Environmental and Earth Sciences

Abstract: Agriculture faces compounding pressures from food insecurity, climate change and resource scarcity, creating urgent demand for scalable analytical tools. This PRISMA 2020-compliant systematic review synthesises 582 peer-reviewed studies on machine learning (ML) and deep learning (DL) in agriculture, retrieved from Scopus for January 2019 to March 2026. Full texts were retrieved for 430 studies (73.9%), which form the analytic subsample for every full-text-derived variable; all such statistics are reported against that denominator. Every study characteristic was independently extracted using a transparent, section-aware rule-based pipeline. Publication volume grew from 6 papers in 2019 to 251 in 2025 (compound annual growth ≈ 86%); the 61 papers from January–March 2026 are a partial year and are excluded from growth comparisons. Convolutional backbones remain dominant (57.2% of full texts), but Transformer-based models rose from 14.3% of full texts in 2022 to 41.2% in 2025. Crop health and protection accounts for 48.3% of the corpus and production forecasting for 30.9%. South Asia and East Asia together contribute 59.3% of output, whereas Sub-Saharan Africa contributes 9 papers (1.5%) and Latin America and the Caribbean 8 (1.4%), against a far larger share of the global hunger burden. Classified at country level, 78.4% of the corpus originates from Global South institutions, but India and China alone supply 67.3% of that output. Two new instruments quantify the translation gap. An evidence-maturity classification shows that 33.0% of studies never leave curated data, 36.5% reach field validation and 30.5% reach an operational prototype; median reported accuracy is 99.0% for studies evaluated on public benchmarks alone but 95.0% for studies evaluated on their own field data (Mann–Whitney p < 0.001). A 0–10 reproducibility index has a median of 3; code is openly available in 7.2% of studies, data in 24.7%, and both in 4.9%. Open data disclosure improves significantly over time (Spearman ρ = 0.131, p = 0.007) whereas code sharing does not. Headline accuracy figures should therefore be read as upper bounds obtained under controlled evaluation. Benchmark standardisation, field-realistic validation, smallholder-relevant design and geographic equity remain the field's most pressing unresolved challenges.

Abstract: With the rapid growth of remote sensing data and multimodal information, foundation models for Earth observation and vision–language geospatial reasoning models have become key drivers of intelligent remote sensing analysis. However, the specific roles and applications of reasoning models in remote sensing remain insufficiently explored. This survey defines RS-Reasoning operationally as task-dependent, multi-step inference whose conclusions are supported by traceable visual, temporal, spatial, or tool-derived evidence and are evaluated beyond final-answer accuracy. We organize the literature into three non-exclusive paradigms according to the dominant mechanism by which reasoning is acquired and executed: supervised reasoning, reinforcement learning–driven reasoning, and agentic or tool-augmented reasoning. We further distinguish reasoning-specific methods from vision–language models that provide semantic alignment, generation, or grounding foundations but do not themselves demonstrate multi-step inference. Across urban analysis, disaster assessment, environmental monitoring, and spatiotemporal question answering, we identify the conditions under which reasoning may add value while separating benchmark capability from validated operational deployment. Our synthesis exposes four persistent limitations: frequent use of synthetic supervision with unevenly reported provenance, outcome-dominant evaluation, weak verification of cross-modal evidence, and fragile long-horizon tool use. We therefore outline a research agenda centered on geography-aware process supervision, multimodal constraint checking, calibrated uncertainty, expert feedback, and robustness tests across sensors, regions, and time. For convenient reference and further research, we maintain a curated collection of related resources at https://github.com/ML4Sustain/Awesome-RS-Reasoning-Models.

Abstract: Whether the European Union Emissions Trading System has influenced the composition of electricity generation remains contested, and the evidence rests largely on aggregate European time series in which carbon prices and clean energy indicators rise together. We show that this design cannot separate a price effect from a common trend: in EU-27 data for 2005–2024, a linear time trend alone accounts for 98 per cent of the variation in renewable electricity, and the allowance price loses all explanatory power once a trend is included. We therefore construct a panel of the 27 Member States over 2005–2024 and identify the effect of carbon pricing from differential exposure, interacting the common allowance price with national fossil generation shares measured over 2000–2004, before the system existed. Three results follow. Carbon cost pressure displaces coal: a one-euro increase in the allowance price reduces the coal share of generation by 0.138 percentage points per unit of pre-ETS coal exposure, and this estimate is robust to country-specific linear trends. The effect is fuel-specific, operating through coal exposure but not gas exposure, as the difference in carbon content implies. Carbon pricing has no detectable effect on renewable deployment; the estimate is a precise null that excludes effects of more than roughly ±0.15 percentage points. Finally, the coal effect is conditional on the policy framework: it is −0.200 in Member States that had adopted a national phase-out commitment by 2019 and statistically indistinguishable from zero elsewhere, with the difference significant at conventional levels. Carbon pricing and phase-out commitments appear to function as complements rather than substitutes, with direct implications for the design of price floors and the extension of emissions trading to jurisdictions without national exit trajectories.

