| 17 Dec 2025
TREED (v1.0): a trait- and optimality-based eco-evolutionary vegetation model for the deep past and the present
Abstract. We present the TREED model (TRait Ecology and Evolution over Deep time), a trait- and optimality-based vegetation model to simulate vegetation structure, carbon cycling and eco-evolutionary adaptation dynamics to climate and CO2 changes across geologic time scales. The global grid-based vegetation model represents plant carbon allocation and trait evolution as a set of carbon economic trade-offs. Based on optimality principles, it is assumed that functional traits of the modelled community-representative average plants evolve towards an optimum that maximizes height growth while maintaining a positive carbon balance. The considered trait trade-offs resolve the potential plant height, leaf carbon pool size, leaf longevity, and phenology as the major axes of plant trait variation. Based on these key traits, whole-plant structure and functioning are derived using functional and allometric relationships. In its eco-evolutionary mode, vegetation-mediated carbon cycling can be tracked over the course of climatic transitions, testing the effects of the speed of evolutionary trait adaptation and dispersal dynamics. Moreover, with its generalized plant physiology, continuous trait space, and lack of pre-defined functional types, the model can be used to calculate metrics of biodiversity, including indices of the functional diversity and species richness potential. With a low computational demand, a flexible time stepping scheme and scalable adaptation parameters, TREED is intended to simulate biological and environmental transitions across time scales spanning from centuries to millions of years. Here, we present the underlying theory and model functions and evaluate model outputs against present-day observations. We show that the trait- and optimality-based approach captures major patterns in present-day vegetation-mediated carbon and water fluxes, biomass carbon storage, vegetation height, leaf traits, as well as the global distribution of plant biodiversity. Finally, we illustrate its application in the context of paleoclimate and palaeoecological research using the Paleocene-Eocene Thermal Maximum as a case study and show how eco-evolutionary adaptation dynamics of terrestrial ecosystems may strongly affect global carbon cycle dynamics during hyperthermal events. The TREED model is a step towards a more self-consistent and parameter-scarce representation of vegetation dynamics under environmental conditions that are fundamentally different from the present. In combination with geochemical and paleobotanical data, the model may help to better constrain the resilience of vegetation-mediated Earth system functions to perturbations in the geologic past and at present.
The TREED model is a vegetation model designed to simulate global vegetation dynamics and biodiversity potential outputting GPP, NPP, evapotranspiration, average plant height, below, and above ground biomass storage. These variables are simulated on an average grid point bases by applying eco-evolutionary optimality principles. The plants in the model adapt to maximize fitness under a steady state or transient climates. Evolving the plants to maximize their fitness lets TREED simulate average plant traits in the grid point without making assumptions with regards to the climate niche. This framework offers a method for understanding how biological systems reorganize during different climates.
The paper is well written and explains the technical parts of the paper in detail. The authors have taken the time to go over all the equations used in the model and describe how and why they are implemented. The paper can benefit from more Thurow evaluation to show where the limitations are, please see comments below. In addition, the authors have carefully provided a limited sensitivity study appropriate for this paper. I hope to see a follow up sensitivity study further exploring the trait evolutions of the plants in different paleo climates, but this would be too much for this paper. I have some minor comments and recommend this paper for publication:
Line 178: I think “ration” should be “ratio”
Table 1: would it be possible to add citations for the values (where applicable) or an indicator, something like “this study” for where you have made your own estimates. In addition, an indicator for the sensitivity level? I understand that this might be subjective at the moment and needs to be further investigated but as a potential user of this model that would help assess if I can/should/want/need to change this value or not.
I think “fumeroot” should be “fine root” and “Rubisco specifity” should be “Rubisco specificity”
Section 5.3: Please expand this analysis to include a more regional evaluation of where the model does well vs. where the model is lacking. In addition, the reasons given for the mismatched between the model and data are too brought and should be more mechanism specific where possible. For example, “Reduced height growth under high NPP levels indicates carbon turnover processes that are not currently represented in TREED and its height optimization.” Which processes are not represented and how does this effect the model? Is this region specific or a global problem? Do this for all three variables, NPP, GPP, and AET. This will help understand when the model is appropriate to use and the model limitations.
Figure 7: A contour plot will work better here showing the density of the points
Figure 8c: Please use a density plot here
Figure 8d: Try to see if a density plot works, it might not because of the
Figure 11: What alphas and dispersal rates are used? (Add numbers to slow, fast, and intermediate)
Section 7: It would be nice to add a section on model limitations. Several limitations are already stated in the appropriate sections however, I think the paper can benefit to list them again and address them in a more systematic way.
Section 9: Please add the GitHub tag (or release name) also in the text, this makes it easier to check out the specific version.