Life on the edge: Monitoring forest community ecotones in a changing climate Todd Lookingbill Department of Geography and the Environment University of Richmond Lookingbill Considerations of scale • Environmental Change Lookingbill & Rocca 2010. in Predictive Modeling in Landscape Ecology. • Ecological Process Lookingbill Climate change and scale Lookingbill Models of ecological response provide inferences at specific spatial scales Mountain monkeyflower, Mimulus tilingii Sebasky, Wu, & Lookingbill. 2010. Evolution. Lookingbill Models of ecological response provide inferences at specific spatial scales MaxEnt (Baldwin. 2009. Entropy) Sebasky, Wu, & Lookingbill. 2010. Evolution. Lookingbill Implications • Two challenges to predicting changes to species distributions with changes in climate: – Choice of sampling design – Choice of inferential model Landscape scale Ecotone scale Lookingbill H.J. Andrews Lookingbill H.J. Andrews Lookingbill Forest sampling – hybrid designs Stevens & Olsen. 2004. J. Amer. Statistic. Assoc. Philippi. 2005. Ecology. 100s-1000s m N = 175 plots Lookingbill Two Communities Western hemlock Douglas fir True firs Silver fir Noble fir Environmental covariates •Elevation (m) •Slope (o) •Taspect (radiation) •Terrain shape index •TCI (wetness) •Mean Litter Depth (cm) •Mean Soil Depth (cm) •St. Dev. Soil Depth •Clay (%) •Silt (%) •Sand (%) Lookingbill •pH •C (%) •N (%) •C:N •P (ug/g) •Ca (cmol(+)/kg) •Mg (cmol(+)/kg) •K (cmol(+)/kg) •Ac (cmol(+)/kg) •ECEC •BS (%) Elevation model Elevation<1203 Hemlock Elevation<1326 Fir TCI<60 Fir Hemlock Elevation < 1443 Fir CART analysis Lookingbill Mt. Hemlock Elevation-based model of current vegetation Lookingbill Temperature sampling design Fridley, 2009. J. Applied Meteor. & Climate 45 Portable Microloggers Stratified approach within watersheds Vanwalleghem & Meentemeyer, 2010. Ecosystems Temperature Tmax Tmean T = β0 - β1 Elevation + β2 ln(dstrm) + β3 Radiation + ε Lookingbill & Urban. 2003. Agric. Forest Meteor. Tmin Comparison of temperature models • Regression-based model is highly correlated with other HJA temperature models for mea and max • Poorly correlated for minimum temperatures Lookingbill Cross-validation Relative Error Temperature-based model 1 0.9 0.8 0.7 0.6 0.5 Proxy xerror Plant-relevant xerror 0.4 0 1 2 3 Number of Splits Lookingbill & Urban 2005. Can J. Forest Res. 4 Moderate change scenario 25 years Lookingbill Current 50 years Lookingbill Rapid change scenario 25 years (moderate) Lookingbill Current 25 years (rapid) Temperature inversion Lookingbill Evidence of temperature inversion Lookingbill Inversion projection Current 50 years Lookingbill Climate change and mountain ecotones review Drought 8 Grazing 8 Edaphic 7 Human disturb 6 Fire 5 Eolian 5 Other factors: Frost Glacial retreat Invasive species Pests and pathogens Lookingbill Ecotone monitoring – multistage sampling (Stohlgren and others) Lookingbill Ecotone monitoring 20m Lookingbill Step 1: Identifying spatial pattern Edge detection Camarero, Gutierrez, Fortin 2006. Global Ecol Biogeog. Step 1: Identifying spatial pattern Point pattern analysis Lookingbill Step 2: Relate species distributions to environmental patterns July.Soil.Moisture<30 Regression tree model of western hemlock seedling counts July.Soil.Moisture<19.5 10 21 57 Step 2: Relate species distributions to environmental patterns Noble fir Low radiation High radiation (t=14.95, d.f. = 391, p <0001) (Charlie Halpern) Lookingbill x Noble fir x Silver fir x Hemlock Depth (cm) Parting shot 100 50 0 0 30 60 90 120 Distance (m) Lookingbill 150 180 Summary and Conclusions • To capture the variety of ecological processes potentially affected by changes in climate, fine-grained data need to be collected over large spatial extents. » GRTS, adaptive clustering, stratified clusters, multi-stage • • • • Nusser, Breidt & Fuller. 1998. Design and estimation for investigating the dynamics of natural resources. Ecol. Applications Philippi. 2005. Adaptive cluster sampling for estimation within local populations of low-abundance plants. Ecol. Stevens & Olsen. 2004. Spatially balanced sampling of natural resources. J. Amer. Statistic. Assoc. Stohlgren, Owen & Lee. 2000. Monitoring shifts in plant diversity in response to climate change: a method for landscapes. Biodiversity and Conservation. Lookingbill Summary and Conclusions • These spatially rich data can be best leveraged by statistical techniques that (1) account for spatial autocorrelation in the data, (2) capture patterns across multiple spatial scales, and (3) explore relationships between variables at each spatial scale. » Mantel correllograms, regression kriging, PLNM, CART, wavelet analysis • • • • • Carl & Kuhn. 2008. Analyzing spatial ecological data using linear regression and wavelet analysis. Stochastic Environ. Research & Risk Assess Dormann. 2007. Effects of incorporating spatial autocorrelation into the analysis of species distribution data. Global Ecol & Biogeogr Goslee & Urban. 2007. The ecodist package for dissimilarity-based analysis of ecological data. J Statistic. Software Peres-Neto, Legendre, Dray & Borcard. 2006. Variation partitioning of species data matrices: Estimation and comparison of fractions. Ecology Wagner & Fortin. 2005. Spatial analysis of landscapes: Concepts and statistics. Ecology Lookingbill Acknowledgements Carly Vendegna-Rameriz, Jonathan Mayfield, Megan Sebasky Monique Rocca, Dean Urban Lookingbill