Modeling net primary productivity in a Cretaceous terrestrial ecosystem: a probabilistic approach

The Late Campanian Big Cedar Ridge (BCR) fossil site provides a well-preserved record of a rapidly buried ecosystem, offering a unique opportunity to reconstruct past environmental conditions. Understanding net primary productivity (NPP) in such ecosystems is crucial for assessing the energy available to sustain large herbivorous dinosaurs. However, quantitative estimates of ecosystem productivity in deep time remain rare, as most reconstructions rely on qualitative indicators or indirect climate proxies rather than mechanistic, data-driven approaches. Existing methods often lack probabilistic treatment of uncertainty and cannot easily be validated against modern ecosystems. Here, we present a probabilistic permutation of an existing carbon cycling model that estimates paleo-NPP using total organic carbon (TOC) from well-preserved paleosols. This approach integrates soil carbon stocks with decomposition rate estimates, providing a probabilistic approach to reconstructing ancient productivity. When tested in modern ecosystems, the model demonstrates strong predictive performance, particularly in high-TOC, oxygen-limited environments. Applied to BCR, the model estimates a mean NPP of approximately 1816 g C m −2 yr −1 , a value comparable to modern tropical and subtropical wetlands. These findings suggest that this Campanian ecosystem was highly productive, capable of sustaining diverse megaherbivore populations and demonstrate the utility of probabilistic NPP modeling for reconstructing the energy budgets of extinct ecosystems and constraining their potential herbivore carrying capacities. By establishing a baseline for estimating primary productivity in deep time, this methodology can provide new insights into the relationship between ecosystem productivity, dinosaur diversity, and paleoecological structure. The model offers a framework for future studies, enhancing our understanding of how primary productivity shaped and was shaped by terrestrial ecosystems during the Mesozoic. Unlike existing paleo-productivity proxies, this approach produces spatially explicit, probabilistic NPP estimates that can be directly validated against modern ecosystems. • Introduces a probabilistic soil‑carbon framework for reconstructing paleo-NPP from paleosols. • Validates soil-derived NPP estimates against MODIS across diverse modern ecosystems. • Demonstrates robust model performance in organic-rich, oxygen-limited environments. • Shows lower productivity in fern-dominated paleosols relative to angiosperm-rich sites. • Quantifies primary production in a Campanian floodplain supporting megaherbivore communities.