Integrated assessment of river development on downstream marine fisheries and ecosystems

The paper addresses how upstream water resource developments (dams and water extraction for agriculture) alter natural river flows and connectivity, and how these changes propagate to estuarine and marine ecosystems and dependent fisheries, particularly in tropical systems where quantitative assessments are scarce. The context emphasizes global modification of the terrestrial water cycle and the paucity of models that quantify downstream marine impacts of altered flows, especially at inter-catchment and basin scales. The purpose is to proactively assess potential downstream ecological, biodiversity, and livelihood impacts of alternative WRD scenarios, prior to infrastructure decisions, and to evaluate trade-offs and mitigation options (for example, pumping thresholds). The study focuses on Australia’s Gulf of Carpentaria (GoC), comprising largely unregulated rivers (Mitchell, Gilbert, Flinders), emblematic of remaining free-flowing systems under development pressure. The research questions include: (1) To what extent do WRDs alter downstream species, habitats, and fishery catches? (2) How do species-specific eco-hydrological responses and cross-catchment connectivity shape outcomes? (3) What WRD settings (e.g., allocation volumes, thresholds, dam portfolios) minimize ecological and fishery risks?

The authors synthesize evidence that dams and hydropower developments have well-documented negative impacts on downstream ecosystems, biodiversity, and fisheries, and that non-strategic dam-by-dam planning sacrifices ecosystem services. Past work has called for basin-scale, multi-objective planning to balance energy and environmental goals, yet most dam-planning processes overlook freshwater needs of marine ecosystems and downstream fisheries. Studies highlight insufficient compensation for downstream social and environmental impacts, a lack of transparency, and systematic overestimation of economic benefits with underestimation of biodiversity and fishery costs. Reviews underscore the importance of river flows for temperate and tropical systems, including impacts on commercial fisheries and threatened species. Prior research in northern Australia linked barramundi and mud crab growth/catches to rainfall/flows and suggested lower threshold flows below which ecological responses become nonlinear, and upper limits at very high flows. The concept of a portfolio effect among rivers stabilizing fisheries (e.g., Bristol Bay sockeye) provides a theoretical basis for cross-catchment connectivity relevant to GoC prawns. Calls for inclusive assessments incorporating Indigenous values and cultural flows are emphasized as a gap in current water planning.

Study area: Australia’s Gulf of Carpentaria comprising eight spatial regions influenced by major catchments (Embley, Mitchell, Gilbert, Norman, Flinders, border region, Roper, Walker). The analysis focuses on three large rivers with scoped WRD options: Mitchell, Flinders, Gilbert. Hydrology: End-of-system flow outputs from river system models (natural flows from 1900; weekly or monthly series) provide flow inputs at the most downstream node. WRD scenarios: Nineteen hypothetical scenarios altering flows via combinations of dam placements and water extraction allocations, pump thresholds (flow level to initiate pumping), and pump durations/rates. Four illustrative scenarios are emphasized: WRD 1 (high allocations; Mitchell and Flinders extraction; two dams on Gilbert), WRD 2 (moderate allocations; one dam on Gilbert), WRD 3 (moderate extraction on Mitchell only), WRD 4 (moderate extraction on Flinders; one dam on Gilbert; no Mitchell development). Modelling approach: An integrated, spatial, multispecies Model of Intermediate Complexity for Ecosystem assessment (MICE) was developed and fitted to long-term fishery and environmental data. Species/groups explicitly represented: common banana prawns (Penaeus merguiensis), barramundi (Lates calcarifer), giant mud crabs (Scylla serrata), largetooth sawfish (Pristis pristis), and habitat groups (mangroves, seagrass; plus supporting groups such as microphytobenthos and meiofauna). Temporal resolution: weekly time steps for prawns; monthly for barramundi, mud crabs, and sawfish. Eco-hydrological relationships: Flow influences recruitment, survival, catchability/availability via functional forms (logistic or parabolic), with lower threshold effects and upper limits. Weekly/monthly flow totals standardized relative to long-term averages. For prawns, a cross-catchment contribution framework estimated relative influence of adjacent rivers (Mitchell, Gilbert, Norman, Flinders) on regional recruitment and catches to detect a river portfolio effect; an ensemble version assumed no cross-catchment connectivity to bound uncertainty. Sawfish recruitment modeled as boom-bust dependent on flow, with additional mortality when flows drop below thresholds (reflecting refuge pool stressors). Habitats: Seagrass growth linked inversely to light attenuation related to flows (sediment/runoff); mangrove growth rate and carrying capacity linked to standardized flow and average annual flow (encroachment/die-back dynamics). Flood productivity: One ensemble version included a flood-induced productivity boost where scoured microphytobenthos during large floods reduces prawn natural mortality (proxy for nutrient-fueled productivity), testing improved fit to historical prawn catches. Model fitting and ensemble: Implemented in AD Model Builder with maximum likelihood, using extensive fishery catch-effort time series (prawns: ~50 years weekly; barramundi and mud crabs: ~30 years monthly) across eight regions. Diagnostics, alternative structures/parameters, and an ensemble of five model versions captured structural and parameter uncertainty; R used for analyses and plotting. Scenario evaluation: Best-fitting eco-hydrological parameters fixed while re-running the MICE under WRD-altered flow series to project counterfactual abundance and catches relative to baseline flows. Risk assessment: Population and fishery risks defined from average declines in abundance and catches under WRDs relative to baseline, mapped to categorical risk levels. Economic risk for the Northern Prawn Fishery (banana prawns) quantified as probabilities of ‘bad’ years (≤2,000 t), successive bad years, and extended sequences rendering operations unviable. Stakeholder engagement: Multi-stakeholder workshops (2019–2021) informed conceptual model, focal species, and assumptions.

