This study investigated the presence of antibiotic-resistance genes in drinking water consumed by the university community in the Peruvian Amazon. Water samples were collected from three primary sources: inflow from the distribution network, a storage cistern, and an underground intake. Conventional PCR was employed to detect genes associated with resistance to erythromycin (ermC), ampicillin (amp), ciprofloxacin (QEP), multidrug resistance (marA), and specific multidrug resistance in E. coli (qEmarA). Physicochemical analysis revealed compliance with most regulatory standards; however, groundwater samples showed lead concentrations exceeding legal limits (0.72 mg/L) and lacked residual chlorine. All sampling points tested positive for the evaluated resistance genes, demonstrating the widespread dissemination of resistance factors in drinking water. Contrary to initial expectations, resistance genes were also prevalent in treated sources. These findings reveal a critical public health risk for the university community, emphasising the need for effective disinfection systems and robust monitoring protocols to ensure water safety. The presence of these resistance genes in water is a critical public health concern as it can facilitate the spread of resistant bacteria, reducing the effectiveness of medical treatments and increasing the risk of infections that are difficult to control.

Antimicrobial resistance is a growing threat to public health and the environment, especially in vulnerable ecosystems such as the Amazon. The confluence of the Marañón, Utcubamba, and Chinchipe rivers, known as the Pongo de Rentema, is a strategic area where water pollution could facilitate the spread of antibiotic resistance genes. This study aims to assess water quality in this region under the “One Health” approach by analyzing physicochemical parameters, heavy metals, and the presence of antimicrobial resistance genes. Water samples were collected from five sampling points during September and October 2024. Physicochemical parameters were analyzed in situ, and heavy metal concentrations were determined using atomic emission spectrophotometry. The presence of Escherichia coli and Pseudomonas aeruginosa was evaluated through selective culture, and the detection of resistance genes (marA, ermC, amp, QEP, and qEmarA) was performed using conventional PCR. Physicochemical parameters were within the limits established by Peruvian regulations, except for total dissolved solids in the Utcubamba River. Elevated levels of lead and chromium were detected at some points. Additionally, resistance genes were identified in E. coli and P. aeruginosa, providing evidence of antimicrobial resistance dissemination in the water. Water pollution in the Pongo de Rentema poses an environmental and public health risk due to the presence of heavy metals and antimicrobial resistance genes. Continuous monitoring and environmental management strategies under the “One Health” approach are recommended to mitigate these risks.

Accurate identification of timber species is essential for sustainable forest management and the prevention of illegal logging, particularly in regions such as the Peruvian Amazon. This study evaluated explainable deep learning models for the classification of cedar (Cedrela montana) and cinchona (Cinchona pubescens) using macro-images of transverse wood sections collected in the Amazonas region of northern Peru. A dataset of wood images was expanded through a sliding-window patch extraction strategy and used to train and compare three architectures: a custom Convolutional Neural Network (CNN), MobileNetV2, and EfficientNetB0. Model performance was assessed using accuracy, precision, recall, and F1-score, while Gradient-weighted Class Activation Mapping (Grad-CAM) was applied to interpret the visual regions influencing predictions. All models achieved high classification performance. EfficientNetB0 obtained the best results, reaching 100% accuracy, precision, recall, and F1-score, while MobileNetV2 and the custom CNN achieved near-perfect performance. Training and validation curves indicated stable convergence without significant overfitting. Grad-CAM analysis showed that the custom CNN generated more localized and interpretable activation regions, whereas MobileNetV2 and EfficientNetB0 focused on broader textural patterns, revealing different feature-learning strategies. These findings demonstrate the potential of explainable deep learning for automated wood identification and support applications in forest monitoring, biodiversity conservation, and illegal logging control.