Ecosystem Health of Andean–Amazonian Rivers: Integrating Macroinvertebrate Diversity, Microbiological Loads and Chemical Signatures Across Anthropogenic Gradients

Abstract

The Western Amazon is a global biodiversity hotspot, yet the Upper Napo River Basin (UNRB) remains understudied regarding aquatic ecosystem health along anthropogenic gradients. We integrated benthic macroinvertebrate assemblages with physicochemical and microbiological indicators across 45 sites to assess ecological quality under four impact scenarios: Few Threats (FT, reference sites; n = 6), Crop/Aquaculture (CA; n = 22), Gold Mining (GM; n = 10), and Wastewater Discharge (WD; n = 7). Analysis of 2285 individuals (62 families) revealed clear degradation across the anthropogenic gradient. Reference sites (FT) exhibited high integrity (q₀ = 24.3 families), establishing the regional baseline for Andean–Amazonian freshwater ecosystems. In stark contrast, GM sites showed catastrophic defaunation (q₀ = 9.9 families) coupled with extreme turbidity (1320 ± 1589 NTU) and heavy metal mobilization (Fe: 430 ± 229 µg/L; Cu: 338 ± 128 µg/L), placing these reaches in “Bad” ecological status (Ecological Quality Ratio, EQR ≤ 0.16). Wastewater sites reached critical fecal coliform levels (33,708 ± 58,047 CFU/100 mL)—165-fold higher than FT sites—indicating severe sanitary impairment and community collapse (EQR = 0.28, dominated by Chironomidae at 80%). The application of ASPT (Average Score Per Taxon) and EQR proved essential for detecting functional shifts toward tolerant assemblages even when raw biotic scores appeared moderate. Crop/Aquaculture sites showed intermediate degradation (EQR = 0.37–0.38), reflecting chronic pesticide exposure and habitat loss. We conclude that gold mining and wastewater discharge are the primary drivers pushing the UNRB toward ecological collapse, with GM exerting the most severe impact on aquatic biodiversity. Safeguarding this global freshwater stronghold requires immediate implementation of multimetric biomonitoring, enhanced mining regulation, wastewater treatment infrastructure, and establishment of Indigenous-led fluvial reserves to maintain long-term connectivity.

1. Introduction

The quality of freshwater habitats is determinant of the richness, distribution, and abundance of aquatic organisms [1]. However, human activities such as mining, agriculture and lack of water treatment pose an increasing threat to aquatic ecosystems, jeopardizing their ecological health, and the benefits they provide to humanity [2,3]. As highlighted by recent research [4], the Western Amazon is critical for the basin’s integrity, harboring 74% of its fish biodiversity and contributing most of the sediment load that reaches the Atlantic Ocean. Despite this, the region is approaching a functional collapse threshold where deforestation and forest degradation could affect up to 47% of the total area by 2050 [5]. Rivers offer essential ecosystem services that underpin societal well-being, yet these depend on the environmental integrity and efficient management of water resources [6]. Sustainable water use practices and water quality monitoring are critical for ensuring these services, particularly in regions experiencing significant anthropogenic pressure [3,7].

Biological monitoring helps track changes in water quality through ecological responses of indicator species [7,8]. Standard physicochemical parameters often fail to capture the full extent of degradation, as biotic indices frequently identify ‘poor’ water quality even when chemical probes meet national standards [9]. This is especially critical given the extreme climatic events of 2023 and 2024, the hottest years on record, which led to severe droughts that concentrated pollutants in the Amazonian–Ecuadorian sector [10,11,12]. Owing to their sensitivity to pollutants, macroinvertebrates effectively reveal ecosystem alterations [13,14]. Biotic indices, including the Biological Monitoring Working Party-Colombia (BMWP-Col), Average Score Per Taxon (ASPT), and Andean–Amazon Biotic Index (AAMBI), have proven effective in distinguishing impacted sites in the Neotropics [14,15], highlighting biodiversity loss and shifts toward tolerant assemblages in polluted environments [16].

The Ecuadorian Amazon is a global biodiversity hotspot [17] and a crucial source of freshwater and climate regulation [18]. Historically shaped by oil and timber extraction, the region is currently facing a rapid expansion of illegal gold mining. By September 2024, confirmed mining expansion across the Amazon reached 37,109 hectares, with significant new fronts detected in Ecuador near the Cofán Bermejo Ecological Reserve [19]. In the Napo province, mining-related deforestation in the Punino area alone increased by 420 hectares during the first half of 2024, reaching a cumulative impact of 1422 hectares in 2019, particularly in the Punino and Anzu River areas [19]. This expansion is the primary driver of mercury contamination, with approximately 150 tons discharged annually into Amazonian rivers [11], threatening biodiversity as well as the health of both indigenous and local non-indigenous communities [9,20]. Simultaneously, agricultural expansion has further degraded these systems; studies in the Napo watershed show that streams in oil palm monocultures exhibit significantly lower family richness than traditional polycultures [21]. Additionally, untreated domestic wastewater in high-altitude Andean watersheds often leads to hypoxic conditions, with dissolved oxygen levels dropping below the 80% regulatory threshold [22].

The Napo River constitutes a significant binational aquatic system and a primary tributary of the Amazon River, acting as a vital corridor for the transport of Andean-derived sediments and nutrients toward the Atlantic [23]. The Napo River Basin (NRB) supplies water to more than 2.5 million people and supports 40% of Ecuador’s electrical energy [14]. Despite its importance, the NRB is threatened by a combination of riverbank deforestation, agricultural expansion, oil extraction, and mining [8,24]. Recent assessments report the severe influence of gold mining, urban discharges, and leachates from non-functional landfills on aquatic biota [25]. Furthermore, contamination by emerging pollutants and microplastics reflects inadequate waste management across the region [26,27,28].

While research in the NRB has historically focused on its main stems, the vast network of lower-order tributaries lacks systematic monitoring. This ecosystemic bias results in a significant knowledge gap regarding freshwater biodiversity and water quality in these critical headwaters. This knowledge gap constrains the assessment of human pressures on riverine ecosystems and hampers the development of effective management and restoration strategies [29]. Therefore, this study aimed to evaluate the ecological quality of the Upper Napo River Basin (UNRB) along an anthropogenic gradient, categorizing sites according to dominant pressures: Crop/Aquaculture (CA), Gold Mining (GM), Wastewater Discharge (WD), and Few Threats (FT). We integrated aquatic macroinvertebrate community assessments with comprehensive chemical and biological indicators. Specifically, we combined multi-taxa diversity indices with microbiological loads (fecal coliforms) to account for sanitary-related impacts, alongside nutrient enrichment and metal mobilization (iron and copper). By adopting the Ecological Quality Ratio (EQR) framework, this multi-proxy approach provides a high-resolution diagnostic baseline to identify areas of greatest impact and to assess how distinct biological and microbiological pressures threaten the health of Andean–Amazonian freshwater ecosystems, thereby supporting informed management and restoration efforts.

2. Materials and Methods

2.1. Study Area

The Napo River Basin is a major transboundary watershed of approximately 100,520 km² distributed among Ecuador (59.6%), Peru (40%), and Colombia (0.4%), and encompasses a marked altitudinal gradient from about 100 up to 6000 m above sea level. Roughly one third of this basin is situated between 200 and 6000 m a.s.l., where steep Andean slopes support a mosaic of high-gradient rivers and associated terrestrial ecosystems in the eastern cordillera of northern Ecuador and southern Colombia. Downstream, the fluvial network traverses extensive lowland rainforests in eastern Ecuador and northeastern Peru before joining the main course of the Amazon River near 100 m a.s.l. Within Ecuador, the basin intersects the Cordillera Real (Eastern Andes), a sector characterized by an abrupt drop in elevation from high mountains to foothills—on the order of several thousand meters over a horizontal distance of about 100 km—and occupying close to one fifth of the country’s eastern region [30,31,32]. This study focuses on the UNRB (~1918 km²), a sub-basin spanning diverse ecosystems from páramo to piedmont rainforest (Figure 1). Draining Andean rivers through protected areas into Amazon tributaries [33], the study sites follow an elevational gradient from 250 to 650 m a.s.l. [25,26,34]. The Napo contributes a mean annual discharge of ~6300 m³/s to the Amazon [32]. This basin is characterized by high precipitation rates, often exceeding 4000 mm/yr in the Andean foothills, which drives significant sediment transport and hydrological connectivity [4,25,26,27,28,35]. For comprehensive details on sampling sites and their anthropogenic impacts, see

Table A1. Detailed characterization of the 45 sampling sites in the Upper Napo River Basin. The table includes geographic coordinates (UTM Zone 18S, WGS84 datum), river/stream information, surrounding land-use activities, assigned anthropogenic impact categories, and the respective data sources (including both unpublished research and peer-reviewed literature).