Abstract: Urban cycling is promoted as a sustainable transport mode and a means of increasing daily physical activity, but increased inhalation of fine particulate matter (PM_2.5) may offset its health benefits. This secondary, scenario-based analysis assessed the balance between physical-activity benefits and PM_2.5-related risks under real-world conditions in Gliwice, Poland. Personal PM2.5 exposure was measured during 162 trips by two com-muters in heating and non-heating seasons using low-cost sensors. Each trip duration was treated as a hypothetical habitual pattern repeated five days per week. Inhaled doses were estimated using measured concentrations, trip duration, and cycling-specific ven-tilation rates. Median cycling duration was 33.0 min/day. Inhaled dose depended on personal exposure, cycling duration, and pulmonary ventilation rather than PM_2.5 con-centration alone. The median physical-activity relative risk was 0.836, corresponding to a 16.4% reduction in all-cause mortality risk, whereas the median PM2.5-related relative risk was 1.00182, equivalent to a 0.182% increase. The combined median net relative risk was 0.840 (IQR: 0.815–0.879), indicating a 16.0% net risk reduction. All trips showed a net benefit under central assumptions, and benefits remained for 160 of 162 trips in sensitivity analysis. These findings support urban cycling promotion alongside measures to reduce cyclists’ exposure.

Abstract: This systematic review, conducted following PRISMA 2020 guidelines, examines satellite remote sensing approaches used to monitor vegetation responses to climate change between 2000 and 2025. Of 757 peer-reviewed studies identified, 455 (60.1%) were available as open-access full text and were extracted using a deterministic, fully auditable regular-expression pipeline; the remaining 302 could be characterised only bibliographically and are excluded from the methodological denominators reported here. Output grew steeply after 2019, with 2021-2025 accounting for 480 studies (63.4%). Landsat (51.4% of full-text studies), the Moderate Resolution Imaging Spectroradiometer (44.6%) and Sentinel-2 (43.5%) dominated; 29.9% used synthetic aperture radar and 62.9% combined two or more sensor families. The Normalised Difference Vegetation Index remained the reference index (76.9%), ahead of the Enhanced Vegetation Index (28.1%), while solar-induced fluorescence appeared in only 4.0%. Random Forest (35.8%) and linear regression (29.5%) were the most common analytical methods; deep learning rose from absent before 2020 to 18.0% of studies published in 2023-2025. First-author institutions were split almost evenly between the Global North (52.4%) and Global South (47.6%), but the latter is dominated by one country: China contributed 208 studies (27.5%) against 152 (20.1%) for all other Global South countries combined. Drylands, shrublands and peatlands were markedly under-represented relative to their global extent. Validation reporting was incomplete, with the coefficient of determination given in 40.9% of full-text studies (within-study median 0.71). Only 8.8% provided a public code repository. The review identifies research gaps and priorities for multi-sensor integration, methodological standardisation, and reproducible practice.

