The study, published in Frontiers in Environmental Science in 2026, was the first comprehensive microplastic assessment of the protected river reach. Investigators collected 35 large-volume water samples, each averaging about 5 cubic metres, from the Thaya/Dyje and two tributaries. The November campaign included 20 samples and the July campaign included 15, with sampling distributed along the river, across its width and, where depth allowed, near the bed.

In total, approximately 171 cubic metres of water passed through a plankton net with a 150-micrometre mesh. Concentrations were relatively low across the study area but ranged widely, from 1.09 to 22.66 microplastic particles per cubic metre. Average surface-water concentration in the main river was 8.18 ± 5.32 particles per cubic metre, while concentrations in the Fugnitz and Kaja tributaries averaged 3.14 ± 0.94 particles per cubic metre.

Fibres made up 56.22% of the particles, followed by fragments at 32.71% and films at 10.09%. Polyamide and polyester were the most frequent polymer categories, each accounting for nearly one-fifth of the total. The findings provide a baseline for a protected river system that remains connected to surrounding catchments and potentially receives microplastics from upstream human activities.

Spatial sampling revealed that the river’s main channel, banks and bed could tell different pollution stories. In November 2024, mean surface concentrations generally decreased downstream, from 12.68 ± 6.58 particles per cubic metre near Vranov to 4.38 ± 3.45 particles per cubic metre at the downstream Lipinská bridge. In July 2025, the pattern instead rose downstream: concentrations reached 5.04 ± 3.91 particles per cubic metre at the Hamerská bridge, 8.00 ± 1.45 after Hardegg and 11.35 after the Kaja junction.

The highest individual value, 22.66 particles per cubic metre, occurred in a low-velocity inner bend, while another peak of 20.14 particles per cubic metre was also recorded near an inner bend. These observations describe spatial and temporal differences; they do not demonstrate a single, persistent downstream trend. Flow velocity showed opposite relationships with concentration in the two campaigns.

In November, microplastic concentrations increased as velocity decreased, a pattern consistent with enhanced retention under reduced flow energy. The linear relationship explained 37% of the variation, with a Spearman correlation of −0.69. By July, concentrations instead rose with increasing velocity; the linear relationship explained 40% of the variation, and the monotonic correlation was positive at 0.59, although its p-value was 0.056.

A multiple-regression model containing an interaction between flow velocity and campaign performed better than an additive model, with an adjusted R² of 0.325 and a significant interaction term of p = 0.014. The authors interpreted this reversal primarily as a response to hydrological context and antecedent flow history rather than as a standalone seasonal effect. The contrast was consistent with the river’s recent history.

The November campaign followed an exceptional flood approximately six weeks earlier, while July brought generally lower and more stable discharge interrupted by occasional pulse releases associated with the upstream Vranov Dam. The flood could have expanded catchment-scale connectivity and mobilized stored sediments and associated particles, whereas summer pulses may have resuspended material temporarily stored within the regulated channel. Between October 2024 and September 2025, discharge at the Vranov–Hamry gauging station ranged from 3.26 to 30.3 cubic metres per second, underscoring the dynamic environment in which the particles moved.

Average discharge during the week before and after sampling was 6.33 ± 1.45 cubic metres per second in November and 4.50 ± 2.03 in July, based on hourly data. Discharge data offered a possible but less certain explanation for the combined observations. Across eight centre-channel samples, travel-time-corrected 10-minute discharge had a moderate positive linear relationship with concentration, explaining 51% of the variation.

However, the result depended strongly on the single highest-discharge July observation. Removing it left a similar estimated slope but reduced the explained variation to 21% and made the relationship statistically non-significant. The dataset also excluded the Czech Klaperův and Lukov creeks because they were dry during both campaigns, and tributary samples were omitted from formal regression models because the dataset was small, uneven and lacked complete discharge information.

Without tributary-specific flow data, the contribution of the Fugnitz and Kaja could not be quantified as particle fluxes. Local channel structure appeared to influence where denser particles were retained. Although meander position did not significantly alter total concentration, elevated readings at low-velocity bends suggest that incomplete lateral mixing and local retention can produce hotspots.

The findings also show why centre-channel sampling alone is insufficient: in July, near-bed concentrations at the Hamerská bridge ranged from 1.19 to 3.20 particles per cubic metre, below the local surface-water value, while a near-bed sample after the Kaja junction reached 8.46 particles per cubic metre compared with 11.35 at the surface. The study did not establish that channel structure influenced particle size. Quality-control results support the relative reliability of the measurements in this low-contamination setting.

Across 22 blank samples, field contamination was negligible, and all environmental results were corrected using measured blank contamination. Recovery averaged 87.3% for polyethylene spherules and 75.2% for low-density polyethylene fragments, with an overall recovery of 83.2% across six spiked samples. Analysing one-quarter of each filter recovered 25.9% of polyethylene microspheres, consistent with the proportion examined.

Thin fibres were recovered less efficiently, however, so the reported fibre concentrations primarily represent the size range retained by the combined 150-micrometre field mesh and laboratory procedure rather than the complete environmental fibre population. The authors also used non-synthetic clothing, glassware, filtered solutions and laminar-flow cabinets to reduce contamination during laboratory work. The results establish a baseline for microplastic monitoring in a protected freshwater system, but they also define limits on what the two campaigns can establish.

The observations are snapshots and cannot fully capture short-lived events such as storms, dam releases or rapid discharge changes. Discharge was reconstructed from gauges about 10 km apart rather than measured directly at every sampling point, and the restricted range of channel forms limited analysis of geomorphological controls. Sediments upstream of the Vranov Dam and within its reservoir were not sampled, leaving the potential long-term storage of microplastics unresolved.

