ABSTRACT The Rural Environmental Registry (CAR, from its Portuguese acronym) is a mandatory public online registry that consolidates information about a property’s area, land use, native vegetation, legal reserve, and vegetation conservation areas. The objective of CAR is to monitor land use and combat deforestation. Currently, CAR registry analyses are conducted manually, requiring specialized personnel and a time-consuming verification process - a challenge amplified by Brazil’s almost 8 million CAR registries and over 8 million rural properties. To address this problem, the use of machine learning techniques for automatic verification is crucial for Brazil’s environmental management. This study aims to develop machine learning-based strategies to classify CAR registries into two categories: ‘Approved’, referring to correct registries, and ‘Canceled’, denoting registries with inconsistencies. Additionally, interpretable machine learning tools were used to determine the impact of each feature on the automatic classification. The best-performing model was the Gradient Boosting Machine, achieving classification accuracy above 90% (using a real test dataset containing more than 6,000 registries) and demonstrating correlations between its feature importance scores and the inconsistencies identified in the literature. These results demonstrate the feasibility of using this approach for registry classification and interpretation.
ABSTRACT Converting depleted mining pits into water reservoirs can reduce water scarcity; however, geochemical dynamics may cause acidification and metal solubilization, posing ecotoxicological risks. This study evaluated the effects of these waters on the germination and early development of sunflower (Helianthus annuus L.) and lettuce (Lactuca sativa L.). A completely randomized design was used in a factorial scheme (two cultivars × five water sources) for sunflower and a unifactorial design for lettuce. The water sources were distilled water (control), three inactive ponds, and one active pond. Germination, Germination Speed Index (GSI), morphometry, biomass, and abnormal seedlings were evaluated. Water from the mining pits did not impair seed viability. Water from one of the inactive ponds increased fresh biomass, whereas the control resulted in a higher number of abnormal lettuce seedlings; for sunflower, the inactive ponds allowed for greater hypocotyl length, while the control resulted in a higher GSI. Among the sunflower cultivars, BRS 422 exhibited the greatest length and fresh biomass, whereas BRS 323 showed the highest dry biomass and GSI. Waters influenced by sand mining had statistically similar or superior effects compared to the control treatment, demonstrating their potential to support well-developed seedlings without phytotoxicity symptoms, thereby fostering opportunities for agricultural sustainability and integrated water-resource management.
ABSTRACT Jacaratia spinosa (Aubl.) A. DC. is a native dioecious tree from the Atlantic Forest with important cultural and biological relevance, although its fruits remain underutilized. This study aimed to characterize the morphological, physicochemical, and nutritional traits of fruits from female genotypes collected in different locations of Misiones, Argentina, and to perform a preliminary assessment of peel-derived pigment extracts. Fifteen female genotypes were georeferenced and evaluated. Significant variability was observed for fruit weight (28.28-81.94 g), fruit length (5.57-7.65 cm), and pulp yield (40.29-56.90%). The fruits also showed high soluble solids content (mean 15.91 °Brix) and relevant concentrations of potassium (777.52 mg 100 g-¹), calcium (87.41 mg 100 g-¹), and magnesium (98.88 mg 100 g-¹). Multivariate analysis did not reveal clear clustering according to geographic origin, suggesting that fruit quality traits were more strongly associated with phenotypic variation among sampled plants than with sampling location. For peel extraction, the highest yield (53.29%) was obtained using 60% ethanol for 90 min. FT-IR analysis indicated spectral differences among extracts and spectral regions commonly associated with plant pigment-related compounds. These results highlight the potential of J. spinosa for selection of superior horticultural materials, horticultural exploitation, and the valorization of fruit processing by-products.
ABSTRACT The composition of the culture medium is a critical factor in the development of plants in vitro, influencing the growth and morphophysiological characteristics of micropropagated plants. This study evaluated the suitability of MS (Murashige & Skoog) and PC-L2 (Phillips & Collins) culture media for the in vitro development of Selenicereus undatus (Haw.) D.R. Hunt based on growth characteristics, photosynthetic responses, mineral composition, and anatomy. Cladodes were cultured for 75 days under a 16-h photoperiod and a photosynthetic photon flux density of 90 μmol m-² s-¹. Shoot and root growth, fresh and dry biomass, photosynthetic pigments, chlorophyll a fluorescence parameters, elemental composition, and anatomical characteristics were evaluated. Plants grown in PC-L2 medium exhibited greater shoot elongation, fresh biomass accumulation, root area, and root diameter, whereas MS medium promoted a greater cladode cross-sectional area and a higher number of vascular bundles. Chlorophyll b concentration was higher in PC-L2 medium, whereas chlorophyll a fluorescence parameters did not differ between the media, indicating the maintenance of photosynthetic performance. Mineral analysis revealed greater Mg and P detection in PC-L2 medium, whereas K predominated in both treatments. Overall, both media promoted the development of S. undatus; however, PC-L2 showed greater potential for promoting in vitro growth under the conditions evaluated.
ABSTRACT Field-scale compaction assessments have mainly focused on harvesting or isolated machine attributes. This study assessed machinery-traffic compaction risk for selected mechanized operations within a monitored multicrop period in a commercial grain field on a clayey Oxisol in southeastern Brazil. A mechanism-based operational exposure assessment combined machine-load characterization, global navigation satellite system (GNSS) traffic mapping for soybean harvesting, geometric traffic-coverage estimates for planting and spraying, measured precompression stress, post-harvest bulk-density measurements, and traffic recurrence. Two combine grain harvesters, one tractor-planter set, and one self-propelled sprayer were evaluated. Simulated contact stresses were compared with measured precompression stress to identify threshold exceedance. The planter transport wheels produced the highest localized stress pathway, reaching 2.14 times the reference precompression stress. Under the spraying recurrence scenario, the sprayer rear axle produced the highest cumulative exposure index (0.530), threshold exceedance accumulated over 18 documented spraying events. GNSS-based harvest mapping showed that 25.5% of the monitored soybean harvest area received wheel contact, and 14.9% and 11.8% of the working areas of the two harvesters were subjected to wheel loads above 5 Mg per wheel. Post-harvest bulk-density contrasts were small and inconsistent, indicating that bulk density alone was not sufficiently sensitive for operational risk ranking. Compaction-risk prioritization changed according to the mechanism considered: localized stress during planting, recurrent threshold exceedance during spraying, and spatially extensive loading during harvesting.
