Ecosystem services in connected catchment to coast ecosystems: Monitoring to detect emerging trends

There is an increasing need for long-term monitoring of ecosystems and their services to inform on-ground management and policy direction. The supply of many ecosystem services relies on connections that span multiple ecosystems. Monitoring the underlying condition of interconnected ecosystems, using established indicators, is therefore required to track effectiveness of past interventions and, ideally, identify impending change. Here we conduct performance testing of ecological indicators for a catchment-to-coast system with the aim of identifying the time-scales over which they respond to change. We chose a case-study of a coastal fishery in Northern Australia that exhibits strong catchment-to-coast connectivity, has long-term available data and is under threat from water resource development. We developed a novel approach to performance testing. Our model drew on state-space modelling to capture ecological dynamics, and structural equation modelling to capture covariation in indictors timeseries. We first quantified covariation among three established ecological indicators: pasture biomass, vegetation greenness and barramundi catch per unit effort. Covariation in the indicators was driven by river flow, with higher values of all indicators occurring in years with greater river flow. We then defined reference bounds for each indicator that accounted for natural variation in river flow. We predicted the emergence times for each indicator, as the time taken for each indicator to emerge from the background of natural variation. Emergence times quantified at 80% and higher confidence levels were >10 years in all cases. Past trends and current status of ecosystem service flows are often used by decision makers to directly inform near-term actions, particularly provisioning services (such as barramundi catch) due to their important contribution to regional economies. We found that the ecological indicators should be used to assess historical performance over decadal timespans, but not as short-term indicators of recent change. More generally, we offer an approach to performance testing of indicators. This approach could be useful for quantifying time-scales of ecosystem response in other systems where cross-ecosystem connections are important.