The study also did not independently quantify connectivity or directly compare microplastic transport with natural sediment transport. The authors therefore treat hydrological connectivity as an interpretive framework involving transport, temporary storage, remobilization and incomplete mixing, rather than as a quantity independently measured by the particles themselves.

Subject of Research: Microplastic transport and hydrological variability in a protected transboundary river

Article Title: Spatio-temporal distribution of microplastics in the protected transboundary Thaya/Dyje River (Central Europe): implications of hydrological connectivity for microplastic monitoring

Article References: Roudbar, S., Young, R., Poeppl, R., Le Heron, D., & Wagreich, M. (2026). Spatio-temporal distribution of microplastics in the protected transboundary Thaya/Dyje River (Central Europe): implications of hydrological connectivity for microplastic monitoring. Frontiers in Environmental Science, 14, Article 1946645. https://doi.org/10.3389/fenvs.2026.1946645

Image Credits: AI Generated

DOI: 10.3389/fenvs.2026.1946645

Keywords: microplastics, river monitoring, hydrological connectivity, Thaya/Dyje River, flow velocity, large-volume sampling, particle remobilization, national park, environmental baseline, spatial variability, Spatio-temporal, distribution

The logic behind the effort is a long-standing problem in restoration ecology. When a river engineering project is completed and scientists later report that vegetation has shifted, critics can always ask: would the change have happened anyway? Floodplains are naturally dynamic, and plant communities fluctuate with floods, droughts and succession. Without a well-documented pre-intervention dataset, post-restoration assessments lack a reliable reference against which observed changes can be interpreted. Publishing baseline data in a standardised, reusable form allows future assessments to distinguish restoration-induced changes from natural temporal variation, and it opens the door to comparative studies and long-term monitoring across other river systems.

The study area itself is no ecological backwater. It lies within the transboundary Middle Danube floodplain shared by Croatia and Serbia, one of the largest remaining riverine wetland systems in Europe and part of the UNESCO Mura–Drava–Danube Biosphere Reserve, a region celebrated for its exceptional biodiversity. Here the interplay between the main channel, side arms, floodplain forests and shallow aquatic habitats creates a mosaic of habitats of remarkable heterogeneity. Previous botanical work in the wider Croatian Danube floodplain, notably in Kopački rit Nature Park, has documented high floristic richness, diverse wetland vegetation and the presence of numerous alien and invasive plant species — a reminder that restored connectivity can be a double-edged sword, reviving native floodplain dynamics while potentially spreading invasive species along the very water corridors being reopened.

The sampling design reflects a careful balance between scientific rigour and practical repeatability. Riparian forest vegetation was surveyed on 20 May 2024 at ten localities. At each locality, botanists established three square plots of 10 metres by 10 metres within representative forest stands, deliberately avoiding recently disturbed microsites and habitat transitions. Every vascular plant species within each plot was recorded, and its cover was estimated using the Domin cover-abundance scale, an ordinal scale running from 0 to 9 that is a staple of European vegetation science. Cover values were then averaged across the three plots to produce a site-level picture of the forest understorey and ground flora.

Aquatic vegetation received its own tailored treatment. On 30 July 2025, the same ten localities were revisited, and at each one three 10-metre belt transects were laid out within the dominant aquatic plant communities, generally parallel to the shoreline wherever the local morphology allowed. This orientation ensured comparable sampling across sites despite the irregular geometry of side-arm habitats. Species cover here was recorded on a five-level ordinal scale, ranging from class 1 for species covering more than 1 to 5 per cent of the transect up to class 5 for those covering more than 75 per cent. As in the forests, values were averaged across the three transects. Boundary cases were handled with a clear rule: an individual was included when the majority of its above-ground biomass fell inside the sampling unit.

One often-overlooked methodological choice deserves attention: every survey was conducted by the same experienced botanist using a single standardised field protocol. Species identification and cover estimation are notoriously observer-dependent, and inter-observer variability can swamp the subtle signals restoration ecologists are trying to detect. By relying on one observer, the team minimised this noise and improved the consistency and comparability of records across all twenty sampling sites. The protocol was also designed with the future in mind — it can be repeated almost verbatim during post-restoration monitoring, enabling direct temporal comparisons rather than vague qualitative impressions.

The dataset goes well beyond a simple species inventory. Alongside the vegetation records, the team compiled event-level environmental descriptors for every site: canopy-cover classes for the forest plots, and field measurements of pH, dissolved oxygen, conductivity and water temperature for the aquatic sites, together with ordinal water-velocity classes distinguishing standing from slow-flowing conditions. Elevation and CORINE Land-Cover categories from the European Union’s Copernicus Land Monitoring Service were recorded for all sampling events, anchoring each site in its broader landscape context. This combination means future analysts can ask not just whether vegetation changed, but whether those changes track shifts in hydrology, light availability or water chemistry.

In total, the published dataset comprises 20 sampling events — ten in riparian forest and ten in aquatic habitats — 251 taxon-by-event occurrence records and 331 measurements. The taxonomic census identified 72 taxa: 40 recorded only in riparian forest habitats, 28 only in aquatic habitats, and four occurring in both, giving 44 forest taxa and 32 aquatic taxa overall. Each record carries an establishment status flag — native, introduced or invasive — a crucial detail for tracking invasion dynamics as connectivity changes. Nomenclature was standardised against Plants of the World Online, and the data were checked for duplicate records, missing identifiers and logical consistency across the linked event, occurrence and measurement tables before publication.

The data are organised in three relational CSV tables following established biodiversity informatics conventions. A sampling-events table holds spatial coordinates, elevation, dates, habitat classifications and protocol details; a species-occurrence table links each scientific name to its sampling event with establishment status and taxonomic references; and a measurement table carries the abundance values and environmental readings, each tied to both its occurrence and its event. The complete dataset has been archived in the Zenodo repository under an Open Data Commons Attribution License, ensuring long-term accessibility and reproducibility, while supplementary files accompany the paper for immediate inspection.

