The vast majority of US rivers lack any protections from human activities, new research finds

University of Washington

Skagit river — image:
The Skagit River, pictured above, runs through northwestern Washington. Nearly 160 miles of the Skagit and its tributaries are protected by the National Wild and Scenic Rivers designation to preserve its scenic value and enhance recreational opportunities.
view more
Credit: University of Washington

The U.S. boasts more than 4 million miles of rivers, peppered with laws and regulations to protect access to drinking water and essential habitat for fish and wildlife. But in the first comprehensive review of river protection, research co-led by the University of Washington shows that the existing regulations account for less than 20% of total river length and vary widely by region.

Freshwater conservation strategies have historically emphasized protections against land use and development on public lands, including National Wildlife Refuges, Wilderness Areas and National Forests. However, protection measures that are specific to lakes, rivers and wetlands are much less common.

Most of the protection afforded to rivers comes from land-based measures, but the growing global consensus is that this isn’t enough. Freshwater ecosystems are losing biodiversity faster than anywhere else. To improve stewardship, researchers first need to map the existing protections and attempt to gauge their benefits.

“We examined the patchwork of different aquatic and terrestrial protection measures that seek to support river resilience to better understand where we are doing well and where there is room for improvement,” said Julian Olden, a UW professor of aquatic and fishery sciences.

Olden co-led this study with Conservation Science Partners and American Rivers. They published the results Jan. 9 in an article in Nature Sustainability, alongside a policy brief on the topic.

Rivers supply clean drinking water and power to millions of Americans. They provide habitat for fish, water for thirsty crops, and create transportation networks for people, goods and animals. But the nature of rivers makes them harder to protect. They cross borders, traverse ecological zones and snake between public and private lands.

Waterways are now represented in some major conservation initiatives, such as the Kunming-Montreal Global Biodiversity Framework — an effort to protect 30% of Earth’s land and ocean by 2030 — but that wasn’t always the case.

“Threats to fresh waters often originate outside the bounds of protected land areas,” Olden said. “So unfortunately no matter how much attention you give an individual stretch of river, it is only as protected as its headwaters.”

Because the mechanism of protection varies depending on the policy or management practice, the researchers developed a river protection index to compare river segments based on water quantity, quality, connectivity, habitat and biodiversity — key ecological attributes supporting freshwater resilience. They categorized segments by protection level to identify gaps and prioritize areas in need of protection.

“We layered local, state and federal protection mechanisms onto the river network to reveal where and how we seek to protect America’s rivers,” Olden said.

The study reported that nearly two-thirds of rivers in the U.S. are unprotected. Just over 19% of total river length in the entire U.S., and 11% in the contiguous U.S., is protected at a level deemed adequate to safeguard the health of river ecosystems. Results varied by region as well. Protections favor high elevation and remote areas, as well as public lands. Low-elevation headwaters and large swaths of the Midwest and South are underprotected.

River-specific protection efforts remain scarce. The Clean Water Act — a seminal freshwater protection measure passed in 1972 — protects just 2.7% of total river length. Habitat bulwarks for endangered species protect 1.3% and approximately 2% receive protection from river-specific designations, such as National Wild and Scenic Rivers.

Land-based regulations, by comparison, apply to a much larger chunk of the total. Federal Wilderness Area designations apply to 6.3% of total river length and river and floodplain protections encompass 14.2% of total river length.

The study also highlighted the potential value of investing in watershed management programs.

“Working to ensure that protected rivers also have protected upstream watersheds supports reliable access to clean water that doesn’t need treatment, which can be expensive, before it hits the faucets of American households,” Olden said.

Beefing up protections doesn’t mean cutting off access to rivers, either.

“We can use regulatory action to support equitable access to the numerous benefits rivers provide human society,” Olden said. “Protected rivers support recreation, freshwater biodiversity and cultural value. It’s a win-win-win.”

For more information, contact Olden at olden@uw.edu.

Additional co-authors include Lise Comte, Caitlin Littlefield and Brett Dickson of Conservation Science Partners; and John Zablocki and David Moryc of American Rivers.

This study was funded by American Rivers.