Sample No.	River/Stream Information	Surroundings Activities	Anthropic Impact	Coordinates (UTM 18S)		Data Source
				X	Y
S1	Via Ahuano	Little Maize crop	Few Threats	212168	9883964	This study (unpublished)
S2	Canuayaca/Via Ahuano	Crop: Maize, Banana, Cacao, Yuca, Sugarcane	Crop or Aquaculture	200008	9883394	This study (unpublished)
S3	Ahuano/San Pedro	Crop: Maize, Banana, Rice, Cacao	Crop or Aquaculture	212786	9879281	This study (unpublished)
S4	Chontapunta	Crop: Rice, Cacao	Crop or Aquaculture	232449	9895107	This study (unpublished)
S5	Chontapunta	Crop: Banana, Cacao	Crop or Aquaculture	230953	9892469	This study (unpublished)
S6	Chontapunta	Crop: Banana, Cacao	Crop or Aquaculture	229652	9890805	This study (unpublished)
S7	Comunidad Chambiro (vía Muyuna)	Crop: Maize, Banana, Cacao, Orillo	Crop or Aquaculture	183954	9890763	This study (unpublished)
S8	Puerto Napo	Crop: Maize, Banana, Yuca, Cacao	Crop or Aquaculture	190730	9881909	This study (unpublished)
S9	Pashimbi Stream	Crop: Orillo	Crop or Aquaculture	181174	9895042	This study (unpublished)
S10	Hatun Sumaku	Crop: Naranjilla	Crop or Aquaculture	210983	9926019	This study (unpublished)
S11	Sumaco Pucuno	Crop: Naranjilla	Crop or Aquaculture	210932	9921730	This study (unpublished)
S12	Marchángara	Crop: Naranjilla	Crop or Aquaculture	199941	9919766	This study (unpublished)
S13	Cotundo (Sardina River)	Crop: Naranjilla	Crop or Aquaculture	189784	9916567	This study (unpublished)
S14	Arosemena Tola	Crop: Banana, Cacao, Coffea, Guayaba, Lemon, Orange, Tangerine	Crop or Aquaculture	181437	9869740	This study (unpublished)
S15	Arosemena Tola	Crop: Cacao	Crop or Aquaculture	180874	9873331	This study (unpublished)
S16	Tributary stream of the Napo River: located in Puerto Napo, near the Jatunyacu river	Fish Farming	Crop or Aquaculture	186419	9883804	[25]
S17	Tena River Tributary	Gold Mining	Gold Mining	180438	9896700	[25]
S18	Estero Paushiyacu	Urban Areas	Wastewater Discharge	187167	9889428	[25]
S19	Tena Landfill, Misahualli River Tributary	Landfill	Wastewater Discharge	186311	9896648	[25]
S20	Pashimbi Stream	With few threats: Located within an agricultural matrix, but no direct point source identified	Crop or Aquaculture	181172	9895033	[25]
S21	Morete Cocha	The mining site was located within a forested area on a small stream that was heavily impacted by extraction activities. Effluents from the tailings pond were discharged directly into the channel, and both machinery and workers were operating at the time of sampling.	Gold Mining	181793	9877381	[26]
S22	Estrella del Oriente	The sampling point was located upstream of an area where mining activity was replaced by tilapia pools. The vegetation on the banks is secondary.	Crop or Aquaculture	179919	9873989	[26]
S23	Estrella del Oriente	Mining area: effluents from the tailings pond were being released into the stream, and both machinery and personnel were actively operating at the site during sampling.	Gold Mining	179888	9873589	[26]
S24	Río Chumbiyacu	A road has been built over the river, specifically constructed to provide access for the mining machinery.	Gold Mining	180783	9876572	[26]
S25	Shiguacocha	Mining area: the sampling reach received runoff from upstream mining operations and corresponded to an abandoned extraction site, where vegetation was beginning to recolonize soils previously altered by mining activities.	Gold Mining	184371	9877252	[26]
S26	Rio Chumbiyacu	Mining area: the sampling site received wastewater discharges from upstream mining operations.	Gold Mining	186691	9877900	[26]
S27	Río Huambuno	Mining area: the sampling site is located upstream of an abandoned mining operation.	Gold Mining	220682	9890162	[26]
S28	Río Huambuno	Mining area: the sampling site receives wastewater discharges from nearby mining operations	Gold Mining	222877	9891792	[26]
S29	Río Tuyano	Mining area: the sampling reach was heavily impacted by mining activities; the riverbed had been completely reshaped to create tailings and settling ponds and to flush alluvial sediments. Heavy machinery was operating at the site during sampling.	Gold Mining	209735	9884928	[26]
S30	Río Yutzupino: near Puerto Napo City	Mining area: the sampling point receives the sewage from mining areas	Gold Mining	187088	9883802	[26]
S31	Toglo river	Native Vegetation	Few Threats	188431	9888314	This study (Karst systems survey)
S32	Castillo stream: Santa Rosa	Transition zone from rural to urban land use near a gas station. Potential impacts on the stream include runoff from a secondary road and discharges from the local sewer system.	Wastewater Discharge	187993	9885936	This study (Karst systems survey)
S33	Castillo stream: Santa Rosa	The site is situated adjacent to a highway and receives direct wastewater discharges caused by damaged sewer infrastructure.	Wastewater Discharge	188155	9886152	This study (Karst systems survey)
S34	Toglo River: Santa Rosa	Discharge: downstream of gold mining, dredging, agricultural activity, and sewage discharges	Wastewater Discharge	188180	9885933	This study (Karst systems survey)
S35	Wamahurco: Norma Aguinda Family	Natural Spring	Few Threats	189531	9892965	This study (Karst systems survey)
S36	Wamahurco: Centro Comunitario Kichwa Tamia Yura	Visible pressures on the site include agricultural activity, wastewater discharges, and tourism in and around the cave	Wastewater Discharge	188266	9892533	This study (Karst systems survey)
S37	Small stream, discharges directly into the Napo River; Nuevo Paraíso community	Corn Crop	Crop or Aquaculture	224405	9887295	[34]
S38	San Pedro de Sumino Community, Sumino River	Corn Crop	Crop or Aquaculture	230851	9892654	[34]
S39	Seasonal stream: near the Arajuno River	Corn Crop	Crop or Aquaculture	212941	9878992	[34]
S40	Puno river: Ahuano	Corn Crop	Crop or Aquaculture	211656	9881921	[34]
S41	Small stream, discharges directly into the Anzu River; “Comunidad Intercultural Naranjalito”	Corn Crop	Crop or Aquaculture	188189	9882211	[34]
S42	Estero Mamallacta	Discharge: sewage discharge on Pano River	Wastewater Discharge	186577	9889257	[34]
S43	Few Threats zone, near Kawsay Yaku spa	Pristine Zone, Few Threats zone	Few Threats	179106	9896248	[34]
S44	Colonso River	Pristine Zone, Few Threats zone	Few Threats	177814	9895309	[34]
S45	Lupi River; El Calvario-Muyuna	Pristine Zone Few Threats zone	Few Threats	176721	9892228	[34]

in Appendix A.

2.2. Sampling and Dataset Integration

A total of 45 sampling sites were selected across the Upper Napo River Basin (UNRB) to represent a comprehensive gradient of anthropogenic pressure (Table A1). To ensure a robust regional assessment, this study integrated primary data from unpublished field surveys (21 sites) [35] with high-quality secondary data (24 sites) sourced from previous peer-reviewed studies conducted in the region [25,26,34]. All macroinvertebrate collections and physicochemical measurements across the 45 sites were conducted by the authors and co-authors of this study, ensuring consistent sampling protocols and taxonomic identification standards within the same research institution. To ensure environmental comparability and the validity of the Ecological Quality Ratio (EQR) framework, all sampled reaches were standardized to low-order streams (1st to 3rd order) with wetted widths not exceeding 10 m. Field campaigns were carried out between February and December 2020 and in November 2021, consistently scheduled at the onset of the dry season to avoid high river discharges, which can redistribute substrate material and transport aquatic macroinvertebrates, thereby altering their community composition. This study presents a novel multi-metric synthesis by integrating diverse biological and physicochemical datasets. While previous research in the region focused on specific stressors—such as pesticides [34], mining toxicity [26], or multi-line indices [25]—our approach expands upon these findings. We apply a unified framework using Hill numbers, Average Score per Taxon (ASPT), and Ecological Quality Ratio (EQR) to assess the entire anthropogenic gradient. This meta-analytical approach allows for the identification of ecological thresholds that were not detectable in the isolated assessments of the original studies. Sites were categorized into four distinct impact groups based on dominant surrounding activities: Few Threats (FT, n = 6), Crop or Aquaculture (CA, n = 22), Gold Mining (GM, n = 10), and Wastewater Discharge (WD, n = 7).