What happens next will be the real test. When the side arm is reconnected, flow conditions, inundation dynamics and habitat structure in the adjacent forests and aquatic habitats are expected to change, and vegetation responses are likely to be gradual and context-dependent, as river-floodplain research has repeatedly shown. Armed with this baseline, the DaWetRest monitoring programme can repeat the identical surveys and measure the difference — separating genuine restoration effects from the background noise of natural variation. The dataset also stands as a resource for invasion biology, floodplain ecology and restoration science far beyond Croatia, offering a template for how a single, well-designed ‘before’ survey can turn a local engineering project into a rigorous, long-running natural experiment on one of Europe’s great rivers.

Subject of Research: Baseline vegetation monitoring for Danube floodplain side-arm reconnection restoration

Article Title: Baseline riparian forest and aquatic macrophyte vegetation dataset from the Danube right bank near Draž (Croatia), to support before-after monitoring of side-arm reconnection (2024-2025)

Article References: Vukov, D., Žunić, D., Prodanović, N., & Sabadoš, V. (2026). Baseline riparian forest and aquatic macrophyte vegetation dataset from the Danube right bank near Draž (Croatia), to support before-after monitoring of side-arm reconnection (2024-2025). Biodiversity Data Journal, 14, Article e190669. https://doi.org/10.3897/bdj.14.e190669

Image Credits: AI Generated

DOI: 10.3897/BDJ.14.e190669

Keywords: Danube River, floodplain restoration, riparian forest, aquatic macrophytes, baseline dataset, side-arm reconnection, biodiversity monitoring, DaWetRest, Croatia, invasive species, hydrological connectivity, vegetation survey

The collection grew out of an international conference called The Fish of Muddy Waters, held in Vienna, Austria, in October 2025. Over three days, researchers presented 52 studies, including 42 oral presentations and 10 posters, all focused on fish species specifically adapted to the floodplain wetlands of European rivers. The eleven papers that made it into the special issue represent only a fraction of that work, but they capture the geographic and thematic breadth of the effort, spanning from the Tagus floodplain wetlands of Portugal to the Danube lowlands of the Pannonian Basin, and from Croatia to northern Germany.

Three species anchor the issue: the European mudminnow (Umbra krameri), the European weatherfish (Misgurnus fossilis), and the crucian carp (Carassius carassius). Despite differences in their ranges and conservation status, these fishes share a defining ecological trait. They are all strongly associated with shallow, structurally complex wetlands that are frequently isolated from larger water bodies or only temporarily connected to them. That specialization makes them excellent indicators of wetland integrity, but it also makes them acutely vulnerable when those habitats disappear, degrade, or become fragmented.

The physics and chemistry of wetland waters explain much of this vulnerability. Because these systems are shallow and packed with vegetation, summer temperatures can climb steeply, and dissolved oxygen fluctuates dramatically, with hypoxic conditions frequently developing in isolated or heavily vegetated zones. Such conditions favor fish species with physiological or behavioral adaptations to low oxygen concentrations and elevated temperatures, while excluding less tolerant species typical of larger rivers and lakes. In other words, the very features that make wetlands hostile to ordinary fish are precisely what allow wetland specialists to persist, which is why hydrological alteration is such a direct threat to them.

That threat is mounting on multiple fronts. Land-use change, pollution, and climate change are degrading wetlands across the continent, and the loss, degradation, and fragmentation of suitable habitats increasingly interact with biological invasions. Remnant populations of native wetland fishes now face competition, predation, pathogen transmission, and hybridization from invasive non-native fishes. Among the experimental studies in the special issue is work examining habitat use and potential competition with invasive Asian weatherfishes, a group that exemplifies how introduced species can exploit the same niches that native specialists depend on.

The methodological range of the new papers is striking. Some analyze species distributions in relation to habitat preferences and habitat deterioration across Portugal, Croatia, northern Germany, and the Pannonian Basin. Others probe population genetics to understand how fragmentation has shaped the diversity of remaining populations. Applied conservation contributions span conservation genetics, the engagement of citizen scientists in monitoring, the development of regional action plans, and ultimately the assessment of conservation translocation success, a critical but often neglected step in determining whether reintroduction programs actually work.

For the European mudminnow, the special issue dedicates several papers covering its genetics, distribution, and translocation outcomes, and additional conference findings were folded into an updated monograph on the species. This depth of attention reflects the mudminnow’s status as one of Europe’s most iconic wetland specialists, a fish confined to small, often ephemeral water bodies whose survival depends on the fine-grained management of local hydrology. The broader lesson, the editors argue, is that conservation requires more than simply protecting the remaining localities where these species still occur.

Effective action, according to the synthesis of the special issue, must combine the maintenance or restoration of appropriate hydrological regimes and habitat structure with invasive species management, preservation of genetic integrity, and long-term population monitoring. Hydrological connectivity deserves particular nuance. Periodic reconnection can restore natural floodplain processes and benefit native specialists, whereas permanent connections may facilitate the spread of invasive competitors and predators into refugial habitats, effectively destroying the isolation that native wetland fishes rely upon. Getting this balance right is one of the central technical challenges facing wetland restoration across Europe.

The findings arrive at a politically significant moment. The authors note that the research is directly relevant to the implementation of the EU Nature Restoration Regulation, particularly for the identification, restoration, and monitoring of freshwater habitats and habitats of species of conservation concern. Translating distributional, ecological, and genetic knowledge into regional action plans, carefully designed translocations, and measurable assessments of conservation success is presented as the pathway from science to policy, ensuring that restoration targets are grounded in what these species actually need rather than in generic habitat goals.

