested in proximate physiological drivers of behavioral variation. Meta-analyses quantifying phenology, differences in body sizes between the sexes (i.e., SSD) and those focused on validating physiological methodology were excluded because the underlying question was not physiological in nature.
Scopus search (17 May 2021 - 212 hits): "( TITLE-ABS-KEY ( meta.analy* OR meta.regr* ) AND ISSN ( ""02698463"" OR ""00220949"" OR ""15407063"" OR ""15222152"" OR ""20511434"" OR ""2397334X"" OR ""1461023X"" OR ""13541013"" OR ""01401963"" OR ""00298549"" OR ""00301299"" OR ""2296701X"" OR ""1744957X"" OR ""1469185X"" OR ""20457758"" OR ""15735184"" OR ""1432136X"" OR ""2397334X"" OR ""09628452"" OR ""15738477"" ) ) AND ( LIMIT-TO ( DOCTYPE , ""ar"" ) OR LIMIT-TO ( DOCTYPE , ""re"" ) ) AND ( LIMIT-TO ( PUBYEAR , 2020 ) OR LIMIT-TO ( PUBYEAR , 2019 ) OR LIMIT-TO ( PUBYEAR , 2018 ) OR LIMIT-TO ( PUBYEAR , 2017 ) OR LIMIT-TO ( PUBYEAR , 2016 ) OR LIMIT-TO ( PUBYEAR , 2015 ) ) AND ( EXCLUDE ( EXACTKEYWORD , ""Ecosystem"" ) OR EXCLUDE ( EXACTKEYWORD , ""Biodiversity"" ) OR EXCLUDE ( EXACTKEYWORD , ""Soil"" ) OR EXCLUDE ( EXACTKEYWORD , ""Plant"" ) OR EXCLUDE ( EXACTKEYWORD , ""Biomass"" ) OR EXCLUDE ( EXACTKEYWORD , ""Plants"" ) OR EXCLUDE ( EXACTKEYWORD , ""Fertilizer"" ) OR EXCLUDE ( EXACTKEYWORD , ""Ecosystem Function"" ) OR EXCLUDE ( EXACTKEYWORD , ""Soil Microbiology"" ) OR EXCLUDE ( EXACTKEYWORD , ""Food Chain"" ) OR EXCLUDE ( EXACTKEYWORD , ""Ecosystem Response"" ) OR EXCLUDE ( EXACTKEYWORD , ""Plant Leaf"" ) OR EXCLUDE ( EXACTKEYWORD , ""Phytoplankton"" ) OR EXCLUDE ( EXACTKEYWORD , ""Soil Carbon"" ) OR EXCLUDE ( EXACTKEYWORD , ""Tree"" ) OR EXCLUDE ( EXACTKEYWORD , ""Crop Yield"" ) )
This search covers publications from 2015-2020 (6 years) in the following 20 journals: Functional Ecology Journal of Experimental Biology Integrative and Comparative Biology Physiological and Biochemical Zoology Conservation Physiology Nature Ecology and Evolution Ecology Letters Global Change Biology Journal of Arid Environments Oecologia Oikos Frontiers in Ecology and Evolution Biological Reviews Ecology and Evolution Reviews in Fish Biology and Fisheries Journal of Comparative Physiology B: Biochemical, Systemic, and Environmental Physiology Nature Ecology and Evolution Evolutionary Ecology Biology Letters Proceedings of the Royal Society B
Retrieved bibliometric records were screened by ML, DN with SN resolving conflicts. Full text screenig? - some papers may be removed at this stage as some abstracts are not very clear?
- Update Table S1 and S2 with new journal names.
- Update search numbers for PRISMA