Caroline Stromberg

Summary: Pooideae, the largest grass subfamily, radiated in the Northern Hemisphere from the Paleocene onward and today dominates cold‐season and high‐altitude grasslands. Fossil phytolith assemblages suggest pooid‐rich ecosystems during the Oligocene–Miocene, but the lack of quantitative phytolith identification has limited reconstruction of early grassland composition and ecology. We combined 3D geometric morphometrics of grass silica short‐cell phytoliths (GSSCPs) with machine‐learning classification and phylogenetically informed climate modelling. A dataset of modern GSSCPs was used to classify Eocene–Miocene phytoliths from Türkiye and the US Great Plains and reconstruct paleoclimate. 3D GSSCP morphology distinguished modern grass subfamilies and tribes. Fossil assemblages contained mainly phytoliths assigned to Bambusoideae and Pooideae (including Poeae I and II). In modern grasses, Crenate and Polylobate phytolith dimensions correlated with temperature and precipitation. Fossil Crenates and Polylobates yielded climate estimates warmer and wetter than today. Ancestral state reconstructions indicated milder winter conditions at early Pooideae nodes, with cold tolerance arising early. Major temperate grass lineages (Poeae I and II) were established earlier than previously recognized, and Pooideae‐rich grasslands were present by the Oligo‐Miocene under warmer, wetter climates than today. These findings demonstrate the value of 3D phytolith morphometrics for interpreting early grassland evolution.

The Miocene Epoch has emerged as an important paleoclimate model intercomparison target due to its dynamic nature, global warmth, and reconstructed atmospheric CO₂ concentrations. Miocene CO₂ estimates range from as low as 300–600 ppmv to as high as 800–1100 ppmv. The formalized Miocene Model Intercomparison Project (MioMIP2) aims to address questions about the mechanisms driving the climatic changes and tipping points seen in the geological record and improve confidence in the ability of climate models to predict future climates and constrain climate sensitivity. Through alignment with PlioMIP and DeepMIP-Eocene efforts, a coordinated MioMIP (DeepMIP-Miocene) effort will allow for cross-time-period studies that sample the full range of Shared Socioeconomic Pathway paleo analogues. Building on the MioMIP1 “MIP of opportunity”, the Miocene community has been developing the experimental design for MioMIP2. The MioMIP1 effort was an ensemble of opportunity that leveraged existing Miocene simulations run with different models, boundary conditions, and reference datasets. MioMIP2 will be the first formal Miocene MIP with common boundary conditions and will target the Miocene Climatic Optimum (MCO; 16.75–14.5 Ma). We present the experimental design for coordinated “CORE” MCO simulations and sensitivity studies. This community effort synthesizes paleogeography, ice sheet, CO₂, and vegetation boundary conditions based on the latest available literature and will use geological data from companion proxy synthesis efforts to evaluate the simulations.

The Miocene Epoch has emerged as an important paleoclimate model intercomparison target due to its dynamic nature, global warmth, and reconstructed atmospheric CO₂ concentrations. Miocene CO₂ estimates range from as low as 300–600 ppmv to as high as 800–1100 ppmv. The formalized Miocene Model Intercomparison Project (MioMIP2) aims to address questions about the mechanisms driving the climatic changes and tipping points seen in the geological record and improve confidence in the ability of climate models to predict future climates and constrain climate sensitivity. Through alignment with PlioMIP and DeepMIP-Eocene efforts, a coordinated MioMIP (DeepMIP-Miocene) effort will allow for cross-time-period studies that sample the full range of Shared Socioeconomic Pathway paleo analogues. Building on the MioMIP1 “MIP of opportunity”, the Miocene community has been developing the experimental design for MioMIP2. The MioMIP1 effort was an ensemble of opportunity that leveraged existing Miocene simulations run with different models, boundary conditions, and reference datasets. MioMIP2 will be the first formal Miocene MIP with common boundary conditions and will target the Miocene Climatic Optimum (MCO; 16.75–14.5 Ma). We present the experimental design for coordinated “CORE” MCO simulations and sensitivity studies. This community effort synthesizes paleogeography, ice sheet, CO₂, and vegetation boundary conditions based on the latest available literature and will use geological data from companion proxy synthesis efforts to evaluate the simulations.