Preventing further losses, the editors conclude, requires coordinated cross-border conservation that combines habitat restoration, invasive species management, long-term monitoring, and, where appropriate, reintroduction programs that respect the local genetic variability of selected species. By raising awareness of the plight of the European mudminnow, the European weatherfish, and the crucian carp, the researchers hope to stimulate further conservation action and help reverse the ongoing population declines of fishes that have, for too long, thrived unnoticed in Europe’s muddy waters.

Subject of Research: Biology and conservation of European wetland fishes

Article Title: Special issue in the journal “Nature Conservation” (Pensoft): biology and conservation of wetland fishes in Europe

Article References: Wanzenböck, J., Šmejkal, M., Kalous, L., Klefoth, T., Geist, J., & Takács, P. (2026). Special issue in the journal “Nature Conservation” (Pensoft): biology and conservation of wetland fishes in Europe. Nature Conservation, 65, 1-5. https://doi.org/10.3897/natureconservation.65.215711

Image Credits: AI Generated

DOI: 10.3897/natureconservation.65.215711

Keywords: wetland fishes, European mudminnow, European weatherfish, crucian carp, floodplain wetlands, invasive species, conservation genetics, habitat restoration, hydrological connectivity, EU Nature Restoration Regulation, Nature Conservation journal, Vienna conference

A new study published in the journal Natural Hazards tackles this puzzle with a grid-based, explainable machine learning framework designed to diagnose the dominant flood-influencing factors under hydrologically heterogeneous conditions. Hyeontae Moon and Gilho Kim of the Korea Institute of Civil Engineering and Building Technology, together with Kyung-Tak Kim of Kyungpook National University, integrated historical flood-trace maps with multi-layer spatial datasets to classify flooded and non-flooded grid cells across Jeju Island at a 100-meter resolution. Their central finding is striking: the dominance of flood-influencing factors varies dramatically across spatial contexts, with hydrometeorological conditions acting as the primary drivers while spatial exposure and proximity-based factors contribute in a distinctly region-specific manner.

The methodological architecture is as important as the results. The researchers discretized the entire island into 100-meter by 100-meter grid cells, which served as the fundamental spatial units for flood labeling and predictor extraction. The target variable was constructed from official flood-trace records for 23 rainfall-induced flood events that occurred between 2006 and 2022, obtained from Korea’s Public Data Portal. Original inundation-depth and flood-extent layers were matched using the reported disaster name and recorded starting and ending dates, then separated into event-specific vector maps. Each event-specific flood-trace map was independently rasterized onto the common grid system, with grid cells satisfying the adopted flood-overlap criterion labeled as flooded and the remaining cells labeled as not flooded for that event. Crucially, the dataset was organized at the event-grid level rather than from a temporally aggregated multi-event flood map, meaning the same geographical cell could appear as a separate observation in multiple events with event-specific flood labels and time-varying predictors.

To represent the physical, hydrological, and socio-economic conditions of Jeju Island, the team selected 30 flood-influencing factors categorized into five functional groups: hazard factors such as rainfall, runoff, groundwater, and wind conditions; hydrogeological factors including elevation, slope, curvature, and land cover; exposure factors such as building counts, population, and property value; vulnerability factors like old buildings and dependent populations; and capacity and proximity-based factors including road and river density, reservoir capacity, and distances to hydrological or infrastructural features. Rainfall variables were calculated separately for each of the 23 events, comprising the 3-hour maximum rainfall, the 12-hour maximum rainfall, and the 5-day antecedent rainfall, with durations selected in accordance with the Korea Meteorological Administration’s heavy-rain advisory criteria. Event-specific rainfall values from 40 observation stations were spatially interpolated using kriging to generate 100-meter gridded rainfall layers matched to the flood labels of the corresponding event.

Five tree-based machine learning classifiers were trained on the gridded dataset: a single Decision Tree as a transparent baseline, Random Forest as a bagging-based ensemble, and XGBoost, LightGBM, and CatBoost as alternative gradient-boosting implementations with different tree-construction, boosting, and regularization strategies. The use of multiple algorithms was intended not to identify a single optimal predictor but to assess the robustness and consistency of inferred flood-influencing factors across different learning architectures. Recursive Feature Elimination was incorporated into the training protocol to examine changes in model performance and feature-importance stability as less informative predictors were progressively removed. Because flooded records accounted for only approximately 0.23 percent of the complete dataset, overall accuracy could be dominated by the majority non-flooded class, so the researchers emphasized recall, F2-score, balanced accuracy, the Matthews correlation coefficient, and the area under the precision-recall curve alongside precision and F1-score.

The Random Forest classifier emerged as the most stable performer, achieving recall and F1-score values of 0.86 and 0.78 respectively for the island-wide dataset, with higher recall values in some sub-basins. Notably, Random Forest performance stabilized with a reduced subset of approximately nine predictors, suggesting that a limited number of hydrometeorological and hydro-geomorphological variables can retain substantial discriminatory power for flood occurrence. CatBoost and XGBoost demonstrated comparable predictive behavior in several sub-basins, though their performance exhibited greater variability across feature-elimination iterations, while the Decision Tree showed moderate and more variable performance and LightGBM displayed unstable patterns in certain sub-basins, possibly reflecting sensitivity to hyperparameter settings or data imbalance under strong spatial heterogeneity.

To directly quantify the incremental value of non-rainfall predictors, the researchers evaluated an island-wide rainfall-only baseline using just the three rainfall variables. At the recall-oriented operational threshold of 0.2, the rainfall-only model achieved precision, recall, F1-score, and F2-score values of 0.605, 0.691, 0.645, and 0.672 respectively, while the multivariate model achieved corresponding values of 0.693, 0.807, 0.746, and 0.781. Threshold-independent discrimination improved as well, with the area under the receiver operating characteristic curve increasing from 0.941 to 0.978 and the precision-recall area from 0.666 to 0.819. These controlled comparisons indicate that the additional predictors provided meaningful incremental discriminatory information beyond rainfall alone, representing atmospheric conditions, subsurface states, topographic settings, and hydrological connectivity that rainfall descriptors cannot fully capture.