• Incorporating flow variability significantly improved model fits to historical fishery catches compared with effort-only models, confirming strong flow–ecosystem linkages. - Cross-catchment river portfolio effect: The model estimated significant contributions from multiple adjacent rivers (Mitchell, Gilbert, Norman, Flinders) to prawn recruitment and catches in the south-east GoC; WRDs on multiple rivers produced cumulative and synergistic (non-additive) negative effects on prawns. - Extraction thresholds matter: Pumping at low river-flow thresholds caused substantially larger negative impacts on catches and abundance than confining extraction to higher flows; short-duration extraction during peak flows can mitigate impacts. - Species- and catchment-specific outcomes: • Common banana prawn biomass and catches predicted to decline by approximately 4% to 40% depending on extraction extent and inclusion of flood-productivity effects. • Barramundi biomass and catches declined by 4%–61% under WRD scenarios 1–4, with relatively larger effects on catches in some cases. • Mud crabs showed substantial sensitivity in most non-perennial rivers, with catch declines up to 83% in some years; perennial systems (e.g., Mitchell) less affected. • Largetooth sawfish, though data-limited, consistently emerged as the most sensitive taxon, with large declines across scenarios due to flow-dependent recruitment and heightened low-flow mortality. • Mangroves declined substantially under affected catchments. • Seagrass responses were small, with slight increases under some WRDs (up to ~7%) and declines up to ~9% in most scenarios. - WRD scenario risk ranking: WRD 1 (highest allocation; multi-catchment development) posed the highest ecological and fishery risks (moderate to intolerable for most components except seagrass), followed by WRD 2 and WRD 4; WRD 3 (Mitchell-only development) was least risky. - Economic risk: For the banana prawn fishery, the probability of an uneconomic ‘bad’ catch year more than doubled under some WRDs. - Floods as productivity drivers: The best-fitting prawn model included a flood-induced productivity effect, highlighting the critical ecological role of flood-driven nutrient inputs; trapping of sediments by dams may further alter estuarine productivity and geomorphology (beyond scope).

The study demonstrates that altering river flows through WRDs translates into substantial downstream impacts on estuarine and marine ecosystems and fisheries, with outcomes strongly mediated by species-specific eco-hydrological responses, catchment characteristics, and cross-catchment connectivity. By fitting a tractable ecosystem model to long time series, the authors quantified the contributions of multiple rivers to sustaining prawn populations, confirming a stabilizing river portfolio effect and revealing that multi-catchment developments can produce synergistic declines that exceed simple additive expectations. The findings directly address the research questions by showing: (1) significant, often large, declines in key species and habitats under plausible WRD settings; (2) the importance of operational settings, particularly pumping thresholds, as mitigation levers; and (3) the value of coordinated, basin-scale planning that incorporates downstream marine ecosystems into WRD decisions. The work underscores the necessity of integrating freshwater and marine management, recognizing floods as crucial for productivity, and using risk-based metrics to identify unsustainable development combinations. These insights are relevant globally for planning in remaining free-flowing rivers and for balancing agricultural water use with biodiversity, fisheries, and cultural values.

This study provides a proactive, quantitative framework that couples river system models with a spatial multispecies MICE to assess downstream ecological and fishery impacts of WRDs. Key contributions include: (1) empirical quantification of flow-dependent responses across taxa and habitats; (2) demonstration of a cross-catchment river portfolio effect for prawns; (3) identification of WRD settings (notably pumping thresholds) that substantially influence risk; and (4) translation of outcomes into ecological and economic risk metrics to inform decision-making. Policy-relevant conclusions are that multi-catchment, high-allocation developments (WRD 1) pose unacceptable risks, while single-catchment, moderate developments with high flow thresholds (e.g., WRD 3) are less risky but still impactful. Future research should incorporate additional mechanisms (sediment/nutrient dynamics, migration barriers, reservoir residence effects), climate-change interactions, and environmental flow rules; refine cross-catchment and oceanographic connectivity; expand taxa coverage; and embed Indigenous values and cultural flows into basin-scale planning.

• Model scope and structure: The MICE focuses on first-order flow–ecology linkages and does not explicitly model full oceanographic processes, wind-driven dynamics, or detailed trophic networks; some predator–prey effects were simplified via proxy relationships. - Data limitations: Sawfish assessments are uncertain due to sparse historical data; mangrove and seagrass parameterization used alternative assumptions due to limited spatiotemporal data. - Excluded processes: Secondary WRD and agricultural impacts (e.g., increased sediments/turbidity, nutrients, agrochemicals), changes in sediment loads and geomorphology, migration barriers, and reservoir retention times were not modeled; river models also did not include migration disruption or sediment trapping effects. - Spatial assumptions: Connectivity among all GoC regions was not fully represented; potential buffering from large unmodified neighboring rivers might lead to slight overestimation of impacts in some locales. - Scenario design: Hypothetical WRDs omit advanced mitigation measures beyond pumping thresholds (e.g., sophisticated environmental flow rules); real-world operations could alter outcomes. Overall, risks may be underestimated where unmodeled impacts are important.