Environmental and biological data from 45 sites were harmonized by converting variables to common units and aggregating macroinvertebrates to the family level [36,37,38] to ensure consistency across datasets (Table A1). At each site, in situ physicochemical parameters (temperature, pH, electrical conductivity, and dissolved oxygen) were recorded using a calibrated YSI Professional Plus multiparameter probe (Yellow Springs Instruments, Yellow Springs, OH, USA). Additionally, surface water samples were collected for laboratory analysis of turbidity and fecal coliforms. Extended chemical and microbiological signatures—including nutrients, BOD, sulfates, iron, and trace metals—were integrated from previously published studies [25,26,34], where detailed laboratory protocols for these specific parameters are described. Macroinvertebrates were sampled using a standardized multi-habitat method with a 500 µm D-frame net over a 25-m stretch for 3 min, following established regional protocols [7,25,26,34,35,36].

To address uneven replication and missing values (denoted by ‘—’ in

Table A2. In situ physicochemical water quality parameters recorded across the sampling sites and compared with international regulatory standards.

Sampling Research	Sampling Number	Anthropic Impact	Physicochemical Parameters						Fecal Coliforms (NMP/100 mL)
			Temperature (°C)	DO (% Sat)	Turbidity (NTU)	TDS (mg/L)	pH	Conductivity (µS/cm)
S1	S1	FT	24	83	2.85	17	7.45	34	-
	S2	CA	24	80	2.26	49.5	7.3	90	-
	S3	CA	24.4	80	55.8	24	4.11	48	-
	S4	CA	24.3	56.2	6.42	28	7.66	55.8	-
	S5	CA	25.3	90.4	2.44	23	6.91	48.5	-
	S6	CA	26.2	102	2.45	31.5	7.11	63.1	-
	S7	CA	24.4	98.6	1.11	15.3	7.9	30.6	-
	S8	CA	24.2	105.15	0.882	15.5	7.47	31.3	-
	S9	CA	23	106.4	0.376	15.5	7.57	31.5	-
	S10	CA	18.3	91.15	0.33	11.3	7.77	22.9	-
	S11	CA	19.5	94.75	0.578	32.5	7.76	64.8	-
	S12	CA	20.6	84	1.2	18	7.59	35.2	-
	S13	CA	21.3	42.75	4.92	7	5.43	14.3	-
	S14	CA	23.1	88.4	22.3	13.5	7.34	26.9	-
	S15	CA	27.2	52.5	1.55	6.3	6.48	13	-
S2	S16	CA	24.9	90.1	4.56	15.5	8.5	-	500
	S17	GM	23	95	40.34	16.5	8.06	-	700
	S18	WD	23.3	57.8	43.4	69	7.53	-	10,000
	S19	WD	24.9	19	46.6	2350	8	-	1000
	S20	CA	22.4	86.2	0.326	22.4	7.8	-	100
S3	S21	GM	23.3	80.6	765	45.5	6.91	67.6	-
	S22	CA	25.6	81.5	10.2	16.26	6.7	25.3	-
	S23	GM	30	75	277	14.95	6.55	25	-
	S24	GM	28.3	76.2	24	33.2	6.67	53.8	-
	S25	GM	31	78.4	246	31.2	7.17	52.1	-
	S26	GM	29.2	76.6	1457	27.3	6.8	45.8	-
	S27	GM	30	74.7	37.3	42.2	6.61	70.9	-
	S28	GM	28.1	76.9	28.2	115	8.06	187.2	-
	S29	GM	26.5	56.5	339	96.2	7.18	152.3	-
	S30	GM	25.1	54.5	5026	57.85	7.37	88.6	-
S4	S31	FT	22.8	107.3	4.46	68	7.8	130.7	208
	S32	WD	24.1	107.9	53	37	6.03	72.1	126
	S33	WD	24.4	134.3	15.7	53	7.71	104.8	800
	S34	WD	25.8	115.6	16.3	57.5	7.77	136.9	501
	S35	FT	22.8	110.7	9.2	84	7.93	160.5	197
	S36	WD	21.9	123.3	3.43	17.5	7.98	221.2	12
S5	S37	CA	23.4	70	-	55.25	7.51	83.3	-
	S38	CA	25	82	-	30.55	7.7	46.6	-
	S39	CA	24.7	2.3	-	46.75	6.68	84.8	-
	S40	CA	25.5	73.8	-	31.85	8.4	49.8	-
	S41	CA	22.7	74.1	-	28.3	7.65	40.1	-
	S42	WD	25	55.5	-	177.45	7.57	432	-
	S43	FT	20.4	82.7	-	22.75	8.36	32.2	-
	S44	FT	18.9	87.7	-	21.45	8.08	29.6	-
	S45	FT	19.7	86	-	30.55	7.9	42.3	-
CCME			22.5-27.5	>80	-	500	6.5–8.5	500	No guideline
TULSMA			22–28	>80	-	1000	6.5–9	1000	200
US EPA			22–28	>80	-	500	6.5–9	500	200

Note: Bold values indicate concentrations that exceed the maximum permissible limits established by both Ecuadorian national regulations (TULSMA) and international water quality standards (US EPA and CCME).

and

Table A3. Detailed microbiological, nutrient, ion, and trace metal parameters in the upper Napo Basin.

Site	Fecal Coliform	NO3	TP	BOD	NH4	NO2	SO4	Na	K	Mg	Ca	As	B	Ba	Cd	Cr	Cu	Fe	Pb	Ag	Al	Be	Co	Mn	Ni	Se
S1	-	0.4	0.08	-	0.07	0.07	0.56	2.52	0.79	1.23	2.42	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S2	-	0.49	0	-	0.07	0.1	0.79	6.47	1.1	2.59	8.75	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S3	-	0.43	0.1	-	0.13	0.05	0.49	2.82	1.07	1.69	3.39	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S4	-	0.45	0.12	-	0.11	0.12	0.59	3.71	1.69	1.84	3.95	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S5	-	0.29	0.11	-	0.03	0.09	0.55	3.46	1.49	1.35	3.02	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S6	-	0.41	0.11	-	0.08	0.08	0.78	2.08	1.33	2.75	5.73	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S7	-	1.01	0.11	-	0.23	0.01	1.45	1.83	0.76	0.62	1.76	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S8	-	0.44	0	-	0.06	0	0.69	1.27	0.6	1.12	2.25	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S9	-	0.43	0.12	-	0.12	0.02	1.62	2.81	0.42	0.46	1.99	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S10	-	0.49	0.09	-	0.14	0.07	0.74	0.83	0.88	0.53	1.42	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S11	-	0.33	0.13	-	0.16	0.02	3.28	4.76	2.81	1.29	4.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S12	-	0.28	0.09	-	0.09	0.02	1.09	2.18	1.04	0.86	2.37	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S13	-	0.34	0.1	-	0.14	0.02	1.29	0.32	0.39	0.28	0.61	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S14	-	0.09	0.2	-	1.43	0.43	0.11	0.24	0.15	0.07	0.17	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S15	-	0.02	0.04	-	0	0.19	0.07	1.5	0.38	0.76	2.22	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S16	500	0.4	0.57	30.45	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S17	700	1.2	0.68	394.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S18	10,000	0.4	1.81	69	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S19	1000	0.01	2.63	346.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S20	100	0.7	0.27	4.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S21	-	-	-	-	-	-	-	-	-	-	-	2.2	<4.0	345.4	0.5	<2.3	24.2	371.9	6.1	0.2	158.1	0.6	2.2	456.3	5.8	0.7
S22	-	-	-	-	-	-	-	-	-	-	-	1.7	<4.0	32.2	0.7	<2.2	6	228.9	10.5	<0.0	132.6	<0.1	0.5	140.5	3.3	0.6
S23	-	-	-	-	-	-	-	-	-	-	-	6.7	<4.0	113.1	0.2	2.6	24.2	558.5	5.7	0.3	287.7	0.2	1.9	456.3	4.3	<0.4
S24	-	-	-	-	-	-	-	-	-	-	-	1.4	<4.0	60.5	0.2	<2.2	6.4	227.5	0.7	<0.0	93.1	0.1	0.5	105.5	2.3	<0.4
S25	-	-	-	-	-	-	-	-	-	-	-	2.3	<4.0	153.8	0.2	<2.2	8	326.1	2.4	0.1	231.2	0.2	1.8	274.7	3.1	0.8
S26	-	-	-	-	-	-	-	-	-	-	-	1.8	<4.0	817.2	0.4	<2.2	12	237	14.5	<0.0	377.6	0.4	4.5	485.6	1.9	0.8
S27	-	-	-	-	-	-	-	-	-	-	-	3.4	<4.0	46.1	0.2	<2.2	5.6	512.8	1.7	0.1	146.9	<0.1	0.4	49.4	5.8	0.9
S28	-	-	-	-	-	-	-	-	-	-	-	2.8	<4.0	61.1	0.8	2.8	11.3	547.6	1.7	0.8	253.1	0.1	0.6	74.4	5.7	0.8
S29	-	-	-	-	-	-	-	-	-	-	-	2.7	<4.0	222.7	0.4	<2.2	10.8	536.3	2.4	<0.0	249.1	0.2	1.7	513.2	4.9	0.8
S30	-	-	-	-	-	-	-	-	-	-	-	5.7	9.8	938.9	0.5	<2.2	9.4	371	4.4	0.1	350.3	0.9	7.7	597.9	8.7	0.7
S31	208	2.3	0.47	-	-	-	-	0.4	4.09	0.84	22.81	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S32	126	2.3	0.35	-	-	-	-	3	1.54	0.93	5.11	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S33	800	0.6	0.13	-	-	0.01	12	-	0.92	1.2	35.79	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S34	501	0	0.34	-	-	0.6	-	4	2.26	1.42	12.12	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S35	197	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S36	12	0.8	0.79	-	-	0.01	0	-	0.42	0.98	47.33	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S37	-	-	-	-	1.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S38	-	-	-	-	1.76	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S39	-	-	-	-	0.04	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S40	-	-	-	-	0.15	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S41	-	-	-	-	10.56	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S42	-	-	-	-	9.12	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S43	-	-	-	-	26.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S44	-	-	-	-	130.27	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
S45	-	-	-	-	5.92	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-