The explainability analysis, conducted using XGBoost with SHAP (SHapley Additive exPlanations) attribution, revealed pronounced spatial heterogeneity across the island’s four official watershed units. At the island-wide scale, rainfall-related predictors dominated, collectively accounting for approximately 50 to 60 percent of total mean absolute SHAP importance, with groundwater level and proximity-based connectivity indicators contributing an additional 20 to 25 percent. In the Western Basin, longer-duration rainfall and antecedent accumulation together accounted for roughly 30 to 40 percent of cumulative SHAP importance, reflecting the sensitivity of low-lying coastal zones to sustained precipitation. In the Northern Basin, which encompasses the dense urban core of Jeju City, short-duration high-intensity rainfall alone contributed approximately 25 to 35 percent of total SHAP importance, with an indicative transition range around 60 to 80 millimeters of 3-hour rainfall within which flood probability tends to increase more rapidly. The model also identified a subsurface threshold in the Northern Basin: when groundwater levels rise above minus 5.0 meters relative to the surface, their contribution to flood susceptibility turns positive, reflecting constrained infiltration capacity during extreme events.

The Southern Basin exhibited a distinct response regime driven by orographic enhancement along the slopes of Hallasan Mountain, where longer-duration rainfall and antecedent precipitation together exceeded 35 to 45 percent of cumulative SHAP importance and the transition to high-risk states was notably associated with 5-day antecedent rainfall exceeding 120 to 150 millimeters. The Eastern Basin showed the strongest inter-model consistency, with rainfall variables, groundwater level, and proximity-based indicators jointly accounting for more than 65 to 70 percent of total SHAP importance. Across all sub-basins, proximity-based variables exhibited consistently negative SHAP contributions, with the highest magnitudes concentrated within a 200 to 300 meter radius of drainage infrastructure, indicating sharply increased flood probability as grid cells approach rivers or detention facilities. Perhaps most provocatively, socio-economic exposure variables such as building density, population, and land-price indicators contributed less than 3 to 5 percent of total SHAP importance across all spatial domains, suggesting that hydrometeorological and geomorphological controls dominate the physical likelihood of flood occurrence while socio-economic factors primarily modulate damage magnitude rather than inundation initiation.

The authors are careful to frame their findings as region-specific diagnostic evidence rather than proof of universal flood-generation mechanisms. A five-fold geographically blocked validation using 10-kilometer spatial blocks revealed that transferability to spatially unseen areas remains limited: at threshold 0.2, the random event-grid-level test yielded an F1-score of 0.746 and balanced accuracy of 0.903, whereas the blocked analysis yielded pooled values of only 0.266 and 0.663, indicating that the random-partition metrics should be interpreted as within-domain or upper-bound estimates rather than evidence of spatial transferability. The reliance on official flood-trace delineations also introduces limitations in spatial completeness and temporal resolution, since smaller or ephemeral inundation patches may not be fully captured and the temporal linkage between targets and predictors was established at the flood-event scale rather than at an hourly grid-cell scale. Nevertheless, the study’s implications for flood risk management are clear: conventional rainfall-centric warning systems implicitly assume a relatively uniform relationship between precipitation and flood occurrence, but the results demonstrate that flood generation on Jeju Island is modulated by hydrological connectivity, antecedent groundwater conditions, and region-specific drainage characteristics. The findings support a conceptual shift toward multi-parameter flood warning frameworks in which region-specific alert criteria are defined using joint conditions reflecting dominant mechanisms, for example triggering risk escalation when short-duration rainfall exceeds a critical level under elevated groundwater conditions, or when prolonged rainfall coincides with limited drainage connectivity in low-lying coastal zones. As hydroclimatic uncertainty grows, the ability to diagnose which combinations of variables matter most in each sub-region offers a pathway toward more physically grounded and locally adaptive flood-risk assessment in geologically complex environments where traditional runoff assumptions simply do not hold.

Subject of Research: Explainable machine learning diagnosis of region-specific flood influencing factors on volcanic Jeju Island

Article Title: Diagnosing region-specific flood influencing factors in a volcanic island using explainable machine learning

Article References: Moon, H., Kim, G., & Kim, K.-T. (2026). Diagnosing region-specific flood influencing factors in a volcanic island using explainable machine learning. Natural Hazards, 122(22), Article 674. https://doi.org/10.1007/s11069-026-08426-w

Image Credits: AI Generated

DOI: 10.1007/s11069-026-08426-w

Keywords: flood prediction, explainable machine learning, Jeju Island, volcanic hydrology, SHAP, groundwater, Random Forest, hydrological connectivity, flood risk management, Natural Hazards, feature selection, typhoon rainfall

The team’s target was net primary productivity, or NPP, the amount of carbon that plants fix from the atmosphere each year. NPP is a fundamental gauge of ecosystem health, and satellite products such as NASA’s MOD17A3HGF.061 dataset, derived from the MODIS instrument on the Terra satellite, now provide annual estimates at a resolution of 500 meters across the globe. For their analysis, the researchers aggregated these satellite observations to the county level, reconstructing annual NPP for 382 counties spanning the Yellow River Basin between 2003 and 2022. This region is a natural laboratory for such work: it encompasses the Loess Plateau, where decades of large-scale revegetation have pushed ecosystems close to sustainable water-resource limits, and its vegetation dynamics respond to both climatic swings and intense human management.

