Optimizing landscape-scale coastal monitoring and reporting through predicted versus observed animal abundance models

Abstract Effective environmental management hinges on monitoring drivers of change and effectively communicating results to stakeholders. While animals are valuable for engagement, few monitoring programs successfully integrate metrics quantifying their assemblages. We studied fish responses to environmental factors (including landscape context and water quality) in a 3-year survey across six ecosystems and 13 estuaries in eastern Australia (for >1800 fish surveys), and developed a novel predicted versus observed approach to monitoring, grading, and reporting on animal populations. Fish species richness and the abundance of five indicator species were explained significantly by at least one spatial attribute of sites (e.g. connectivity with mangroves and the ocean), and at least one water quality metric reflecting annual median water conditions (especially turbidity, dissolved oxygen (DO), and chlorophyll a concentration). For our grading approaches, predicted values were calculated for each replicate using best-fit models for each indicator, thereby accounting for natural spatiotemporal variation and standardizing site-to-site comparisons. Evaluating six methods for translating values into graded scores for each estuary, we recommend a simple metric: the percentage of sites with observed values above predictions. We discuss this approach as useful and complementary to programs with predominantly physical parameter monitoring, and discuss challenges in establishing ongoing protocols.