Macrosystems ecology, the study of ecological systems at large spatial, temporal, or taxonomic extents, requires four interdependent components: networks of observational data in time and space; informatic systems for aggregating, curating, and delivering that data; mechanistic models of system processes; and statistical methods to make inferences and estimates of uncertainty from the data, models, and their fusion. In paleoecology, community curated data repositories (CCDRs) have arisen in response to the desire to do global-scale science from local-scale data, which requires the careful assembly of many individual site-level paleoecological time-series into larger networks. This talk describes the close partnership between two distinct efforts: the Neotoma Paleoecology Database (Neotoma, www.neotomadb.org), a CCDR that focuses on paleoecological data from the recent geological past, and the Paleoecological Observatory Network (PalEON, https://www3.nd.edu/~paleolab/paleonproject/), which brings together paleoecologists, ecological statisticians, and terrestrial ecosystem modelers in order to improve the model parameterization and simulation of slow forest processes, operating at timescales of decades to centuries. Key PalEON data sources include fossil pollen data from Neotoma, tree ring data from the International Tree Ring Databank, and settlement-era forest observations from the Public Land Survey (PLS). I describe recent advances on both fronts, current directions and challenges, and next steps.
Neotoma is designed around a principle of consolidated infrastructure and distributed governance. Its data model is facilitated by the many commonalities of paleoecological data, which usually consist of measurements of proxies in various geological archives by depth, from which we estimate time. Distributed scientific governance, represented by virtual Constituent Databases within Neotoma, accommodates the dispersal of paleoecological scientific expertise among taxonomic groups and proxies, time periods, regions, and questions of interest. Neotoma data are closely integrated with PalEON through scientific workflows that pull fossil pollen records from Neotoma, update age-depth models, and pass them to the STEPPS hierarchical Bayesian model to reconstruct with uncertainty past forest composition. Analyses indicate significant shifts in tree-climate relationships over the last two centuries due to historic land use and climate change. STEPPS reconstructions demonstrate subtle but significant trends in forest composition in the north-central US over the last two millennia, possibly in response to regional cooling. Current work is focusing on assimilating these STEPPS-based reconstructions with terrestrial ecosystem model simulations for the last millennium, driven by downscaled CMIP5 paleoclimatic simulations, in order to better estimate the state of latent variables such as carbon sequestration and the parameterization of slow processes in terrestrial ecosystem models.