The methodological heart of the paper is what the authors call a lag-controlled residual benchmark. The idea is elegant in its simplicity. Because vegetation productivity exhibits strong temporal persistence — this year’s NPP in a county is highly correlated with last year’s — any model that predicts NPP directly can look impressive simply by echoing the previous year. To strip away that advantage, the researchers first built a baseline model using only the lag-1 value, meaning the NPP recorded one year earlier. They then trained their graph models not to predict NPP itself, but to predict only the residual: the portion of this year’s productivity that last year’s value fails to explain. If spatial connectivity genuinely carries additional information, the graph models should be able to squeeze signal out of those residuals. If they cannot, the apparent skill of complex spatial models is an illusion created by temporal persistence.

Four graph architectures entered the contest, each encoding a different hypothesis about how counties are connected. The Euclidean-Static graph linked neighboring counties by straight-line distance, the most common and least informative assumption. The Hydro-Static graph connected counties that share hydrological relationships through the river network, reflecting the intuition that water flows, and the ecosystems along it, tie upstream and downstream regions together. The Hydro-Dynamic graph went further, allowing the strength of those hydrological connections to change over time, for instance as rainfall and drought conditions shift the coupling between counties. Finally, an identity-skip graph served as a control, essentially testing whether any learned connectivity mattered at all. Each model was trained to predict the residual NPP beyond the lag-1 baseline, and the experiments were repeated across 20 random seeds to guard against the luck of random initialization, with an independent test period from 2020 to 2022 held out entirely from training.

The results were unambiguous, and for advocates of graph learning, uncomfortable. The lag-1 baseline alone achieved a root mean square error of 0.0404, with an R-squared of 0.8977, meaning it explained nearly ninety percent of the variance in annual county-level NPP using nothing but the previous year’s value. The four graph models, despite their elaborate connectivity structures, produced root mean square errors between 0.0497 and 0.0506 — consistently and substantially worse than the baseline they were supposed to improve upon. In other words, when the models were forced to explain only what temporal persistence could not, they failed to find meaningful signal in the spatial structure of the basin.

The researchers also probed whether the hydrological graphs at least held an edge under particular conditions. They stratified their evaluation by wetness regime, comparing dry, normal, and wet years, and by subbasin, reasoning that dynamic hydro-climatic connectivity might matter most where water stress is severe or where riverine coupling is strongest. No group showed a robust dynamic advantage. A paired Wilcoxon test, a nonparametric statistical comparison suited to repeated runs across seeds, found that the Hydro-Dynamic graph did not significantly outperform the simpler symmetric Hydro-Static graph, with a p-value of 0.0696. The extra complexity of time-varying connectivity bought nothing statistically detectable.

To rule out the possibility that the problem lay with graph neural networks specifically rather than spatial modeling in general, the team also benchmarked conventional residual-target spatial models. Geographically weighted regression, a classical technique that fits locally varying relationships across space, performed best among these alternatives, achieving a root mean square error of 0.0443. That is better than the graph models, and it shows that the classical spatial econometric toolkit retains real value, but it still fell short of the lag-1 baseline. The pattern held across every family of methods: at the annual county scale, increasingly complex representations of spatial connectivity did not translate into stable predictive gains beyond the temporal benchmark.

Why does this matter beyond the Yellow River Basin? The implications reach into a broader debate about how environmental machine learning is evaluated. Surveys of spatio-temporal graph neural networks document explosive growth in their application to traffic, climate, hydrology, and ecology, yet rigorous comparisons against simple temporal baselines remain rare. When a target variable is strongly autocorrelated in time — as annual productivity, temperature, or streamflow often are — a naive persistence forecast can be remarkably hard to beat, and models that blend spatial and temporal information can appear skillful while contributing little beyond what persistence already provides. The lag-controlled residual framework offers a reproducible diagnostic for separating these two sources of skill, and it costs almost nothing to implement: train a lag baseline, compute residuals, and test whether spatial structure explains any of what remains.

The study also carries a practical message for anyone modeling vegetation in a changing climate. Annual NPP in this basin is dominated by memory — carried by standing biomass, soil conditions, root systems, and slow-recovering vegetation cover — rather than by year-to-year information flowing between neighboring counties through river networks or climatic teleconnections. If researchers want to detect genuine spatial coupling, they may need finer temporal resolution, where seasonal anomalies and extreme events such as droughts propagate through landscapes, rather than annual aggregates in which persistence swamps everything else. Drought-legacy research has shown that ecosystem responses to water stress can echo for years, and capturing such effects may require the very kind of dynamic modeling the graph architectures were meant to provide — but only once the temporal baseline is properly controlled.

For a field racing to publish ever-larger and more elaborate models, the paper is a reminder that the most important competitor is often the simplest one. The authors, who report no external funding and no competing interests, frame their contribution not as an attack on graph learning but as a benchmarking discipline: a way to ask, honestly and reproducibly, whether spatial connectivity adds anything at all. In this case, the answer was no — and that negative result, rigorously established across seeds, architectures, wetness regimes, and subbasins, may prove more influential than any positive one. As graph-based geospatial modeling matures, studies like this one will help distinguish genuine advances in our understanding of environmental systems from elaborate repackaging of temporal inertia.