Note: Units for the parameters reported in this dataset are as follows: Fecal Coliform in Colony Forming Units per 100 mL (CFU/100 mL); Nutrients (Nitrates-NO₃, Total Phosphate-TP, Ammonium-NH₄, Nitrite-NO₂, Sulfate-SO₄), Biochemical Oxygen Demand (BOD), and Major Ions (Sodium-Na, Potassium-K, Magnesium-Mg, Calcium-Ca) in milligrams per liter (mg/L); and Heavy Metals/Trace Elements (As, B, Ba, Cd, Cr, Cu, Fe, Pb, Ag, Al, Be, Co, Mn, Ni, Se) in micrograms per liter (µg/L). The dash (-) denotes parameters that were not recorded at specific sites due to logistical accessibility constraints or technical limitations during field campaigns in remote Amazonian reaches. Heavy metal concentrations in sites affected by gold mining (GM) were analyzed to determine the multi-elemental signature of the activity.

) typical of remote Amazonian hydrological conditions, we employed a Multiple Lines of Evidence (LOE) framework. This approach integrates high-resolution biotic responses—including Hill numbers, ASPT, and Ecological Quality Ratios (EQR)—with available chemical signatures. To ensure mathematically valid diversity comparisons, coverage-based rarefaction and extrapolation were applied, providing a robust diagnostic assessment of ecological thresholds across the anthropogenic gradient regardless of the unbalanced design.

2.3. Data Analysis

2.3.1. Ecological Water Quality Indices

Four biotic indices were calculated for each sampling site: AAMBI, BMWP-Col, EQR-A, and EQR-B. The Andean–Amazon Biotic Index (AAMBI) [13] assigned tolerance values to macroinvertebrate families from 1 (contamination-tolerant) to 10 (highly sensitive) [26,35], classifying water quality into five categories: excellent (>100), very good (75–90), good (50–74), regular (25–49), and bad (0–24) [14,25].

The Colombian adaptation of the Biological Monitoring Working Party index (BMWP-Col) was also applied, as Colombia and Ecuador share relatively similar environmental conditions, making this regional index an appropriate reference for ecological water quality assessments in Ecuadorian rivers. The BMWP-Col index, based on macroinvertebrate community composition, assigns each taxon a tolerance score, from 1 (for tolerant taxa) to 10 (for sensitive taxa); water quality is then classified as good (≥100), moderate (61–100), poor (36–60), bad (16–35), and very bad (0–15) [14,39]. Recent studies in Ecuadorian Andean and Amazonian watersheds have re-validated the use of BMWP-Col and AAMBI, demonstrating their high sensitivity in detecting organic and mining-related pollution gradients [9,22,25]. Additionally, the Average Score Per Taxon (ASPT) was calculated for both AAMBI and BMWP-Col by dividing the total index score by the number of families identified at each sampling site. The ASPT provides a normalized measure of taxonomic sensitivity that is less influenced by sampling effort or total richness, as it represents the average quality of the taxa present [9,14,25].

Finally, to ensure international comparability and alignment with the EU Water Framework Directive (WFD) guidelines, the ecological status of each site was expressed as an Ecological Quality Ratio (EQR). The EQR represents the relationship between the observed (O) values of the biological parameters (AAMBI and BMWP-Col) and the expected (E) values under reference conditions. Two sub-indices were calculated: EQR-A (Observed AAMBI/Expected AAMBI) and EQR-B (Observed BMWP-Col/Expected BMWP-Col). EQR values range from 0 to 1, where values close to 1 represent high ecological status (reference conditions) and values close to 0 indicate severe anthropogenic impact. Ecological status was classified into five classes: High (0.90–1.00), Good (0.75–0.89), Moderate (0.60–0.74), Poor (0.45–0.59), and Bad (<0.45) [40,41].

2.3.2. Statistical Analysis

To assess alpha diversity across anthropogenic gradients (CA, GM, FT, WD), macroinvertebrate family abundance data were analyzed using Hill numbers at three diversity orders: q = 0 (family richness), q = 1 (exponential of Shannon entropy), and q = 2 (inverse Simpson diversity). Hill numbers provide a unified framework expressing diversity in units of ‘effective family equivalents’. While sampling effort was standardized across all sites, we employed this framework following Chao & Jost [42] to ensure comparisons are based on sample completeness (coverage) rather than just sample size, accounting for inherent differences in community complexity. Sample completeness was evaluated through coverage-based rarefaction and extrapolation (R/E) curves. Diversity comparisons were standardized to the lowest observed sample coverage value, with 95% confidence intervals (CI) estimated via bootstrapping (999 iterations) to determine statistical significance [42,43,44,45]. Finally, community structure was characterized using rank-abundance curves based on the logarithm (Log₁₀) of family abundances.

To examine the multivariate relationships between environmental parameters (temperature, DO%, turbidity, pH, and TDS), biological diversity (Hill numbers: q0, q1, and q2), and biotic indices (AAMBI, BMWP-Col), we performed a Principal Component Analysis (PCA) [46]. Other chemical parameters, such as nutrients and trace metals, were excluded from the PCA because they were not available for all 45 sites; their inclusion would have necessitated the exclusion of numerous sampling locations (listwise deletion), thereby reducing the statistical power and representative scope of the regional analysis. This exploratory approach allowed for the identification of the main gradients of variation and the correlation structure among variables. The number of significant principal components was determined by evaluating their respective eigenvalues, following the criteria for ecological data reduction [46,47].

Beyond this exploratory analysis, differences in community composition among impact categories were evaluated using a beta-diversity framework. We calculated Bray–Curtis dissimilarities from a community matrix with abundances square-root transformed to reduce the influence of dominant taxa. Multivariate structure was explored through non-metric multidimensional scaling (NMDS) with two dimensions (k = 2) and up to 200 random starts to ensure convergence. Ordination quality was assessed using Kruskal’s stress value. Statistical significance of these compositional differences was formally tested using a Permutational Multivariate Analysis of Variance (PERMANOVA) via the adonis2 function, based on 9999 permutations. Pairwise comparisons among land-use categories were subsequently performed, and p-values were adjusted using the Benjamini–Hochberg procedure to control the false discovery rate [48].

All statistical procedures were performed in R version 4.5.2 [49]. Hill numbers and R/E curves were generated using the iNEXT package (v3.0.2) [50]. Rank-abundance and community quantitative structure analyses were computed with the BiodiversityR package (v2.17-4) [51]. Multivariate environmental analyses were implemented via the prcomp() function from the base stats package and the PCA() function from the FactoMineR package (v2.12) [52]. Community composition differences were evaluated through NMDS and PERMANOVA using the vegan package (v2.7-2) [48]. One-way comparisons across disturbance categories used Kruskal–Wallis tests (non-parametric ANOVA) followed by Dunn’s post hoc tests (Bonferroni-adjusted vs. FT reference) implemented via kruskal.test() and dunnTest() (FSA package v0.9.5) [53]. For family-level macroinvertebrate abundances, additional pairwise contrasts among all impact categories were computed with p-values adjusted using the Benjamini–Hochberg procedure to control the false discovery rate with the emmeans package [54]. Detailed values for all monitored environmental parameters are provided in Appendix A Table A2 and Table A3, and the full taxonomic composition and mean abundances of aquatic macroinvertebrates across impact categories are summarized in Appendix A 

Table A4. Taxonomic composition and mean abundance (Mean ± Standard Deviation) of aquatic macroinvertebrates across anthropogenic impact categories in the Upper Napo River Basin.