Like many of Earth’s subtropical open ecosystems, the African savanna originated in the Miocene Epoch (23.0 to 5.3 Ma), one of the best historical analogs for future climate states. The first C ₄ grass, C ₃ tree, and shrub savanna developed in the subtropics after ~10 Ma in northwest Africa, but the terrestrial sedimentary record of the region is not well preserved, making it difficult to determine when, how, and why the African landscape opened. Here, we leverage plant-wax biomarkers extracted from the marine sedimentary archive to document the origination and history of open ecosystems through Miocene northwest Africa. We show that a C ₃ grass-rich savanna with no modern analog replaced closed woodlands and forests during the middle Miocene (~15 to 14 Ma), a period of global cooling and aridification. This nonanalog ecosystem transitioned into the C ₄ savanna in stages in the late Miocene (most rapidly between 7.4 to 6.4 Ma) as global temperatures cooled and aridity increased, roughly coincident with several C ₄ expansions in other subtropical regions. Our new vegetation history, put in the context of other records, links initial ecosystem opening and C ₄ expansion to periods of global cooling and drying. It is consistent with a nonlinear system of feedbacks between climate, consumers (herbivory and fire), and plant-trait interactions that allows small changes in one parameter to have dramatic effects on regional vegetation structure and composition. The geologic record indicates that the response of subtropical open ecosystems to future climate perturbations will depend on the nature and strength of these feedbacks.

The Miocene Epoch has emerged as an important paleoclimate model intercomparison target due to its dynamic nature, global warmth, and reconstructed atmospheric CO2 concentrations. Miocene CO2 estimates range from as low as 300–600 ppmv to as high as 800–1100 ppmv. The formalized Miocene Model Intercomparison Project (MioMIP2) aims to address questions about the mechanisms driving the climatic changes and tipping points seen in the geological record and improve confidence in the ability of climate models to predict future climates and constrain climate sensitivity. Through alignment with PlioMIP and DeepMIP-Eocene efforts, a coordinated MioMIP (DeepMIP-Miocene) effort will allow for cross-time-period studies that sample the full range of Shared Socioeconomic Pathway paleo analogues. Building on the MioMIP1 “MIP of opportunity”, the Miocene community has been developing the experimental design for MioMIP2. The MioMIP1 effort was an ensemble of opportunity that leveraged existing Miocene simulations run with different models, boundary conditions, and reference datasets. MioMIP2 will be the first formal Miocene MIP with common boundary conditions and will target the Miocene Climatic Optimum (MCO; 16.75–14.5 Ma). We present the experimental design for coordinated “CORE” MCO simulations and sensitivity studies. This community effort synthesizes paleogeography, ice sheet, CO2, and vegetation boundary conditions based on the latest available literature and will use geological data from companion proxy synthesis efforts to evaluate the simulations.

The Trans-Amazon Drilling Project (TADP) is reconstructing the late Cenozoic history of Amazonian geology, climate, rivers, and forests. Drilling in the Acre Basin of western Brazil in 2023 recovered an 860 m drill core characterized by sediments that were deposited in a large paleo-river system. The overall sequence includes sandstones, siltstones, and mudstones that underwent varied degrees of weathering and pedogenesis. Here, we describe the ongoing geochronologic, geochemical, mineralogical, geophysical, and biotic analyses of the sedimentary record and present some preliminary inferences of the environmental history based on these initial results. Except for the uppermost ∼ 12 m, sediments from the drill core represent a single lithostratigraphic unit, assigned to the Solimões Formation, which is dominated by feldspar-rich sands of Andean origin. The pollen assemblage is quite different from Early to Middle Miocene floras that have been analyzed from a few sites elsewhere in the western Amazon. The novel pollen assemblage and new geochronological results suggest that the sequence may span the latest Miocene and all of the Pliocene Epoch, an interval that currently is not well represented in existing regional records and that is crucial for understanding the evolution of Amazonian biodiversity, as well as landscape transformations driven by Andean uplift and global climate change.