Subject of Research: Benchmarking graph-based geospatial models against a lag-1 temporal baseline for predicting county-level vegetation net primary productivity in the Yellow River Basin

Article Title: A lag-controlled residual benchmark for graph-based geospatial modelling of vegetation productivity

Article References: Wang, X., Wang, X., Cao, D., & Wang, Y. (2026). A lag-controlled residual benchmark for graph-based geospatial modelling of vegetation productivity. Earth Science Informatics, 19(10), Article 169. https://doi.org/10.1007/s12145-026-02232-5

Image Credits: AI Generated

DOI: 10.1007/s12145-026-02232-5

Keywords: graph neural networks, net primary productivity, Yellow River Basin, temporal persistence, residual benchmark, geospatial modelling, MODIS remote sensing, hydrological connectivity, geographically weighted regression, ecosystem forecasting, spatial machine learning, model evaluation

Freshwater ecosystems rank among the most biodiverse and most threatened environments on Earth, and headwater streams sit at the top of the conservation agenda because of their outsized influence on the biodiversity, functioning, and connectivity of entire river networks. In Mediterranean mountains, where irrigated agriculture and climate change have extensively reshaped flow regimes, these small streams face mounting pressure. Sierra Nevada, a designated Biosphere Reserve and a recognized Mediterranean biodiversity super hotspot, offers a striking case study. There, careo ditches have been documented since the eighth to tenth centuries, diverting snowmelt from headwater reaches and distributing it across upper slopes to artificially recharge aquifers in weathered fractured rock, extending water availability for lowland farms and towns through the dry season.

A research team led by scientists from the University of Almeria, the Andalusian Centre for Global Change, and the Geological and Mining Institute of Spain examined five first- and second-order headwater streams on the southern face of the park, in the Alpujarra region, at altitudes between 1800 and 2000 meters. In each stream, a small, rudimentary weir built of loose rocks and gravel diverts water into a careo ditch. The researchers compared upstream reference reaches with downstream flow-impacted reaches, sampling benthic macroinvertebrates in spring during snowmelt and in autumn after the prolonged dry period, while continuously monitoring discharge and water temperature from November 2022 onward.

The magnitude of diversion was severe. Proportional flow reduction downstream of the weirs most frequently ranged from 76 to 98 percent of upstream discharge, with mean reductions exceeding 0.85 for much of the year in the Cáñar, Mecina, and Bérchules streams. In absolute terms, downstream flows often fell below 10 liters per second. Yet despite these extreme withdrawals, the impacted reaches never dried. They remained perennial throughout the year, maintaining the longitudinal hydrological connectivity that ecologists consider critical for stream life. The team attributes this persistence to two concurrent processes: seepage returning from the leaky weirs and unlined or poorly sealed ditches, and groundwater effluents from the region’s weathered hard-rock aquifers.

The groundwater signal was unmistakable in the temperature records. Downstream reaches showed significantly higher winter temperatures despite their lower discharge, along with lower summer maxima and reduced spring daily fluctuations, a thermal fingerprint of groundwater-fed baseflow. Water chemistry, by contrast, remained remarkably stable across reach types, with only slightly lower oxygen and higher salinity downstream, differences that were not statistically significant. Total dissolved nitrogen was actually higher upstream, suggesting stronger surface runoff influence at reference sites. This chemical stability matters, because it means that any ecological differences between reaches could be traced primarily to physical habitat alteration, such as reduced flow diversity and increased sedimentation, rather than to water quality degradation.

The biological consequences were nuanced. Macroinvertebrate density was substantially reduced downstream, particularly in spring, when it was roughly 50 percent lower than at upstream sites, a pattern consistent across all functional feeding groups and supported by medium effect sizes. The prolonged duration of the reduction, which persisted for at least a full year, contracted the wetted channel width by 63 to 85 percent, slashing total benthic abundance per reach length to about one-tenth of reference values. The authors warn that this decline in benthic production, together with reduced insect drift and emergence, likely creates severe food shortages for trout and riparian insectivorous predators that depend on aquatic insects crossing the land-water interface.

Surprisingly, local alpha diversity and interlocal beta diversity, measured as taxa richness, Shannon diversity, and community turnover, did not differ significantly between upstream and downstream reaches, and community composition showed no detectable separation in ordination analyses. The researchers propose that the maintenance of hydrological connectivity prevented the impacted reaches from degrading into isolated pools, a fate that typically traps invertebrates, intensifies predation, and triggers prey population collapse. Groundwater subsidies may also have mitigated local diversity losses, since moderate upwelling is known to enhance macroinvertebrate abundance and richness in alpine streams by moderating the harsh abiotic stress imposed by meltwater.

However, the buffering was incomplete, and the damage emerged at larger scales. Regional gamma diversity, estimated from rarefaction curves, was significantly lower in the sets of downstream reaches in both seasons, indicating that even small, non-significant losses of local richness can accumulate into substantial regional declines. Sensitive components of the community also eroded: the taxonomic richness of collector-gatherers dropped significantly in spring, and several biotic indices based on the pollution- and disturbance-sensitive insect orders Ephemeroptera, Plecoptera, and Trichoptera, the EPT taxa, declined significantly at impacted sites. Notably, the IBMWP index, the regulatory standard for Iberian river biomonitoring, classified both reach types as being of good quality and failed to distinguish them, suggesting that current monitoring frameworks may be blind to the early functional erosion caused by flow diversion unless they incorporate EPT-driven metrics.

The study’s most consequential implication concerns the global push toward irrigation efficiency. Modern water policy has favored concrete-lined canals and sealed infrastructure that minimize seepage, yet the Sierra Nevada findings suggest that these very inefficiencies, the leaky weirs and porous ditches of the traditional system, are essential to sustaining downstream flow and biodiversity, particularly in reaches that lack groundwater inflow. The authors invoke the well-documented irrigation efficiency paradox, in which saving water at the infrastructure scale can paradoxically intensify overall water consumption and ecological harm, and they argue that reversing the efficiency-first trend is urgent to halt river degradation in biodiversity-rich mountain regions.

Ultimately, the researchers argue that preserving as many stream reaches as possible under natural or ecologically compatible flow regimes is mandatory in biodiversity hotspots, because moderate local losses compound into regional extirpation. As climate change intensifies aridity across the Mediterranean and agricultural water demands climb, the centuries-old careo systems, increasingly framed as nature-based solutions for aquifer recharge, offer a rare example of human water infrastructure coexisting with, and even underwriting, freshwater conservation. The lesson from Sierra Nevada is that sometimes the most valuable feature of an irrigation system is not how efficiently it moves water, but how much it lets slip away.