Order	Family	Sample Sites				Total Collected
		FT (n = 6)	CA (n = 22)	WD (n = 7)	GM (n = 10)
Seriata	Planariidae	0.67 (1.6)	0	0	0	4
Tricladida	Dendrocoelidae	0	0	0.14 (0.4)	0	1
Basommatophora	Physidae	0.17 (0.4)	0	0.14 (0.4)	0	2
Sorbeoconcha	Hydrobiidae	0	0	0.14 (0.4)	0	1
Neotaenioglossa	Thiaridae	0	0	5.86 (15.5)	0	41
Architaeniglossa	Ampullariidae	0	0.18 (0.5)	0	0	4
Pulmonata	Planorbidae	0	0	1.86 (4.9)	0	13
Sphaeriida	Sphaeriidae	8.33 (6.1)	0.09 (0.3)	0.43 (1.1)	0	55
Rhynchobdellida	Glossiphoniidae	0	0.18 (0.5)	1.14 (3.0)	0	12
	Piscicolidae	0.17 (0.4)	0.86 (1.9)	3.43 (9.1)	0	44
Arhynchobdellida	Erpobdellidae	0	0	0.43 (1.1)	0	3
Decapoda	Palaemonidae	0.67 (1.2)	0.45 (0.9)	0	0	14
Acari	Hydrachnidae	0.17 (0.4)	0.05 (0.2)	0	0	2
Ephemeroptera	Baetidae	5.00 (4.2)	1.00 (1.4)	0	0	52
	Caenidae	0.33 (0.8)	0.05 (0.2)	0	0	3
	Euthyplociidae	0.33 (0.8)	0.45 (1.2)	0.86 (2.3)	0	18
	Leptohyphidae	5.00 (4.5)	0.50 (0.9)	0	0	41
	Leptophlebiidae	6.33 (5.7)	1.95 (3.1)	1.14 (2.1)	0.60 (1.3)	95
	Oligoneuriidae	0.17 (0.4)	0	0	0	1
Odonata	Aeshnidae	0.33 (0.8)	0	0	0	2
	Coenagrionidae	1.00 (1.1)	0.23 (0.6)	0	0.40 (0.7)	15
	Gomphidae	2.83 (2.1)	0	0	0.30 (0.7)	20
	Libellulidae	0	0.64 (1.5)	0.14 (0.4)	0.10 (0.3)	16
	Megapodagrionidae	0.50 (0.8)	0	0.14 (0.4)	0	4
	Platystictidae	0.17 (0.4)	0	0	0	1
	Polythotidae	0.17 (0.4)	0	0	0	1
Plecoptera	Perlidae	3.17 (2.4)	0.68 (1.2)	0	0.50 (0.7)	39
Trichoptera	Calamoceratidae	9.00 (11.2)	2.50 (4.1)	0.43 (1.1)	0	1
	Glossosomatidae	0.17 (0.4)	0	0	0	24
	Hydrobiosidae	0	0	3.43 (9.1)	0	3
	Hydropsychidae	0.33 (0.5)	0.09 (0.3)	0	0	112
	Hydroptilidae	0	0.05 (0.2)	0	0	1
	Lepidostomatidae	0.50 (0.8)	0	0	0	4
	Philopotamidae	0.33 (0.5)	0	0.14 (0.4)	0	3
Hemiptera	Belostomatidae	0	0	0.29 (0.8)	0	2
	Gerridae	0	0.05 (0.2)	0	0	1
	Hebridae	0.17 (0.4)	0	0	0	1
	Helotrephidae	0	0.05 (0.2)	0	0	1
	Naucoridae	0.17 (0.4)	0.41 (1.1)	0	0.10 (0.3)	11
	Notonectidae	0.17 (0.4)	0	0.14 (0.4)	0	2
	Veliidae	0	0.05 (0.2)	0	0	1
Megaloptera	Corydalidae	1.00 (1.1)	0.14 (0.4)	0.14 (0.4)	0.40 (0.7)	14
Coleoptera	Dytiscidae	0	0.14 (0.5)	0.14 (0.4)	0	4
	Elmidae	14.83 (9.1)	1.82 (2.4)	0	0.20 (0.4)	131
	Hydrophilidae	2.00 (1.8)	0	0	0	12
	Hydroscaphidae	0	0.14 (0.6)	0	0	3
	Psephenidae	0.67 (0.8)	0.05 (0.2)	0	0	5
	Ptilodactylidae	0.17 (0.4)	0.09 (0.3)	0	0	3
	Scirtidae	0.33 (0.8)	0	0.86 (2.3)	0	8
	Staphylinidae	0.33 (0.5)	0	0	0	2
Lepidoptera	Crambidae	0.17 (0.4)	0	0.29 (0.8)	0	3
	Erebidae	0	0.05 (0.2)	0	0	1
	Noctuidae	0	0.09 (0.3)	0	0	2
Diptera	Ceratopogonidae	1.50 (2.4)	0	0	0.30 (0.9)	12
	Chironomidae	27.83 (10.5)	21.64 (15.3)	97.57 (52.4)	0.40 (0.8)	1330
	Culicidae	0.17 (0.4)	0	1.00 (2.6)	0	8
	Dolichopodidae	0	0	0.14 (0.4)	0	1
	Empididae	0.17 (0.4)	0	0.14 (0.4)	0	2
	Limoniidae	0	0	0.57 (1.5)	0	4
	Psychodidae	0	0	0.57 (1.5)	0	4
	Simuliidae	9.50 (12.3)	0	0	0	57
	Tipulidae	1.17 (1.5)	0.05 (0.2)	0	0	8
Total abundance (individuals/site)		106.17 (48.3)	34.68 (25.1)	121.71 (68.5)	3.30 (2.1)	2285
Richness (S) (Families/site)		15.17 (4.1)	7.50 (3.2)	9.29 (2.8)	2.10 (1.5)	110

Note: CA: Crop/Aquaculture; FT: Few Threats (reference); GM: Gold Mining; WD: Wastewater Discharge (n = number of sampling sites per category). Except for ‘Total Collected’, values are mean individuals per site ± SD. Total Abundance and Richness (S) are category means of the sites within each sampling category.

.

3. Results

The environmental characterization of the UNRB revealed distinct pollution signatures across the anthropogenic gradient (

Table 1. Physicochemical and microbiological water quality parameters [Mean (SD)] across anthropogenic impact categories (Kruskal–Wallis and Dunn’s post hoc, p < 0.05) in Upper Napo River Basin.

Parameter (Unit)	FT (n = 6)	CA (n = 22)	WD (n = 7)	GM (n = 10)	Regulatory Limit (ECU/CAN)
Temperature (°C)	21.4 (2.0) *	23.6 (2.2)	24.2 (1.3)	27.4 (2.9) * †	Natural ± 3/Natural ± 3
Dissolved Oxygen (% sat)	92.9 (12.7)	78.7 (23.6)	87.6 (43.4)	74.4 (11.6)†	≥80% sat (≥6 mg/L)/≥80% sat (≥6 mg/L)
pH	7.9 (0.3) *	7.2 (1.0) *	7.5 (0.7)	7.1 (0.5) *	6.5–9.0/6.5–9.0
Turbidity (NTU)	5.5 (3.3)	6.9 (13.7)	29.7 (20.4)	824.0 (1543.3) * † #	Natural + 5 NTU/Site-specific (+5 NTU)
TDS (mg/L)	40.6 (28.2)	24.4 (13.2)	394.5 (863.8) * †	48.0 (33.3)	No specific/No specific
Conductivity (μS/cm)	71.6 (58.3)	45.3 (22.7)	193.4 (144.5) *	82.6 (53.2)	Site-specific/Included in WQI
Fecal Coliform (CFU/100 mL)	202.5 (7.8)	300.0 (282.8)	2073.2 (3901.8)	700.0	200 NMP/100 mL **/No guideline

Note: CA: Crop or Aquaculture; FT: Few Threats (reference); GM: Gold Mining; WD: Wastewater Discharge. Values are reported as Mean ± Standard Deviation (SD). For fecal coliforms in Gold Mining (GM), only one measurement was available (700 NMP/100 mL), thus no SD reported; SD omitted where single measurements preclude calculation. Statistical significance (p < 0.05): (*) indicates a significant difference compared to FT; (†) indicates a significant difference compared to CA; (#) indicates that the mean value exceeds at least one regulatory limit. Sources: TULSMA Book VI, Table 3 (criteria for freshwater flora/fauna preservation) [55]; Appendix 1 (additional limits) [55]; CAN = CCME, Canadian Water Quality Guidelines for the Protection of Freshwater Aquatic Life [56]. ** TULSMA Book VI, Appendix 1 assumes an approximate 1:1 equivalence (factor 0.9–1.1) between NMP/100 mL and CFU/100 mL for regulatory comparisons.

