Forest sites- Since 2008, we have been conducting large-scale monitoring and experimental work involving the collection of demographic data pertinent to understanding tree species recruitment along regional (two latitudes in the Great Lakes region, Michigan), landscape (different levels of fragmentation, isolation and urbanization at the southern latitude) and site-level gradients (edaphic conditions and levels of canopy openness). These forest stands are distributed along the latitudinal gradient of Michigan and include the main forest types found in the entire Great Lakes region: temperate deciduous, northern hardwoods and coniferous biomes. The combination of varying climate, physiography, and soil give rise to the current distribution of forest ecosystems in the region. Through our experimental approach, we have capitalized on this heterogeneous landscape to evaluate species responses to a wide range of conditions. In this region, microclimate variability in a particular site (~7-8C) can be larger than the average forecasted change due to global warming (~3-5C), and the variety of soils at any location, from outwash sands to lacustrine clays, provides a wide range of soil moisture conditions and nutrient levels. Plant performance along those gradients have provided us insight into tree species recruitment performance under climate change while still accounting for other drivers of change (e.g., fragmentation, urbanization, invasive species). However, this information has been collected only for the life-stages involved in recruitment, i.e., seed production, seedling establishment, and seedling growth and survival. We lack information on tree species performance from the sapling stages through adulthood and senescence. To fulfill this essential data gap we are proposing to set up a network of Forest Inventory Plots at the forest sites were we have extensive information on recruitment. We will establish 1 ha forest plots at eight forest sites, in two latitudes; these are forest sites that represent the major forest types in the Great Lakes region. The selected forest sites share a large number of species making it possible to develop comparisons within and between species across a large area and thus inform about their performance under a wide array of environmental conditions. Also, some of these forest sites have been fully or partially surveyed, data that we will take into account in our inventories. Forest plots and inventories-The size of forest plots varies according to the type of forest to be sampled. Tropical forests required several hectares to be able to account for the large levels of diversity and the low abundance of most species. Temperate forests, like ours, can be fully surveyed (i.e., individuals of all species and sizes are sampled in sufficient numbers) in much smaller plots. Thus, we will establish 1 ha (100 x 100m) stands in each of the selected forest sites. At each forest plot we will set up a 10x10 m grid marked by pvc poles at each intersection delimiting 100 m2 quadrats what will be numbered to facilitate the tree census. Elevation at each node of the grid will also be recorded to generate high-resolution topographic maps of the plots. The surveying and GPS equipment necessary is already available (Dick ESGR and Ibáñez lab). In each hectare plot we will tag (with unique numbered tags), identify and measured (diameter at breast height, dbh, 1.37 m) all trees with at least a dbh of 1 cm. Tags will be placed at the point-of-measured (POM, 1.37 m) for consistent measurements in further censuses, for small trees POM will be marked with tree paint. Trees with several stems at 1.37 cm height will be marked accordingly and measured. Location in the plot of each tree will also be recorded. In further censuses we will also include a 'status' field in the data set (alive, dead, damage). Field teams will be trained each year to ensure proper and consistent data gathering. Data analysis-The demographic data collected across the network of forest plots will be the basis for the PhD student dissertation work. Our research group have developed models following several analytical frameworks for this type of data, e.g., including antecedent effects used to assess lag effects of both intrinsic and extrinsic variables; accounting for spatial effects to explain spatial correlation; we have also developed hierarchical models which allow to link scales (individual trees, sites, landscapes, and regions), and make inferences about species performance at each of these scales and as a function of the many covariates we consider to affect natural processes. These analyses provide parameter estimates at each level, and associated estimates of their uncertainty, and, in addition, quantify the sources of variability associated with each level (or scale) in the hierarchical analysis. We will explore hierarchical estimations of the parameters (P) involved in demographic performances (Pspecies Pregion Planscape Psite) to understand how the covariates (biotic and abiotic) may affect tree performance along the latitudinal gradient, the local landscapes, and a variety of forest sites. This approach will allow us to make inferences about forest dynamics under a large set of environmental conditions, including those predicted for the region. Data availability- The stem maps and demographic tree data will be made available to the UofM research community and collaborating scientists, managers and policy-makers interested in the region forest dynamics. We will work with the ESGR, UMBS, and possible SNRE and/or MBGNA to create the online infrastructure for data sharing and curating of the data. Data usage will require acknowldegement of funding sources.