Abstract Quantifying vegetation biomass is essential for understanding the structure and energy balance of terrestrial ecosystems, yet robust estimates for deep-time ecosystems remain elusive. Most palaeovegetation proxies provide qualitative or relative signals and lack a direct connection to ecosystem-scale energetics. Here, we present a probabilistic framework to reconstruct ancient vegetation biomass using fossil insect assemblages, treating insect biomass as a primary energy-based state variable. We estimated individual insect body mass from fossil morphological traits of insect fossils using relationships calibrated with modern insects, and inferred population densities based on metabolic scaling theory. These estimates were integrated to reconstruct fossil insect community biomass while explicitly accounting for sampling bias. Herbivorous insect biomass was then translated into vegetation biomass through a trophic transfer model incorporating realistic ranges of feeding rates, trophic efficiencies, and biomass accumulation ratios. All uncertainties were implemented by Monte Carlo simulations to generate full probability distributions of vegetation biomass. Sensitivity analyses reveal that insect biomass is the dominant contributor to variance in reconstructed vegetation biomass, with standardized regression coefficients substantially exceeding those of trophic transfer parameters. Ecological parameters primarily modulate vegetation biomass within constrained bounds but do not determine its order of magnitude. Comparisons with modern global vegetation datasets indicate that the reconstructed biomass corresponds to productive terrestrial ecosystems with identifiable modern biome analogues. Our results demonstrate that fossil insect assemblages encode fundamental energy-based constraints of ancient ecosystems and provide a quantitative pathway for reconstructing palaeovegetation biomass.
Chen et al. (Thu,) studied this question.