Abstract: The north-eastern hilly region of Bangladesh is characterized by subtropical evergreen hilly forests, haors, tea estates, and nationally designated protected areas, and is undergoing rapid land-use transformation. This study analyzes spatiotemporal changes in land use and land cover (LULC) in Moulvibazar and Habiganj districts of Bangladesh over three decades (1994-2024) using Landsat imagery and supervised Maximum Likelihood Classification in QGIS and predicts future LULC for 2054 using the Cellular Automata-Artificial Neural Network (CA-ANN) model in the MOLUSCE (Modules for Land Use Change Evaluation) plugin. Five major land cover classes were examined: croplands, water bodies, forests, tea gardens, and settlements. Croplands declined substantially from 379,716.66 ha (72.07%) in 1994 to 303,873.97 ha (57.68%) in 2024, a net loss of 14.40%. Tea gardens nearly doubled, expanding from 28,523.84 ha (5.41%) in 1994 to 69,600.19 ha (13.21%) in 2024. Forests initially declined by 1.33% between 1994 and 2004; however, then subsequently recovered, with a net gain of 4.46% by 2024. Settlement areas expanded from 17,103.54 ha (3.25%) to 26,795.58 ha (5.09%), indicating ongoing urbanization. NDVI indicates improved vegetation health, while the NDWI indicates surface water fluctuation. The projections indicate continued tea garden and settlement expansion alongside further cropland and forest cover decline, emphasizing the need for sustainable cropland management and long-term forest management to ensure ecological sustainability in this region.

Abstract: Silica aerogels are among the most extraordinary porous materials produced through sol–gel chemistry, distinguished by ultralow density, exceptionally high porosity, large specific surface area, and extremely low thermal conductivity. Despite these characteristics, widespread application of conventional silica aerogels has been constrained by inherent brittleness, poor mechanical strength, and moisture sensitivity. Significant research has therefore focused on silica aerogel composites, in which reinforcing or functional phases — fibers, polymers, carbon nanomaterials, metal oxides, and biopolymers — are integrated into the silica network to enhance mechanical robustness, flexibility, hydrothermal stability, electrical conductivity, catalytic activity, and multifunctionality while largely preserving the parent aerogel's desirable properties. This chapter reviews the synthesis, characterization, properties, and applications of silica aerogel composites. Sol–gel processing and drying technologies are discussed, followed by composite-formation strategies and the advanced techniques used to evaluate structural, mechanical, thermal, surface, and functional properties. The effects of reinforcing phases on mechanical performance, thermal conductivity, and hydrothermal stability are analyzed, and current and emerging applications in thermal insulation, environmental remediation, catalysis, acoustic damping, aerospace systems, biomedical engineering, and energy storage are highlighted. Finally, key challenges and future directions involving multifunctional materials, green synthesis, and data-driven materials design are discussed.

Abstract:

Background: Urban Green Spaces (UGS) are vital for urban sustainability, providing ecological, environmental, and social benefits. Growing urbanisation, climate change, and biodiversity loss highlight the need for better integration of ecological principles into planning. However, existing approaches are fragmented and poorly capture hierarchical relationships between urban form, ecological processes, and ecosystem outcomes. Methods: This review synthesises interdisciplinary research from urban ecology, landscape ecology, urban forestry, ecosystem services, and spatial planning. It introduces the Integrated Ecological Framework of Urban Green Spaces (IEF–UGS), structured into six levels: urban morphology, green infrastructure, ecological processes, ecological functions, ecosystem services, and urban resilience. Results: Urban morphology is identified as the key structural driver shaping ecological functioning via habitat configuration, connectivity, hydrology, and microclimate. Green infrastructure forms the spatial basis for ecological processes, which generate functions supporting ecosystem services and resilience. The framework clarifies causal links across levels and resolves conceptual inconsistencies in prior research. Conclusions: IEF–UGS offers both theoretical and practical value, supporting ecological assessment, green infrastructure planning, urban forest management, and climate adaptation. It enables integrated indicators and improves evidence-based decision-making, providing a robust foundation for future research on sustainable urban systems.

Abstract: Private forest land-use designations can contribute to climate mitigation; however, their carbon benefits are difficult to estimate because protected and less strictly regulated forest lands often differ in site and environmental conditions. Evidence on whether designation-associated carbon differences vary with forest structure is also limited. This study evaluated carbon stock differentials between protected areas in private forests (PA) and matched semi-conservation mountainous districts (SCA) in South Korea using the InVEST Carbon Storage and Sequestration model. Propensity score matching was used to improve comparability, followed by average treatment effect on the treated (ATT) estimation and regression analysis of the matched sample. Alternative matching and regression specifications were examined as robustness checks. The mean carbon stock was 4.97 Mg C ha⁻¹ higher in PAs than in comparable SCAs (95% CI: 4.45–5.49, p < 0.001). The PA-SCA difference varied by forest structure: it was greater in broad-leaved forests than in coniferous forests, showed a nonlinear pattern across canopy density classes, and declined with increasing age class. These findings indicate that carbon stock differences associated with private forest land-use designation are structurally heterogeneous. While they should be interpreted as cross-sectional associations, they provide evidence for differentiated land-management, compensation, and conservation prioritization strategies.