). Wastewater Discharge (WD) sites exhibited critical fecal coliform levels (33,708 ± 58,047 CFU/100 mL) and the highest variability in total dissolved solids (TDS), with extreme values reaching 2350 mg/L. In contrast, Gold Mining (GM) sites were characterized by extreme turbidity (1320.1 ± 1589.2 NTU) compared to reference conditions. Few Threats (FT) sites showed the lowest values for temperature (22.5 ± 1.6 °C), turbidity (5.5 ± 3.5 NTU), and fecal coliforms (203 ± 32 CFU/100 mL), justifying their selection as reference sites.

3.1. Benthic Macroinvertebrate Communities

We examined 2285 individual benthic macroinvertebrates collected, identified, and enumerated across all study sites, representing 21 orders and 62 families. Specimens were predominantly in immature stages, consistent with typical Neotropical stream assemblages. Diptera was the dominant order, comprising 62.4% (1426 individuals) of total abundance, followed by Ephemeroptera (210 individuals, 9.2%) and Coleoptera (168 individuals, 7.4%). At the family level, Chironomidae was by far the most abundant taxon (1330 individuals, 58.2% of all specimens), with markedly higher densities at Wastewater Discharge (WD) sites than at Few Threats (FT) and Crop/Aquaculture (CA) sites, as indicated by significant pairwise differences in the Benjamini–Hochberg-adjusted post hoc tests (FT–WD, CA–WD, CA–GM; p_adj < 0.05; Figure 2). In contrast, sensitive families such as Elmidae and Leptophlebiidae reached their highest abundances at FT sites and were significantly reduced or nearly absent in GM and WD streams, reflecting the strong filtering imposed by mining and wastewater impacts (Figure 2). Other relatively common families, including Hydropsychidae and Baetidae, showed intermediate responses and persisted mainly in FT and CA categories, whereas GM sites were characterized by uniformly low abundances across most families. Additional details on family-level composition are provided in Appendix A, Table A4.

3.2. Ecological Water Quality

The ecological status of the sampling sites varied significantly across the anthropic gradient. FT sites exhibited the highest biological water quality, with mean AAMBI (82.3 ± 14.5) and BMWP-Col (88.0 ± 13.5) scores indicating “Very Good” and “Good” conditions, respectively (

Table 2. Synthesis of ecological quality status based on biotic index scores, Average Score Per Taxon (ASPT), and Ecological Ratios (EQR) [Mean (SD)] across anthropogenic disturbance categories (n = 45) in the Upper Napo River Basin.

Index/Category	Mean (±SD)	ASPT (Mean ± SD)	EQR (O/E)	Ecological Status
	AAMBI
FT (Reference)	82.3 (14.5)	6.5 (1.1)	1	Excellent/Very Good
CA	30.6 (16.5) *	5.1 (1.9) *	0.37	Regular
WD	23.4 (14.3) *	3.9 (1.4) *	0.28	Bad
GM	13.2 (14.4) *	4.6 (2.8) *	0.16	Bad
	BMWP-Col
FT (Reference)	88.0 (13.5)	7.3 (1.9)	1	Moderate/Good
CA	33.5 (17.5) *	5.0 (2.8) *	0.38	Bad/Regular
WD	24.3 (15.6) *	4.0 (1.5) *	0.28	Bad
GM	12.1 (14.2) *	4.9 (3.3) *	0.14	Very Bad

Note: FT = Few Threats; CA = Crop or Aquaculture; GM = Gold Mining; WD = Wastewater Discharge. ASPT (Average Score Per Taxon) is calculated as the total biotic score divided by the number of families present at each site. The asterisk (*) indicates a significant difference compared to the reference group (FT) at p < 0.05. EQR (Ecological Quality Ratio) was calculated as the ratio of the observed (O) mean of the category to the expected (E) reference mean from FT sites. AAMBI quality ranges: >100 Excellent, 75–90 Very Good, 50–74 Good, 25–49 Regular, <24 Bad. BMWP-Col quality ranges: ≥100 Good, 61–100 Moderate, 36–60 Poor, 16–35 Bad, <15 Very Bad.

). The high ASPT values for both indices in these reference sites (ASPT_AAMBI = 6.45 ± 1.12; ASPT_BMWP = 7.25 ± 1.93) reflect a community dominated by highly sensitive taxa. These sites achieved an Ecological Quality Ratio (EQR) of 1.00, representing the regional baseline for high ecological integrity.

In contrast, GM and WD sites showed impairment. Gold mining (GM) sites were characterized by the lowest absolute richness, while wastewater discharge (WD) sites exhibited the lowest average sensitivity per taxon (ASPT_AAMBI = 3.92 ± 1.35; ASPT_BMWP = 4.03 ± 1.45), confirming the collapse of sensitive assemblages and the dominance of highly tolerant taxa (Table 2). Notably, CA sites showed a significant drop in ASPT compared to FT, although their raw scores remained in the “Regular” category. This dual-index ASPT approach enhanced the diagnostic resolution of the study by confirming that even when raw scores were moderate, the intrinsic quality of the remaining taxa was significantly degraded.

Nutrient analysis confirmed a eutrophication risk in WD and CA sites (

Table 3. Nutrient enrichment and organic matter parameters [Mean (SD)] across anthropogenic disturbance categories (Kruskal–Wallis and Dunn’s post hoc, p < 0.05) and comparison with aquatic life standards.

Parameter (Unit)	FT (n = 6)	CA (n = 22)	WD (n = 7)	GM (n = 10)	Ecological Threshold	Regulatory Limits (ECU/CAN)
Nitrates (mg/L)	1.07 (1.00)	0.50 (0.22)	0.40 (0.17)	0.73 (0.32)	>2.0 (Moderate)	No specific/No specific
Total Phosphate (mg/L)	0.32 (0.17) *	0.65 (0.56) #	1.79 (1.15) * + #	0.68	>0.1 (Eutrophication)	No specific/0.03 (CCME soft water)
Ammonium (mg/L)	0.66 (0.36)	0.33 (0.50)	1.71 (1.46) * + #	0.44 (0.22)	>0.5 (Pollution)	0.02/0.019 (CCME, cold water)
Nitrite (mg/L)	0.01 (0.01) *	0.11 (0.14) #	0.30 (0.42) * + #	0.09 (0.09)	>0.1 (Pollution)	0.06/0.06 (CCME)
BOD (mg/L)	0.47 (0.33)	5.83 (13.85) #	0.58 (0.30)	0.68	>5.0 (Hypoxia Risk)	No specific/No specific
Sulfate (mg/L)	0.77 (0.39)	1.14 (0.87)	4.13 (0.22) * +	1.16 (0.66)	--	No specific/No specific
N:P Ratio	3.34	0.77	0.22 * +	1.07	7.0–10.0 (Balanced)	7.0–10.0 (Balanced)/10:1 (healthy aquatic ecosystems)

Note: CA = Crop/Aquaculture; FT = Few Threats (reference); GM = Gold Mining; WD = Wastewater Discharge. n = number of sampling sites per category. Values are presented as Mean ± SD. Significant differences were determined using Kruskal–Wallis followed by Dunn’s post hoc test (p < 0.05): (*) vs. FT; (+) vs. GM; (#) exceeds ≥1 aquatic life limit. Sources: ECU = Unified Text of Secondary Environmental Legislation of Ecuador (TULSMA 2015) [55], Book VI, Appendix 1, Table 3 [55]: Freshwater flora and fauna preservation; CAN = Canadian Council of Ministers of the Environment (CCME, Freshwater Aquatic Life Guidelines) [56]. N:P < 7.0 indicates N-limitation (P excess, eutrophication risk).

). The N:P ratio in WD (0.22) and the high ammonium levels (1.71 ± 1.46 mg/L) represent a sewage signature that directly correlates with the dominance of tolerant macroinvertebrate families.

