Sensitivity of remote sensing indicators to the composition and dynamics of wetland plant communities
Wednesday, August 4, 2021
Link To Share This Presentation: https://cdmcd.co/gQ63mP
Sophie Taddeo, Negaunee Institute for Plant Conservation Science and Action, Chicago Botanic Garden, Glencoe, IL, Sophie Taddeo and Alexandra Touloupas, Plant Biology and Conservation, Northwestern University, Evanston, IL
Negaunee Institute for Plant Conservation Science and Action, Chicago Botanic Garden Glencoe, IL, USA
Background/Question/Methods Plant diversity is paramount to maintaining the functions, stability, and resilience of wetland ecosystems but is threatened by large-scale habitat loss and degradation. Developing low cost, robust, and scalable methods for the consistent monitoring of wetlands is key to managing their diversity and rapidly identifying where interventions are most needed. Remote sensing can provide such data at large scales and low cost and could therefore become central to national conservation strategies. To advance this capacity, we test the sensitivity of remote sensing indicators to heterogeneity and shifts in the plant composition of freshwater wetlands. We combine non-metric multidimensional scaling (NMDS) and clustering to identify broad patterns of species composition and dynamics across 180 wetland sites monitored since 1997 by the Illinois Natural History Survey’s Critical Trends Assessment Program. Using aerial images from the National Agriculture Imagery Program, we generate 17 texture metrics describing the degree of spectral heterogeneity in each of these sites and aerial image captured between 2004-2019. We then assess whether texture metrics can significantly separate the different plant associations identified in the dataset and detect changes in their species composition and dominance. Results/Conclusions Preliminary results suggest that texture metrics can signal shifts in plant dominance, growth forms, and diversity levels. Across our sample, wetland sites experiencing woody encroachment tend to see an increase in the heterogeneity of their spectral values, while the proliferation of aggressive introduced species generally results in their greater uniformity and correlation. A partial linear regression suggests that texture metrics can help estimate the degree of similarity in the species composition of reference and restored sites. These results show that texture metrics derived from open-source aerial images could complement field surveys to rapidly identify signs of ecosystem shifts and biological invasions. Texture metrics could also help assess whether restored sites are progressing towards plant composition and diversity targets.