Abstract: This study evaluates the optimal technological pathway for the Brussels Intercommunal Transport Company (STIB) to progressively decarbonize its urban surface transit fleet. A comprehensive, region-specific evaluation framework is developed, comparing Battery Electric Buses (BEBs) and Fuel Cell Electric Buses (FCEBs) across three key pillars: environmental performance, economic feasibility, and operational practicality. The methodology integrates a cradle-to-grave Life Cycle Assessment (LCA), a discounted Net Present Value Total Cost of Ownership (TCO) model, semi-structured interviews with transit engineering experts, and a final Multi-Criteria Decision-Making (MCDM) matrix. The LCA results demonstrate that environmental benefits are highly contingent upon upstream energy pathways; under an optimized low-carbon Belgian electricity mix and renewable electrolysis, FCEBs achieve lower lifetime emissions than BEBs (0.21 versus 0.30kgCO₂e/km), though this advantage is entirely neutralized if fossil-derived grey hydrogen is used. Conversely, the TCO analysis outlines a severe economic penalty for hydrogen fleets; over a 15-year design lifespan and 675,000 km of operation, BEBs achieve a standardized unit cost of 1.19bad hbox compared to 1.84bad hbox for FCEBs, a gap driven by current green hydrogen market price premiums relative to grid electricity. Empirical interview results confirm that BEBs exhibit superior localized practicality due to established infrastructure compatibility, mature depot electrification roadmaps, and high power supply reliability, which offset the FCEB’s theoretical advantages in range and refueling logistics. Finally, the multi-criteria analysis confirms that BEBs represent the optimal near-to-mid-term transition pathway for STIB, securing a global suitability score of 0.94 compared to 0.76 for FCEBs.

Abstract: White oak (Quercus alba L.) is among the most ecologically and economically important tree species of the eastern United States, yet its populations are increasingly threatened by climate change. This study predicts the white oak mortality (WOM) rate under alternative future climate projections across the eastern United States. WOM rate was derived from declining basal area in multicycle U.S. Forest Inventory and Analysis data (1998–2019) and related to seasonal precipitation and temperature. Future climate was represented by five statistically downscaled CMIP6 general circulation models (EC-Earth3, GFDL-ESM4, GISS-E2-1-G, MIROC-ES2L, and MPI-ESM-1-2-HR) under the SSP2-4.5 (middle-of-the-road) scenario for three 25-year intervals: early (2025–2049), mid (2050–2074), and late (2075–2099). Predicted WOM rates were mapped and compared with current conditions to identify areas of increase, decrease, and no change. Across all models and intervals, WOM rate is projected to increase most strongly in the southern part of the study area (notably Alabama and Arkansas) and in parts of the central region, while decreases are more prevalent in the north. Differences among models underscore the uncertainty inherent in climate projections. These results identify where white oak mortality risk is likely to concentrate under a changing climate and can inform adaptive silviculture, habitat restoration, and climate-informed forest conservation.

Abstract: Snowmaking systems are increasingly used to serve winter tourism in mountain regions, yet their role in the environmental dissemination of antimicrobial resistance (AMR) remains poorly understood. Here, we investigated how snowmaking infrastructure affects the transport, retention and reduction of antibiotic residues, antibiotic resistance genes (ARGs) and phenotypically resistant bacteria along two interconnected water chains feeding technical snow production in mountain catchments. Over two winter seasons (2023/24 and 2024/25), we quantified a panel of clinically relevant antibiotics, ARGs and resistance phenotypes in wastewater effluent, intake water, reservoirs, technical snow, aged snow, meltwater, receiving streams and reservoir sediments. Wastewater and intake water exhibited the highest antibiotic loads and ARG richness, dominated by fluoroquinolones, macrolides and β-lactam resistance determinants. Reservoirs accumulated substantial amount of antibiotics but their reduction occurred further along the snowmaking chain. Fresh technical snow still contained multiple antibiotics and ARGs, and harbored resistant Enterobacteriaceae and Aeromonas, indicating that snowmaking systems can redistribute AMR determinants into the snowpack. Aged snow, meltwater and receiving streams showed nearly complete loss of antibiotic residues and reduced ARG richness, due to processes of snow transformation and downstream transport. By integrating chemical, molecular and phenotypic data with the technical layout of the infrastructure, we identify specific components as AMR retention hotspots and clarify where barrier functions are most effective. Our findings demonstrate that snowmaking infrastructure can both concentrate and distribute antibiotic residues and resistance determinants in mountain catchments, highlighting the need to consider snowmaking systems in environmental AMR risk assessments and One Health frameworks.