The Nyakach Formation of western Kenya provides a rare Middle Miocene (ca. 14-16 Ma.) terrestrial record bearing the monkey, Victoriapithecus macinnesi, and associated vertebrate faunas. Here we integrate sedimentology, clay mineralogy, paleosol geochemistry, phytoliths, and δ13C of soil organic matter to refine the age, depositional setting, and paleoecology of the Nyakach Formation. Stratigraphic and lithological evidence indicates subaerial deposition within a rift-margin fluvial system influenced by episodic volcanism, with eastern and western sites reflecting multiple sediment sources and variable degrees of reworking. Smectite-rich, base-saturated paleosols with high CaO concentrations point to carbonatitic parent material and a seasonally variable, but overall humid hydroclimates. Proxy-based precipitation estimates from paleosol geochemistry yield subhumid–humid conditions (ca. 1100–1400 mm/yr), which are substantially wetter than earlier reconstructions, but similar to consensus values from the Miocene Model Intercomparison project (MioMIP1). Phytolith assemblages dominated by forest-indicator morphotypes, together with δ13C value indicating predominantly C3 vegetation and only limited C4 biomass, support the reconstruction of a warm, tropical, seasonally wet forest with localized grassy openings. These findings highlight a heterogeneous but forest-dominated ecosystem during the Langhian stage, shaped by volcanism, tectonism, and fluvial dynamics, and provide refined ecological context for Middle Miocene primates and faunas inhabiting the Nyanza Rift region.

Climate models require boundary condition information, such as vegetation and soil distributions because they influence the mean state climate, and feedbacks can significantly influence regional climate and climate sensitivity to CO ₂ forcing. Information about past distributions comes primarily from the paleobotanical record, which is often supplemented by a vegetation model to fill data gaps. For recent past periods such as the Pliocene, a quantitative suitability assessment of these vegetation model simulations is sufficient. However, the Miocene Climate Optimum spanning 16.9–14.7 Ma was the warmest period on Earth over the last ∼25 million years and models struggle to reproduce those conditions for the range of paleogeographies and CO ₂ concentrations tested, particularly at high latitudes. Here we bring together the Miocene modeling and data communities to update previous vegetation reconstructions used for climate modeling with a new regional approach that relaxes the requirement for a single model simulation to be used, blending instead simulations forced by different paleogeographies and CO ₂ concentrations. This ensures the simulated vegetation is first, and foremost, consistent with the paleorecord and provides a baseline for future comparisons. The reconstruction shows global increases in forest cover at all latitudes as compared to today and extensive C ₃ grasslands across the high northern latitudes. Data gaps at high latitudes are filled with vegetation models forced by higher CO ₂ concentrations than were required at lower latitudes consistent with the inability of current models to simulate Miocene high latitude warmth.

Premise: Grasses support critically important ecosystems, occupy a large portion of Earth's land surface, and provide essential resources, including maize, other cereal crops, and bamboo. Their radiation and global spread have been documented primarily by microfossils and paleosol isotopic studies, whereas reconstruction of phylogenetic relationships has relied primarily on living species. However, hypothetical dates associated with evolutionary events can be tested only by fossils. We report new, dated Pharoideae phytoliths from Ethiopia (21.73 Ma), pushing back the first occurrence of that subfamily in the Old World by 10 million years. We also formally describe a rare Pharoideae leaf impression reported from Kenya (12.6 Ma). We discuss the relevance of the fossils to phylogenetics, paleobiogeography, and paleoecology.Methods: Phytoliths and the macrofossil were prepared using standard techniques. Fossils were identified by comparison with online, herbarium, and research collections.Results: The phytoliths possess characters exclusive to living Pharoideae. The fossil leaf compares favorably to one species in each of the two living Paleotropical Pharoideae genera, Leptaspis and Scrotochloa. Thus, the subfamily was present in Africa by the early Miocene, improving on previous middle Miocene estimates used to date the Neotropical‐Paleotropical crown group split. The fossil sites were forests, indicating that Pharoideae occupied the same niche as today; the broad leaf of the macrofossil is consistent with that of living C₃ forest grasses.Conclusions: Early Miocene (21.73 Ma) Pharoideae phytoliths from Ethiopia provide a new, earlier, first occurrence datum for Poaceae trees. The subfamily occupied a forest niche as it does today.