3.3. Alpha Diversity and Community Structure

Sample completeness, assessed via coverage-based rarefaction, reached high values (C ≈ 0.99) across all impact categories, indicating that the sampling effort was sufficient to robustly characterize and compare macroinvertebrate communities. Alpha diversity metrics followed a clear anthropogenic gradient; the highest family richness (q = 0 ≈ 24.3) and Exponential Shannon diversity (q = 1 ≈ 12.1) were recorded at FT sites, reflecting stable and heterogeneous habitats. CA sites showed intermediate richness (q = 0 ≈ 15.8), whereas GM and WD sites exhibited the lowest taxonomic diversity (q = 0 ≈ 9.9 and 10.7, respectively) (Figure 3).

Community structure analysis through rank-abundance curves (Figure 4) revealed high dominance in WD sites (q = 2 ≈ 1.5), where Chironomidae accounted for nearly the entire assemblage. In contrast, GM sites showed relatively high evenness (q = 2 ≈ 8.1) despite their low richness. This pattern likely stems from the extremely low overall abundance (33 individuals total), which prevented any single taxon from achieving absolute dominance, possibly due to the limiting effects of sediment toxicity and extreme turbidity.

3.4. Multivariate Analysis of Environmental Drivers

The PCA biplot (explaining 57% of the total variance) demonstrated a clear separation of FT sites along the positive PC1 axis, which was strongly associated with higher dissolved oxygen (DO) and family richness, as well as superior performance in biotic indices (Figure 5).

In contrast, CA, GM, and WD sites clustered toward the center and the negative PC1 axis, correlating with increased water temperature, total dissolved solids (TDS), and turbidity. The 95% confidence ellipses confirmed that FT sites constitute a distinct ecological group, whereas impacted categories exhibited significant overlap. This pattern indicates a homogenized response to cumulative anthropogenic stressors, specifically thermal stress, nutrient enrichment, and increased sediment loads.

The Non-metric Multidimensional Scaling (NMDS) ordination (Stress = 0.14) showed a clear separation of macroinvertebrate community structure across the four sites (Figure 6). FT sites clustered tightly together, showing distinct composition relative to the more dispersed and overlapping clusters of CA, GM, and WD sites along the NMDS1 axis. PERMANOVA analysis confirmed significant differences in community composition among all groups (p < 0.05). Pairwise comparisons revealed the strongest differentiation between CA and GM (F = 5.32, R² = 0.16, Padj = 0.0018) and between FT and CA (F = 2.90, R² = 0.10, Padj = 0.006). In contrast, the comparison between CA and WD showed the lowest degree of variation (F = 2.13, R² = 0.07, Padj = 0.0251), although it remained statistically significant.

The impact of gold mining (GM) was further corroborated by the significant mobilization of heavy metals (

Table 4. Heavy metal concentrations, macroinvertebrate biotic and community structure indices [Mean (SD)] in relation to anthropogenic disturbance (n = 45 sites).

Indicator	FT	CA	WD	GM	Aquatic Life Limits (TULSMA/CCME)
METALS (µg/L)
Iron-Fe	--	170 (153)	--	430 (229) * † #	300/300
Copper-Cu	--	117 (60)	--	338 (128) * †	20/2
Manganese-Mn	--	--	--	263 (161) * † #	100/430
Aluminum-Al	--	--	--	230 (77) * † #	100/5–100
Lead-Pb	--	8.3 (2.8)	--	7.5 (6.2)	10/1–7
Cadmium-Cd	--	0.6 (0.2)	--	0.3 (0.2)	1/0.09
BIOTIC INDICES
AAMBI	85.3 (20.6) *	25.5 (7.8)	21.7 (17.2) †	22.1 (14.3) †	>100 (Excellent)
BMWP-Col	90.5 (18.6) *	27.3 (10.0)	23.8 (18.2) †	23.0 (15.9) †	≥100 (Good)
EQR-A (AAMBI-based)	1.0 (0.0) *	0.8 (0.1)	0.6 (0.1) † ‡	0.8 (0.1) †	~1.0 (Optimal)
EQR-B (BMWP-based)	1.0 (0.0) *	0.7 (0.1)	0.5 (0.2) † ‡	0.6 (0.2) †	~1.0 (Optimal)
COMMUNITY STRUCTURE
Richness (families)	41	31	27	11	>40 (Ref)
Exp. Shannon Diversity exp(H’)	12.1	4.7	2.5	8.8	>10 (Ref)
Chironomidae (%)	10%	62%	80% *	30%	<15%
EPT taxa (%)	15%	55% *	9% †	5% †	>50%

Note: CA (Crop/Aquaculture); FT (Few Threats); GM (Gold Mining); WD (Wastewater Discharge). Mean ± SD. Kruskal–Wallis + Dunn’s vs. FT: (*) p < 0.05 vs. FT; (†) vs. CA; (‡) vs. GM; (#) exceeds ≥1 aquatic life limit. CCME guidelines for Cu, Mn, Pb, and Cd are hardness-dependent; values represent the most stringent criteria for soft water. Aluminum limits are pH-dependent (5 µg/L if pH < 6.5; 100 µg/L if pH ≥ 6.5). Aquatic Life Limits: Unified Text of Secondary Environmental Legislation of Ecuador (TULSMA 2015) [55], Book VI, Appendix 1, Table 3 [55] (Freshwater flora and fauna preservation); Canadian Council of Ministers of the Environment (CCME, Freshwater Aquatic Life Guidelines) [56]. AAMBI quality ranges: >100 Excellent, 75–90 Very Good, 50–74 Good, 25–49 Regular, <24 Bad. BMWP-Col quality ranges: ≥ 100 Good, 61–100 Moderate, 36–60 Poor, 16–35 Bad, <15 Very Bad. EQR based on Observed/Expected ratio (1.0 = reference condition).

). Concentrations of iron (430 ± 229 µg/L) and copper (338 ± 128 µg/L) at GM sites were two to three times higher than in other categories. These elevated concentrations coincided with the lowest biotic index scores and the near-total loss of sensitive EPT taxa.

4. Discussion

Our multimetric assessment across the anthropogenic gradient (CA, GM, WD, FT) in the Upper Napo River Basin confirms that gold mining (extreme turbidity >1320 NTU; Fe 430 µg/L; Cu 338 µg/L) and wastewater discharge (fecal coliforms >33,708 CFU/100 mL) drive the most severe ecological degradation (EQR = 0.28, Bad status), with GM showing near-total defaunation (q0 = 9.9) and WD extreme Chironomidae dominance (q2 = 1.5). ASPT (≈4.0) and EQR outperformed raw AAMBI/BMWP-Col scores by detecting functional shifts toward tolerant taxa despite moderate raw values, validating our approach to address the research question of pressure-specific ecological baselines. These findings reveal clear diagnostic thresholds for Andean–Amazonian rivers, where FT sites establish regional reference conditions (EQR = 1.0, q0 = 24.3) now threatened by cumulative anthropogenic impacts.

4.1. Benthic Community Shifts and Taxonomic Loss Across the Disturbance Gradient

The macroinvertebrate assemblages in the UNRB showed a sharp decline in diversity along the studied anthropogenic gradient. Sites categorized as FT acted as regional references of high ecological integrity, supporting the highest richness (q0 ≈ 24.3) and a balanced community structure (q1 ≈ 12.1), dominated by sensitive taxa such as Elmidae and Leptophlebiidae (Figure 3, Table 2). These findings align with the recent 2025 baseline for Western Amazonian rivers, which identifies the region as an “ecological stronghold” harboring 74% of the basin’s ichthyofauna and providing critical connectivity for sediment and nutrient transport [4].

In contrast, the transition from FT to impacted sites (CA, GM, and WD) was marked by a predictable loss of sensitive EPT (Ephemeroptera, Plecoptera, Trichoptera) families and an increase in tolerant groups, particularly Chironomidae. This pattern confirms the sensitivity of benthic communities to land-use changes in the Neotropics, where the loss of riparian forest and increased nutrient loads disrupt the base of the aquatic food web [14].

4.2. Eutrophication Signatures and the Diagnostic Resolution of ASPT and EQR Indices

Wastewater Discharge (WD) sites exhibited the most extreme ecological impairment, characterized by an overwhelming dominance of Chironomidae (80%) and the lowest Exponential Shannon diversity (q1 ≈ 2.5). The integration of microbiological data revealed critical fecal coliform levels (33,708 ± 58,047 CFU/100 mL), which are 56 times higher than the Unified Text of Secondary Environmental Legislation of Ecuador (TULSMA) limit for human consumption (600 CFU/100 mL) and over 160 times higher than reference conditions (Table 1). A significant finding of this study is the enhanced diagnostic power provided by the simultaneous application of the ASPT and the EQR.