Abstract: This research investigates the determinants of perceived outdoor thermal comfort (POTC) in a coastal area of Athens, Greece, with emphasis on seasonal transitions between cooler and warmer periods, utilizing a comprehensive survey dataset completed in late 2023. To capture the dynamic nature of human-environment interactions across varying thermal conditions, the dataset was segmented into a combined spring–autumn transitional period, winter, and summer. This seasonal stratification was validated by integrating survey data with meteorological conditions to calculate the Predicted Mean Vote (PMV) for available temporal cases. Canonical Correlation Analysis (CCA) was executed for the spring-autumn and winter periods to map the complex interrelationships between a set of demographic, experiential, environmental, and spatial predictors with four subjective dependent variables. These variables acted as surrogates for POTC, capturing the combined influence of meteorological parameters alongside perceptual and adaptive factors. The spring-autumn CCA model revealed that POTC is defined by a balance, where comfort-enhancing psychological and behavioral adaptations — relaxation space, proactive clothing adjustments, and an urban identity — successfully counteract the physiological and environmental indicators of localized heat stress. The winter CCA model demonstrated that POTC is driven by deliberate cognitive evaluations rather than generalized physical sensations, with satisfaction maximized during sunny, calm daytime conditions particularly among male respondents, and primary environmental stress determined by the convective cooling effects of wind rather than humidity. Due to sample size limitations specific to the summer cohort, analysis was restricted to descriptive methods and it was found that a strong urban identity, evening exposure, physical activity, and optimal airflow functioned as the primary positive indicators of outdoor thermal comfort, whereas peak afternoon exposure, stagnant air conditions, and an explicit demand for shading and cooling infrastructure served as the primary markers of thermal discomfort. Findings demonstrate that POTC is shaped by a dynamic interplay of adaptive, cognitive, and microclimatic factors, with the comparison of these findings with the calculated PMV values confirming that the four subjective dependent variables serve as robust, reliable surrogates for POTC. Behavioral regulators — such as urban identity, psychological relaxation, and clothing adjustments — actively mitigate physiological strain across transitional and summer periods, while a highly resilient, educated, urban-oriented demographic consistently prioritizes structural spatial benefits over seasonal microclimatic, acoustic, or convective stressors. This research suggests that climate-responsive urban design must combine targeted physical modifications — such as wind shelters in cooler months and cool pavements or strategic shading in summer — with a human-centric framework that accounts for diverse, subjective urban thermal experiences.

Abstract: Rehabilitation of mine features associated with abandoned mines is a global challenge in many countries, including South Africa, which has had a mature mining industry since the 1860s. Most existing rehabilitation approaches do not adequately address the physical, chemical, and environmental hazards associated with abandoned mines. In this study, the abandoned mine features at the Louis Moore and Klein Letaba abandoned gold mine sites were identified, assessed, scored, and ranked based on their existing and potential negative impacts on the environment and human health. Soil around the studied abandoned mine features was collected at a depth of 10-30 cm to assess contamination levels. The tailings at the abandoned mines show the decreasing order of A s˃ Co ˃ Ni ˃ Cu ˃ Zn ˃ Pb ˃ Cd. The maximum concentrations of Cu (81mg/kg), Zn (465 mg/kg), As (694 mg/kg), Ni (1213 mg/kg), and Pb (109 mg/kg) exceed the maximum permissible limits in natural soils, indicating that the tailings material was contaminated by these metals. The total physical hazard scores computed for abandoned mine features at Louis Moore and Klein Letaba are 40.96 and 16.04, respectively. Mine features for rehabilitation are prioritized as mine shafts, mine tailings, rock dumps, water reservoirs, silos, ball mills, and old buildings. The highest potential environmental hazard and risk, with a score of 9, is assigned because flooding has high potential to pollute water systems. The flooding of abandoned mine shafts, ore spillages, and toxic metals of abandoned mine dumps contaminate nearby water systems and soil.

