COS 238-6 - Precision of allometric coefficients used in popular dynamic food web models

Thursday, August 18, 2022

11:15 AM – 11:30 AM EDT

Location: 518A

Presenting Author(s)

Penelope Blyth

PhD Student
University of Sheffield, United Kingdom

Coauthor(s)

ED

Eva Delmas

Post doctoral research associate
Université du Québec en Outaouais, Canada
CG

Chris Griffiths

Researcher
Swedish University of Agricultural Sciences, Sweden
TJ

Thomas Frederick Johnson

University of Sheffield
Sheffield, United Kingdom
BS

Benno I. Simmons

University of Exeter, United Kingdom
AB

Andrew Beckerman

Professor
University of Sheffield, United Kingdom

Background/Question/Methods

Allometry is a foundational aspect of modern ecology - allowing the scaling of a diverse array of traits (e.g. biological rates) with body mass. A particular component of allometry - allometric coefficients (or normalisation constants) of metabolic rates with body mass - are widely used in ecology, but they were derived from taxonomically restricted data and developed using outdated statistical methods. Here, we introduce an updated allometric coefficient, derived in two stages: 1) We begin by compiling data on metabolic rates and body masses for all non-bacterial heterotrophs (e.g. including species ranging from amoeba to moose); 2) Using this data we then apply a newly developed model which captures measurement error in estimates of body mass and metabolic rates, as well as phylogenetic non-independence driven by trait similarity in closely related species. We then compare our updated allometric coefficients to those in the literature and discuss the significance of our findings.

Results/Conclusions

We extracted over 7500 metabolic rate and body mass data points for 1850 species from 1200 genera, possessing 14.7-fold more data than in the previously derived allometric coefficient calculations. This data compilation effort was particularly valuable for some of the most diverse taxonomic groups. For instance, we increased the number of data points 220-fold for ectothermic invertebrates. In isolation, this increase in data quantity increased the metabolic rate~body mass coefficient values substantially for ectothermic invertebrates and endothermic vertebrates but decreased it for ectothermic vertebrates. However, after also implementing our Bayesian phylogenetic mixed model which accounts for missing data through imputation, we found the allometric coefficients of ectothermic invertebrates and ectothermic vertebrates decreased, whilst endothermic vertebrates increased. Even very modest changes in these coefficients can have a substantial impact on down-stream analyses, emphasising the value of our results. However, our most substantial finding is that by accounting for phylogenetic relatedness and missing data, the estimates of uncertainty around the constants increased by an order of magnitude. While this raises some issues about the confidence in previous point estimates, these estimates of uncertainty pave the way to explore the sensitivity of population, dynamics and stability to these values.