While raw AAMBI and BMWP-Col scores for WD sites suggested a state of “Bad” quality, it is the ASPT values (3.92 and 4.03, respectively) that provide the necessary normalization to confirm that the community is composed almost exclusively of highly tolerant taxa (Table 3). This filters out the richness-based bias where traditional indices might be slightly inflated by the presence of multiple tolerant groups. Furthermore, the EQR (0.28) effectively captures the severe magnitude of the deviation from the expected reference community. This multi-metric resolution has also been documented in other urbanized Ecuadorian Amazonian streams, such as the Orienco, where macroinvertebrates proved to be better indicators of long-term organic pollution than standard chemical analysis [10]. In the present study, the application of EQR and ASPT allowed for a deeper diagnostic resolution of the data originally reported by Galarza et al. [25], confirming that the 30% to 53% decline in ecological quality is driven by a functional shift toward tolerant taxa like Chironomidae, which dominate in systems characterized by severe hypoxia and untreated sewage [25].

4.3. Synergistic Physical and Chemical Impairment Driven by Alluvial Gold Mining

Gold mining (GM) represented the most detrimental impact in the UNRB, resulting in the lowest absolute abundance, with only 33 individuals collected across 10 sites. GM sites were characterized by extreme turbidity (NTU)—exceeding the TULSMA limit of 100 NTU by 13 times—and the mobilization of heavy metals, including iron (430 µg/L) and copper (338 µg/L), at concentrations 2.5 to 3 times higher than reference sites. The actual toxicity of these metals is intrinsically linked to the water’s physicochemical matrix; such extreme sediment loads suggest that suspended particles could act as a sink for Fe and Cu through adsorption, potentially limiting their dissolved bioavailable fraction. However, the magnitude of these concentrations indicates that even a partially bioavailable fraction may exert significant physiological stress on macroinvertebrates, especially when coupled with the physical destruction of their habitats [25,26]. The extreme reduction in macroinvertebrate abundance observed at GM sites is mechanistically explained by the high sediment toxicity and heavy metal mobilization (Fe and Cu) identified in the primary assessment of these locations [26]. Our integrated analysis confirms that the near-total defaunation is not merely a localized event but a systemic response to the toxicological threshold reached in mining-impacted sediments, as evidenced by previous bioassays in these same reaches [26]. Since dissolved organic matter (DOM) also regulates metal toxicity in Amazonian systems [57], its measurement should be prioritized in future assessments to define biogeochemical breakpoints. These findings are consistent with acid mine drainage (AMD) signatures and current reports on the rapid expansion of illegal gold mining across the Ecuadorian Amazon, including the Punino and Anzu sectors of Napo province (total impacted areas >1422 ha by late 2024) [19].

The near-complete absence of macroinvertebrates in GM sites is likely a synergistic effect of physical habitat destruction and indirect ecological constraints. High concentrations of suspended solids cause the clogging of interstitial spaces and the scouring of substrates; furthermore, increased turbidity limits light penetration, drastically reducing primary productivity and the availability of food sources for scraper and collector functional groups [58]. This physical degradation is compounded by chemical toxicity from mercury and other metals [59]. In the Ecuadorian Amazon, approximately 150 tons of mercury are discharged annually into river systems, posing an unprecedented threat to biodiversity and the health of riverside communities like the Kichwaruna [59]. Although mercury concentrations were not analyzed in the present study due to budgetary and logistical constraints for trace metal laboratory testing, its presence remains a primary concern in the UNRB. Given the extent of alluvial mining activities in the region, we strongly recommend that future biomonitoring programs incorporate mercury and other heavy metal analyses to better characterize the chemical stress on aquatic communities. This persistent contamination poses a risk not only to biodiversity but also to human health, as shown by detectable metal levels in clinical analyses of local populations [20].

4.4. Landscape Transformation and the Breach of Ecological Resilience Thresholds

Crop and Aquaculture (CA) sites presented an intermediate state of degradation. Although these sites retained some families from the EPT group, such as Baetidae and Hydropsychidae, these taxa are known to be more tolerant to moderate pollution compared to other sensitive EPT members [13,39]. Interestingly, while groups like Perlidae and Leptophlebiidae are highly sensitive to organic enrichment, they showed a persistent presence in some sites impacted by mining (GM) (Figure 4). This suggests that certain sensitive taxa may exhibit higher tolerance to metal-related stress than to organic pollution, provided that some microhabitat features remain available [60]. Nevertheless, the overall reduction in taxonomic richness observed at CA sites is consistent with the chronic pesticide exposure previously documented in these areas [34]. Reassessment of these biological assemblages through Hill numbers and EQR reveals that biological impairment closely parallels the chemical risk (ms-PAF) reported by Cabrera et al. [34], providing a clear biological signal of the chronic pressure exerted by fungicides and insecticides in the agricultural frontier. Recent studies in the Napo watershed (2024) specifically comparing streams in oil palm monocultures versus traditional polycultures (Kichwa chakras) found that monocultures consistently support lower macroinvertebrate richness due to the loss of buffer strips and pesticide runoff [21]. As agricultural expansion remains the dominant driver of deforestation in Ecuador, the Western Amazon is approaching an irreversible ecological threshold; models suggest that if current rates persist, up to 47% of the forest area could be degraded by 2050, permanently altering the region’s hydrological stability [5]. Furthermore, the resulting hypoxic conditions, with dissolved oxygen levels falling below the 80% regulatory threshold in impacted reaches, further limit the survival of stenothermic Andean–Amazonian taxa [22].

4.5. Policy Frameworks and Management Priorities for Basin-Wide Resilience

The consistent results across all lines of evidence underscore the urgent need for a shift in water management in the UNRB. First, the construction of wastewater treatment plants (WWTPs) in urban areas is an immediate priority to reduce the critical coliform and ammonium loads that currently collapse aquatic communities. Second, the declaration of Indigenous-led Fluvial Reserves, such as the one established in 2023 for the Nushiño-Curaray-Villano area, represents a viable strategy to maintain longitudinal connectivity and protect biodiversity refuges from mining and oil expansion [61], complemented by broader National System of Protected Areas (SNAP) protections.

However, it is equally imperative that units within the SNAP officially implement biological corridors and buffer zones, which are currently lacking in many Amazonian and Ecuadorian reserves. The absence of these transition zones exposes protected areas to the ‘hard edges’ of land-use change, evidenced by deforestation rates reaching up to 25.5% in 5-km buffer zones [62]. Adopting longitudinal fluvial connectivity models, such as the proposed Jondachi–Hollín–Misahuallí–Napo corridor by the Napo River Foundation and Ecuadorian Rivers Foundation (private foundations led by Napo residents), alongside terrestrial corridors like Sangay–Podocarpus and Llanganates–Sangay, is essential to link isolated SNAP patches and maintain biodiversity flow under increasing anthropogenic pressure [63].

Future monitoring programs should prioritize filling critical data gaps identified in this study by establishing long-term strategic monitoring at key sentinel sites across anthropogenic gradients. While family-level identification of macroinvertebrates, as used in this study, is a robust and cost-effective tool for regional assessments in the Neotropics [64], we recognize its limitations in detecting species-specific responses to environmental stress. In line with international standards, such as the European Water Framework Directive (WFD), species-level identification is often required to provide higher diagnostic resolution for biological quality elements [65]. Therefore, future research in Andean–Amazonian basins should strive to enhance taxonomic resolution through the development of regional keys and the integration of molecular tools (e.g., DNA metabarcoding). Multimetric biomonitoring frameworks combining taxonomic diversity (Hill numbers), Ecological Quality Ratios (EQR), and chemical stressors will provide robust baselines for adaptive management of these sensitive ecosystems [10,21].

5. Conclusions

This study establishes multimetric ecological baselines for Andean–Amazonian rivers, revealing critical degradation thresholds in the Upper Napo River Basin. Gold mining emerges as the primary driver of ecosystem collapse (EQR = 0.16, q₀ = 9.9), followed by wastewater discharge (EQR = 0.28) and agricultural impacts (EQR = 0.37). Normalized indices (ASPT, EQR) provided superior diagnostic resolution over raw biotic scores, identifying functional collapse where physicochemical parameters appeared deceptively moderate.

These findings highlight three critical management priorities: (1) stringent regulation of legal and illegal alluvial gold mining, (2) construction of wastewater treatment infrastructure, and (3) establishment of Indigenous-led, private, and state-supported fluvial reserves featuring longitudinal connectivity corridors. Absent immediate intervention, ongoing deforestation and mining expansion risk permanently undermining this vital freshwater stronghold, jeopardizing essential ecosystem services for 2.5 million people dependent on Napo River waters. The ecological thresholds established in this study offer indispensable benchmarks for adaptive management and policy transformation across the Western Amazon.