Abstract: Given the inherent rarity of anomalous readings in environmental sensor data, tree-based models often provide high weighted classification metrics but low anomaly-class recall values, leading to excessive missed anomalies. This study proposes that integrating Bayesian surprise with a random forest classifier can simultaneously enhance weighted classification metrics and class-specific classification accuracy when classifying environmental sensor data. To test this approach, real air-, reservoir-, and river-water quality datasets are used to evaluate the new ensemble method. Bayesian surprise, which is defined using relative predictive surprise, is standardized and serves as an additional anomaly-informed feature. A random forest classifier is employed to classify raw sensor data augmented with the resulting relative predictive surprise values. Empirical experiments indicate that relative predictive surprise is robustly more effective than Rousseeuw–Croux estimators, Tukey interquartile ranges, and percentile-based rules in identifying anomalous sensor measurements. Moreover, augmenting the random forest classifier with relative predictive surprise values improves anomaly-class recall values and weighted classification metrics, indicating fewer missed anomalous sensor measurements. In conclusion, the relative predictive surprise is a computationally simpler and more effective anomaly measure compared to four baseline metrics. This anomaly measure and a random forest classifier emerge as a novel tool for balancing weighted classification metrics and class-specific classification accuracy in classifying environmental sensor data.

Abstract: Terrestrial LiDAR provides detailed three-dimensional (3-D) observations of forest structure, yet wood-foliar semantics remains challenging because of structural heterogeneity, occlusion, and variability among forest ecosystems and terrestrial LiDAR platforms. Existing deep learning (DL) approaches are commonly developed for localized forest conditions or complex multi-class semantic taxonomies that often exhibit limited transferability across structurally diverse forests. We present a terrestrial LiDAR platform-agnostic DL framework for wood–foliar semantics using globally harmonized benchmark datasets acquired from multiple terrestrial LiDAR systems representing broadleaf, coniferous, mixed, regenerating, and structurally complex forests. Three point-based architectures (PointNet++, PointNeXt, and PT) were benchmarked using identical training and evaluation protocols. PT consistently achieved the highest performance, with overall accuracy (OA) of 92–94%, balanced accuracy (BA) of 88–94%, mean Intersection-over-Union (mIoU) of 0.78–0.87, macro F1-score of 0.87–0.93, and Matthew’s correlation coefficient (MCC) of 0.74–0.86 across four independent benchmark datasets. Species-level, vertical-profile, and qualitative cross-ecosystem evaluations further demonstrated stable preservation of wood-foliar semantics throughout the 3-D fuel continuum, while revealing that many apparent disagreements originated from incomplete manual annotation of fine branches in complex forest ecosystems. The resulting wood–foliar predictions were subsequently decomposed into ecologically relevant fuel classes using a geometric framework, achieving F1-scores of 0.98 for stems and 0.92 for foliage, thereby demonstrating that binary semantic abstraction preserves sufficient structural information for hierarchical fuel characterization.

Abstract: Suspended particulate matter (SPM) plays a fundamental role in water bodies by controlling the transport of sediments, nutrients, trace elements, radionuclides, and organic contaminants. The mineral fraction of SPM is particularly important because it governs many of these physicochemical processes and strongly influences contaminant mobility and bioavailability. However, its characterization remains challenging due to the small particle size, complex organo-mineral associations, poor crystallinity of some phases, and the dynamic nature of aquatic particles. This review provides a comprehensive overview of the principal methodologies currently available for investigating the mineral fraction of SPM. First, the advantages and limitations of the main sampling strategies are critically discussed. Then, analytical techniques are reviewed from bulk characterization methods to high-resolution approaches. This review emphasizes that no single analytical technique is sufficient to fully characterize natural SPM. Instead, integrating complementary methods across multiple spatial scales provides the most robust framework for understanding particle composition, reactivity, and contaminant dynamics. Future progress will rely on the harmonization of sampling and analytical protocols, the development of international SPM databases, and the combination of advanced laboratory experiments with field observations to improve our understanding of SPM processes under changing environmental conditions.

