Retrieval of intertidal biofilm quantity and nutritional quality through field spectroscopy provides potential for regional mapping of shorebird food resources

Abstract:

Microbial biofilm communities, composed of bacteria, diatoms, protozoa and fungi, inhabit the surface of intertidal mudflats and can comprise a large proportion of shorebirds' diets. Given their major role in intertidal food webs, understanding biofilm distribution, quantity and nutritional value for shorebirds is of vital importance. In San Francisco Bay, California, the largest tidal wetland restoration project in the Western U.S., the South Bay Salt Pond Restoration Project (SBSPRP), will transform 6,110 hectares of former salt ponds to a mosaic of wetland habitat types. One of the project’s key uncertainties is whether the restoration will impact mudflat resources such as biofilm. To address this uncertainty, we are using a multi-scalar approach (1 meter plot to regional scale) to model biofilm parameters of quantity [chlorophyll-a (chl-a; mg/m²; indicator of biomass)] and quality [total lipids, total organic carbon (TOC) and glucose (mg/g)] with field spectroscopy and satellite remote sensing. In summer 2019, we collected in-situ mudflat and biofilm reflectance spectra concurrently with biofilm parameters using an ASD Inc. full spectrum portable spectroradiometer. We modeled chl-a using an existing hyperspectral optical model developed by Launeau and others of the chlorophyll absorption coefficient (α) at 673 nm that accounts for background reflectance. Relationships between field spectra and lipids, TOC and glucose were explored through partial least squares regression (PLS). Spectra exhibited low reflectance values at 673 nm, ranging from 7.4% to 12.3%. The Launeau optical model was successful in estimating chl-a. A regression of α with in-situ chl-a values produced an R² of 0.54 (n=16, p=0.001). PLS shows potential to estimate nutritional quality with biofilm spectra; PLS of lipids: R² = 0.64 with 4 components, PLS of TOC: R² = 0.64 with 1 component, and PLS of glucose: R² = 0.59 with 1 component (for all models, n=16). Given heterogeneity of biofilm across the mudflat surface, next steps will be to develop methods to scale biofilm parameter estimates with hyperspectral or multispectral imagery. Results will support managers’ need to map and quantify biofilm areas for shorebirds and understand the influence of restoration activities on biofilm resources.