The scale of the challenge is enormous. The Furnas drainage basin covers roughly 52,500 square kilometers, and the reservoir itself spans 1,440 square kilometers at an altitude of about 830 meters, supplying water, sanitation, leisure and tourism to dozens of neighboring municipalities. Around the water, seasonal semideciduous forest has been converted to approximately 55 to 60 percent agriculture and pasture. Coffee plantations cluster near the reservoir, soybean and citrus operations dominate areas farther away, and the most heavily disturbed sites sit close to urban centers. Hydropower generation, agribusiness, fisheries, water supply and reservoir tourism all compete within the same basin, and until now no assessment could weigh their combined effects on the ecosystem at a resolution useful to managers.

The research team, led by scientists at the Federal University of Minas Gerais in partnership with the electric utility Axia Energia, NGOs and state and federal agencies, deliberately assembled the framework from established methodologies rather than inventing one from scratch. From causal-chain accounting schemes such as the Pressure–State–Response and DPSIR frameworks, they borrowed the ordering of the assessment, so that socioeconomic drivers, catchment pressures and biological state occupy defined, separable positions. From the reference-condition tradition of multimetric biological assessment, pioneered by the Index of Biotic Integrity, they took the machinery for measuring ecological state against least-disturbed benchmarks. From composite-indicator methodology they adopted normalization, aggregation rules and the requirement that alternative weighting schemes be tested. And from social–ecological systems science they adopted genuine co-production, with specialists from multiple disciplines, the utility company, NGOs and public agencies jointly deciding which metrics matter.

The fieldwork itself was designed to be statistically defensible. Using a spatially balanced random survey design known as GRTS, originally developed for the US Environmental Protection Agency’s national stream surveys, the team sampled 40 stream sites in low-order watercourses and 40 littoral sites around the reservoir perimeter. At each site they measured water quality parameters in situ, including dissolved oxygen, pH, turbidity, conductivity and chlorophyll-a, alongside laboratory determinations of nutrients, bacteria and biochemical oxygen demand. They quantified physical habitat structure using roughly 50 metrics covering channel morphology, riparian vegetation, substrate and shoreline disturbance. And they sampled biodiversity exhaustively: benthic macroinvertebrates collected with kick-nets, fish captured with seines and hand nets, and crustaceans and molluscs identified to the lowest possible taxonomic level, with all specimens deposited in a reference collection at the university.

From this mountain of raw data, the biological component of the index was distilled through a rigorous screening funnel. Starting from 191 candidate metrics, sequential filtering for range, discrimination, responsiveness and redundancy yielded nine final metrics describing richness, tolerance, non-native taxa, diversity and life-history traits, which were scaled from 0 to 100 and averaged into a Multitaxa Multimetric Index. The framework’s authors stress that raw abundances never enter the index directly; every count is first aggregated into site-level metrics, a deliberate and documented simplification that preserves exactly the information, such as species-level tolerance and native status, needed to distinguish reference from degraded sites.

The physical environment contributed three further components. A Water Quality Index, adapted by Minas Gerais state authorities, summarized nine chemical and microbiological parameters. An Integrated Disturbance Index combined local riparian disturbance with catchment-scale land use, weighting urban and mining areas four times, agriculture twice, and pasture once. And an Environmental Fragility Index, built through the Analytic Hierarchical Process within a geographic information system, overlapped natural landscape factors such as rainfall, slope, geology and elevation with anthropogenic pressures including road density, proximity to highways and remaining natural cover. The consistency ratio of the weighting matrix was 0.0049, indicating highly coherent expert judgments. Socioeconomic data from the Brazilian Demographic Censuses, the 2017 Agricultural Census and municipal GDP accounts supplied the final domain, capturing population density, agricultural production, tourism revenue and fish farming.