Subject of Research: Ecological effects of traditional irrigation water diversion on benthic macroinvertebrate biodiversity in Sierra Nevada headwater streams

Article Title: Leakage from Traditional Irrigation Systems and Groundwater Upwelling Buffer the Impact of Heavy Streamflow Diversion on Benthic Biodiversity

Article References: Casas, J. J., Fenoy, E., Rubio-Ríos, J., Villegas, J., Salinas-Bonillo, M. J., Zakaluk, T., Martos-Rosillo, S., & Cabello, J. (2026). Leakage from Traditional Irrigation Systems and Groundwater Upwelling Buffer the Impact of Heavy Streamflow Diversion on Benthic Biodiversity. Water Resources Management, 40(11), Article 517. https://doi.org/10.1007/s11269-026-04861-3

Image Credits: AI Generated

DOI: 10.1007/s11269-026-04861-3

Keywords: benthic macroinvertebrates, headwater streams, irrigation ditches, groundwater upwelling, hydrological connectivity, water abstraction, Sierra Nevada, EPT taxa, biotic indices, environmental flows, Mediterranean mountains, biodiversity conservation

The term “aerial rivers” describes narrow, elongated bands of concentrated moisture in the atmosphere. They are often compared with atmospheric rivers because they transport enormous quantities of water vapor from one region to another, sometimes across thousands of kilometers. When these moisture corridors encounter mountains or weather systems, the air is forced upward, cooling the water vapor and producing intense precipitation. That rainfall can feed streams, recharge soils and wetlands, and alter the timing and volume of runoff entering major river basins.

South America is an especially important natural laboratory for studying this process. The continent contains the Amazon Basin, the world’s largest river system by discharge, as well as the Andes, the Pantanal wetlands, the La Plata Basin and extensive drylands in the west and south. Moisture does not simply arrive from the Atlantic and fall locally. It can be recycled through forests, redirected by mountain ranges and transported across national borders before returning to the surface as rain or snow. The study investigates how these aerial pathways correspond to different hydrological regimes and drainage patterns across the continent.

A hydrological regime describes the seasonal rhythm of water movement through a landscape. In a river system, that rhythm may be controlled by rainfall, snowmelt, glacier melt, evaporation, groundwater exchange or a combination of these processes. In the Andes, for example, precipitation stored as seasonal snow and ice can delay the delivery of water to downstream valleys. In tropical basins, rainfall may be more closely linked to atmospheric moisture transport and convection. By connecting aerial rivers with these contrasting regimes, the research helps explain why neighboring watersheds can respond very differently to the same large-scale climate conditions.

The drainage system is the landscape’s physical network for moving water. It includes headwater streams, tributaries, floodplains, wetlands and main river channels, all shaped by topography and geology. Atmospheric moisture can influence this network in several ways. Persistent wet-season transport may sustain broad floodplains and high river discharge, while short-lived but intense moisture pulses can trigger flash floods, landslides and rapid channel changes. In mountainous terrain, the location and direction of aerial rivers may determine which slopes receive the heaviest precipitation and which valleys experience the greatest hydrological stress.

One of the study’s broader scientific contributions is its effort to view atmospheric and terrestrial water systems as a single connected process. Conventional maps often show rivers as blue lines drawn across the ground, while atmospheric circulation is analyzed separately using weather data. Yet the water in a river may have traveled through a complex chain of evaporation, atmospheric transport, condensation and runoff before reaching a gauge station. Tracking that chain can reveal why rainfall anomalies in one region may produce flooding or drought far downstream, even when local weather appears ordinary.

This perspective is increasingly important as climate change reshapes the water cycle. A warmer atmosphere can hold more water vapor, potentially intensifying moisture transport when conditions favor the formation of atmospheric rivers. At the same time, warming can reduce snow and glacier storage in the Andes, changing the seasonal timing of runoff. Some locations may face heavier rainfall and flooding, while others could lose dependable dry-season water supplies. Understanding the baseline behavior of aerial rivers provides a foundation for detecting how these systems are shifting and for identifying communities and ecosystems most exposed to the changes.

The findings also carry implications for forecasting and water management. River discharge models become more useful when they account not only for rainfall over a basin but also for the pathways that deliver moisture to it. Improved monitoring of atmospheric rivers could strengthen early warnings for floods, landslides and extreme rainfall, particularly in regions where weather stations are sparse. It could also support decisions about reservoir operations, agricultural planning, hydropower generation and wetland conservation. Because many South American watersheds cross political boundaries, understanding shared atmospheric sources may be as important as mapping shared rivers.

By placing South America’s drainage networks beneath the lens of atmospheric moisture transport, the research turns an invisible movement of water vapor into a visible part of the continent’s geography. The study’s central message is that rivers do not begin only where water touches the ground. They also begin in the atmosphere, in moving corridors that connect oceans, forests, mountains and cities. As climate pressures grow, tracing those aerial rivers may become essential for predicting where water will arrive, when it will arrive and how powerfully it will reshape the landscapes below.

Subject of Research: Hydrological regimes and atmospheric moisture transport through aerial rivers across South America

Article Title: Hydrological regimes and drainage systems of aerial rivers across South America

Article References: Weng, W., Fu, P., Hung, H.T. et al. “Hydrological regimes and drainage systems of aerial rivers across South America.” Nature Communications (2026). https://doi.org/10.1038/s41467-026-76303-y

Image Credits: AI Generated

DOI: 10.1038/s41467-026-76303-y

Keywords: aerial rivers, atmospheric rivers, South America, hydrology, drainage systems, atmospheric moisture transport, Amazon Basin, Andes, climate change, extreme rainfall