Journal

Nature Sustainability

DOI

10.5281/zenodo.17279334

Method of Research

Data/statistical analysis

Article Title

National assessment of river protection in the United States

Article Publication Date

9-Jan-2026

AI turns water into an early warning network for hidden biological pollutants

Biochar Editorial Office, Shenyang Agricultural University

image:
A review of AI-driven monitoring, forecasting, and source attribution of aquatic biocontaminants
view more
Credit: Qinling Wang, Yiran Zhang, Wenze Wang, Xinyi Wu, Hailing Zhou, Ling Chen & Bing Wu

Artificial intelligence is quietly transforming how scientists monitor and manage invisible biological pollutants in rivers, lakes, and coastal waters, and a new review explains how this technological shift could better protect ecosystems and public health.

In a paper published in the open access journal Biocontaminant, researchers from Nanjing University outline how AI can turn water quality management from a reactive, after the fact process into a proactive early warning and control system for harmful microbes, algal toxins, parasites, and antibiotic resistance genes in aquatic environments. These living “biocontaminants” are highly dynamic, able to grow, evolve, and spread with changing temperature, nutrients, and hydrology, which makes them far harder to track than traditional chemical pollutants.

“Our work shows that artificial intelligence has the potential to serve as an intelligent nervous system for aquatic environments, sensing subtle biological changes, learning from them, and triggering timely responses before risks escalate,” said lead author Qinling Wang from the School of Environment at Nanjing University. “The ultimate goal is to move from passively discovering problems in water bodies to actively preventing ecological and health crises.”

From static snapshots to real time sensing

Conventional monitoring of microbial and algal contamination often depends on periodic sampling and lab analysis, which can miss fast developing events like harmful algal blooms or pathogen outbreaks. The review describes how new intelligent sensors combined with edge computing and embedded machine learning models can now analyze signals directly in the field for near real time water quality assessment.

By integrating AI models into fluorescence, electrochemical, and Raman spectroscopy based sensors, devices evolve from simple data collectors into on site diagnostic terminals that recognize characteristic “fingerprints” of contaminants. In pilot studies, such AI enhanced sensing systems have been able to rapidly identify multiple pathogens or discriminate harmful algal species with high accuracy while operating on low cost, low power chips positioned directly at monitoring sites.

Forecasting blooms and outbreaks before they strike

Beyond detecting what is currently in the water, AI is also being used to forecast when and where biological hazards are likely to appear. According to the review, models such as deep neural networks, recurrent networks, and gradient boosting trees can learn complex relationships between environmental drivers for example temperature, nutrients, turbidity, and weather and the growth of algae, bacteria, and viruses.

These models have already been applied to predict harmful algal blooms days to months in advance, estimate pathogen concentrations in drinking water sources, and identify threshold conditions under which contamination risks rise sharply. When coupled with explainable AI techniques that highlight which factors matter most, such forecasts can guide practical decisions like reservoir operation, beach closures, or adjustments in water treatment.

Tracing invisible sources and pathways

A third frontier covered in the article involves using machine learning to trace where biocontaminants come from and how they move through interconnected water, sediment, biofilm, and infrastructure networks. By analyzing “microbial fingerprints” from high throughput DNA sequencing, AI based microbial source tracking tools can estimate how much of the contamination in a river or reservoir originates from sources such as human sewage, livestock, or wildlife.

The review also highlights AI studies that map the spread of antibiotic resistance genes across multiple environmental media, identify key microbial hosts, and reveal how stressors like microplastics or industrial chemicals can accelerate horizontal gene transfer. When combined with hydrological, land use, and wastewater data, spatiotemporal models can reconstruct contamination events and support wastewater based epidemiology for tracking community disease trends.

Promise, pitfalls, and the path ahead

Despite the promise, the authors emphasize that AI is not a magic solution. Biocontaminants are living, evolving systems, and high quality data on rare pathogens, emerging resistance genes, and long term ecological change are still scarce, which can limit model reliability.

Another major challenge is that many powerful AI models behave as black boxes, providing little insight into the underlying biology and offering few guarantees when conditions change beyond the range of past data. The review argues that future research should focus on adaptive sensing systems that continuously learn from new observations, hybrid models that embed ecological mechanisms such as growth and competition into neural networks, and dynamic network based risk assessment that considers whole ecosystems instead of single pollutants in isolation.