All nine components were standardized to a common 0-to-1 scale, aligned so that higher values always meant better environmental quality, and summed without differential weighting. The team tested alternatives, including principal component analysis, multicriteria weighting and z-score normalization, but concluded that assigning weights at this early stage could introduce artificial biases. The resulting index was then classified into four management regions using its empirical quartiles: very bad areas below 4.4536 designated as cold spots requiring immediate restoration, priority restoration areas between 4.4536 and 4.9155, conservation hotspots between 4.9155 and 5.6470, and reference areas of Maximum Ecological Potential above 5.6470 serving as long-term monitoring benchmarks.

The maps that emerged tell a striking story. Degradation cold spots clustered in the southern and southwestern basin, coinciding with intensive agriculture, poor water quality, high environmental fragility and severe siltation. The Sapucaí branch of the reservoir emerged as a particular invasion hotspot, showing higher richness and biomass of non-native fish than the Grande branch, a pattern the authors link to cage aquaculture and shorter water residence times. By contrast, the best-preserved sites lay within Serra da Boa Esperança State Park, where dense riparian canopy shades streams, water quality is high, and sensitive aquatic insects such as mayflies, stoneflies and caddisflies flourish alongside rare native fish. The least-disturbed benchmark sites coincided with the legally protected area, exactly as the team had predicted.

Perhaps the most consequential findings concern invasive species. Of nine non-native species detected in the reservoir, eight showed significant correlations with measurable human pressures. The peacock bass Cichla kelberi tracked human water consumption; the tilapia Coptodon rendalli avoided forested buffers; the aquarium-trade guppy Poecilia reticulata was associated with irrigation projects in streams; and the invasive golden mussel Limnoperna fortunei, an oligotrophic specialist with enormous ecological and economic impacts, correlated positively with groundwater extraction permits. Agribusiness emerged as the dominant correlate of invasion across the basin, consistent with the way intensive agriculture degrades habitat through fine-sediment accumulation, eutrophication and oxygen depletion. Because the design is correlational, the authors caution that causality cannot be proven within this system, but the spatial alignment is difficult to ignore.

The framework’s practical promise lies in its transferability. Because hydropower supplies close to half of South America’s electricity and dozens of regional reservoirs resemble Furnas, the authors argue that the analytical process, indicator selection, standardization, aggregation and spatial classification, can be replicated elsewhere, provided metrics and thresholds are recalibrated to local ecological, socioeconomic and regulatory conditions and comparable partnerships between researchers, utilities and agencies are in place. Their recommendations span the full gradient of degradation: basic sanitation and erosion control in the worst areas, riparian restoration and agroecology in intermediate zones, and long-term monitoring, citizen science and strengthened conservation corridors in the best. As companies worldwide adopt nature-positive commitments, the Furnas index offers something rare in freshwater science: a single number, grounded in field data rather than satellite proxies, that tells decision-makers exactly where to act first.

Subject of Research: A transdisciplinary framework for assessing human impacts and sustainability in tropical hydropower reservoirs

Article Title: A transdisciplinary framework for assessing human impacts and sustainability in tropical hydropower reservoirs

Article References: Callisto, M., de Castro Solar, R. R., Manzano, F. V., Linares, M. S., Pompeu, P. S., Domingues, G. F., Macedo, D. R., Mascarenhas Alves, C. B., Salvador, G. N., Sulzbacher, R., Caiafa, L., Golgher, A. B., Monteiro Amaral, P. H., de Oliveira Tourinho, T. C., Formagio, P. S., de Pádua Bueno, A. A., Madureira, K. H., & Rocha, A. S. (2026). A transdisciplinary framework for assessing human impacts and sustainability in tropical hydropower reservoirs. Environmental and Sustainability Indicators, 32, Article 101493. https://doi.org/10.1016/j.indic.2026.101493

Image Credits: AI Generated

DOI: 10.1016/j.indic.2026.101493

Keywords: hydropower, freshwater biodiversity, invasive species, environmental indicators, Furnas reservoir, Brazil, multimetric index, water quality, agribusiness, ecosystem services, transdisciplinary assessment, river basin management