An image analysis method to study the relationship between seagrasses and their epiphytes

Seagrasses are globally threatened due to increasing environmental stressors in coastal ecosystems. Excessive accumulation of algal epiphytes is suggested to be harmful to seagrasses, but an understanding of the dynamics of epiphyte accumulation relative to host leaf growth and senescence is still limited, owing to poor spatiotemporal resolution of biomass measures. This study developed and validated color scanning and image analysis methods to characterize epiphyte accumulation relative to Thalassia testudinum seagrass host morphology. Spectral Angle Mapper (SAM) algorithms within the ENVI Program distinguished the pixels of many epiphytes collectively from epiphyte-free leaf areas based on each pixel’s spectral vector, derived from Red, Green, and Blue color channels (RGB). Classification accuracy was evaluated by the misclassified pixels and the correlation of biomass and morphology metrics vs. image-based metrics for more than 2,000 seagrass and epiphyte images. Many variations of four major color groups of epiphytes on the seagrass surface were identified. Misclassified pixels were < 10% in 94% of the analyzed images. Image-derived seagrass metrics strongly correlated with seagrass biomass and leaf area. Linear regressions of epiphyte biomass vs. epiphyte area (pixels), and of biomass ratio of epiphyte to seagrass vs. epiphyte coverage (epiphyte pixels/leaf pixels), were only moderate and suggestive of environmental influences. The robust validation result indicated that this image analysis method can provide insight into the spatiotemporal accumulation of epiphytes on the seagrass host and has the potential to be adapted to detect environmentally-induced changes in the community composition of epibiota.