“AI systems for water management must be as adaptive as the ecosystems they monitor,” said senior author Bing Wu of the State Key Laboratory of Water Pollution Control and Green Resource Recycling at Nanjing University. “By integrating real time monitoring, ecological theory, and machine learning, we can move toward truly predictive management of aquatic health and safeguard both biodiversity and public health in a changing world.”

===

Journal reference: Wang Q, Zhang Y, Wang W, Wu X, Zhou H, et al. 2025. A review of AI-driven monitoring, forecasting, and source attribution of aquatic biocontaminants. Biocontaminant 1: e025

https://www.maxapress.com/article/doi/10.48130/biocontam-0025-0025

===

About Biocontaminant:
Biocontaminant (e-ISSN: 3070-359X) is a multidisciplinary platform dedicated to advancing fundamental and applied research on biological contaminants across diverse environments and systems. The journal serves as an innovative, efficient, and professional forum for global researchers to disseminate findings in this rapidly evolving field.

DOI

10.48130/biocontam-0025-0025

Method of Research

Literature review

Subject of Research

Not applicable

Article Title

A review of AI-driven monitoring, forecasting, and source attribution of aquatic biocontaminants

Himalayan balsam’s damaging impact on rivers revealed in new Stirling study

A three-year study led by Dr James Hardwick, a Lecturer in Earth Sciences at the University’s Faculty of Natural Sciences, has shown that the invasive species is doing more than just replacing native plants

University of Stirling

Pioneering University of Stirling led research has revealed the diverse and damaging impact Himalayan balsam has on river ecosystems.

A three-year study led by Dr James Hardwick, a Lecturer in Earth Sciences at the University’s Faculty of Natural Sciences, has shown that the invasive species is doing more than just replacing native plants.

The new research, carried out in partnership with Newcastle University, has shown that the species can disrupt the physical stability of ecosystems in ways that can have far-reaching consequences. Including significantly weakening riverbanks over winter.

By outcompeting native plants and then dying back completely each year, Himalayan balsam makes invaded riverbanks more susceptible to erosion during high-flow periods.

When riverbanks erode, more fine sediment enters their waters. This damages habitats, harms wildlife, and even changes the shape and structure of rivers.

This can then have a negative impact on water quality, biodiversity, and the benefits that healthy rivers provide.

Lead researcher Dr James Hardwick explained: “Our study reveals a previously overlooked seasonal process that illustrates how this plant weakens riverbanks. Himalayan balsam is often assumed to increase riverbank erosion simply because it dies-back leaving riverbanks bare over winter, but very little research has tested this relationship.

“Our work provides the first evidence that the impact of Himalayan balsam on the riverbank is not just about dieback. It’s about the way it suppresses native vegetation during summer, creating a loop that weakens riverbanks in winter. This seasonal mechanism has important implications for both river restoration and the management of invasive species.

“Understanding how invasive plants influence riverbank stability is essential for designing effective strategies that reduce erosion risks and protect river habitats.”

The research team hope that the findings will help river managers, conservation charities, local authorities, and environmental regulators better understand the risks associated with Himalayan balsam. Allowing more informed decisions to be made, particularly on intervention strategies.

Himalayan balsam was first introduced to the UK in 1839 as an ornamental garden plant. It was planted in places like Kew Gardens before escaping and spreading rapidly into the wild, especially along riverbanks and damp areas.

Researchers combined three years of field measurements with advanced statistical modelling. Over this period, they surveyed vegetation and measured riverbank shear strength - an indicator of how resistant the bank is to erosion - at multiple sites with and without Himalayan balsam.

To uncover these processes, researchers used Piecewise Structural Equation Modelling - a statistical method that breaks complex cause-and-effect links into smaller models - to explore how Himalayan balsam and native vegetation interact across seasons and the impact of these changes on riverbank stability.

Together, the field data and modelling approach enabled the team to quantify, for the first time, the indirect effects of Himalayan balsam on winter riverbank strength.

Invasive non-native plants indirectly destabilise riverbanks was published in the journal Biological Invasions as a cross-institution project between the University of Stirling and Newcastle University.

It was funded by Newcastle University as an internal PhD studentship with support from the Forth Rivers Trust and Tees River Trust who helped obtain site access.

A vegeation survey is carried out as part of the research.

Himalayan balsam growing by a riverbank in the UK

Credit

University of Stirling