A framework for multiscale intertidal sandflat mapping: A case study in the Whangateau estuary

Abstract

We introduce a multiscale approach to population estimation of maldanid polychaetes on intertidal sandflats. Animals that live in marine sediments play important roles in ecosystem processes and are frequently the target of environmental monitoring but are notoriously time consuming to sample. Our goal was to enable the mapping of patterns of abundance of estuary lifeforms over long tracks at spatial scales and temporal frequencies not amenable by traditional quantitative methods. At each of four study sites of 9 m2, we used several forms of imagery at increasing resolution, obtained with a low-cost camera system and UAVs, to estimate the presence of the maldanid polychaete macroclymenella stewartensis. This involved defining the micro-scale topographic features on the sediment surface linked to the worms feeding activity. At each site, counts of polychate faecal mounds were estimated by quadrat counts. High resolution imagery was used to construct 3D depth maps using Structure from Motion, which were then segmented to obtain a local count of M.stewartensis faecal mounds. Texture features were then extracted from the high resolution imagery and used to create a linear model using the local counts. Low resolution imagery was then scaled, texture features extracted and passed into the linear model to provide count estimates with an error of 4% at best and 35% at worst for best orientation. The results demonstrated the potential of our multiscale texture based approach to model data at scale on sediment topographic features linked to the activity of functionally important species. These methods in turn help to advance ecological research and estimate ecosystem services potential as well as provide an additional tool for large scale rapid ecological assessment and estuarine management.