Abstract: Sacred forests play an important role in biodiversity conservation and climate change mitigation, yet their carbon storage characteristics remain poorly understood. This study quantified aboveground biomass (AGB) and aboveground carbon (AGC) stocks and examined their relationships with soil carbon–nutrient characteristics across edge-to-core spatial zones in Kaya Kauma Sacred Forest, a UNESCO-recognized sacred forest on the Kenyan coast. Forest inventories from ten permanent plots (385 trees representing 42 indigenous species) were used to estimate AGB and AGC, while soil samples collected at four depths (0–10, 10–20, 20–50, and 50–100 cm) were analyzed for total organic carbon (TOC), total nitrogen (TN), soil organic matter (SOM), soil inorganic carbon (SIC), and calcium carbonate (CaCO₃). Mean AGB and AGC were 204.2 ± 38.2 Mg ha⁻¹ and 96.0 ± 18.0 Mg C ha⁻¹, respectively, with no significant differences among forest zones (Kruskal–Wallis, H = 2.373, p = 0.305). Soil TOC, TN, and SOM declined with increasing depth, while deadwood exhibited the highest C:N ratios and regeneration material contained the highest nitrogen concentrations. No significant relationships were observed between AGC and soil variables (all p > 0.05). These findings provide baseline information on aboveground carbon stocks and soil carbon–nutrient characteristics in Kaya Kauma Sacred Forest, highlighting the importance of conserving sacred forests for biodiversity conservation and climate change mitigation.

Abstract: Coffee production and trade are globally important, and Brazil is the world's largest producer and second-largest consumer. However, climate change poses a major challenge to coffee production because the crop is highly sensitive to irregular rainfall distribution and temperature fluctuations. We evaluated temperature and precipitation fluctuations from 1984 to 2023 at two coffee-producing units in southern Minas Gerais, Brazil, to characterize regional climate dynamics and their impacts on coffee yield and total production. We used the Mann–Kendall test and Sen’s slope estimator to identify trends and estimate rates of change in maximum, minimum, and mean air temperature and in total, dry-season, and wet-season precipitation. We used Pearson’s and Spearman’s correlation coefficients to assess relationships between climatic and agronomic variables, selecting the appropriate coefficient based on prior Shapiro–Wilk normality tests. All temperature variables showed warming trends in September, during the transition from the dry to the wet season. Cooling trends were also detected in April, May, and December. At one production unit, dry-season rainfall declined over the historical series. At the other, precipitation declined in May but showed no trend in either the dry or wet season. Relationships between climatic and agronomic variables differed between production units: correlations were positive at one unit and negative at the other. Regionally, these trends indicate vulnerability during critical phenological stages that coincide with these seasonal windows, resulting in pollen tube desiccation, greater consumption of photoassimilates during physiological rest, and poor bean development under reduced dry-season rainfall. The contrasting correlation patterns were attributed to regional climate fluctuations and to the microclimatic, land-use, and land-cover characteristics of each production unit. Under this scenario, mitigating these impacts is essential to minimize damage to coffee crops. Recommended strategies include the implementation of agroforestry systems, rigorous monitoring of pests favored by climate fluctuations, and the adoption of more climate-resilient cultivars. Ultimately, this study highlights that crop management and mitigation strategies designed to buffer these climate impacts must be tailored to the distinct conditions of each production unit.

Abstract: Phosphonates in the ocean can serve as an alternative phosphorus (P) source for microorganisms when phosphate is scarce. Dinoflagellates cannot utilize phosphonates, but some associated bacteria can degrade these compounds and release phosphate. However, the community composition of these bacteria and their phosphonate degradation pathways remain poorly understood. In this study, the dinoflagellate Amphidinium carterae was cultured with 2-aminoethylphosphonic acid (2-AEP) as the exclusive P source. Metagenomic and genomic analyses were conducted to identify bacterial taxa and genes related to phosphonate utilization (phn genes). Specific strains were isolated from 2-AEP cultures to characterize phosphonate degradation gene clusters. Ten of the 20 most abundant genera were found to possess the genetic potential for phosphonate utilization, with Labrenzia, Phycocomes, Pseudosulfitobacter as the dominant genera. The identified phn genes constituted multiple pathways, including the C-P lyase, PhnW-PhnX, and PhnW-PhnY-PhnA pathways, indicating that A. carterae-associated bacteria can utilize phosphonates through diverse mechanisms, with some strains even possessing more than one pathway. Five isolated bacterial strains were confirmed to be capable of degrading 2-AEP. These findings underscore the ecological significance of bacterial diversity and metabolic versatility in helping dinoflagellate hosts adapt to P scarcity, providing novel insights into bacteria-algae interactions and marine P cycling.