Search Results (40)

The selection of environmental variables with different spatial resolutions is a critical factor affecting the accuracy of machine learning-based fishery forecasting. In this study, spring-season survey data of Decapterus macarellus in the South China Sea from 2016 to 2024 were used to construct six machine learning models—decision tree (DT), extra trees (ETs), K-Nearest Neighbors (KNN), light gradient boosting machine (LGBM), random forest (RF), and extreme gradient boosting (XGB)—based on seven environmental variables (e.g., sea surface temperature (SST), chlorophyll-a concentration (CHL)) at four spatial resolutions (0.083°, 0.25°, 0.5°, and 1°), filtered using Pearson correlation analysis. Optimal models were selected under each resolution through performance comparison. SHapley Additive exPlanations (SHAP) values were employed to interpret the contribution of environmental predictors, and the maximum entropy (MaxEnt) model was used to perform habitat suitability mapping. Results showed that the XGB model at 0.083° resolution achieved the best performance, with the area under the receiver operating characteristic curve (ROC_AUC) = 0.836, accuracy = 0.793, and negative predictive value = 0.862, outperforming models at coarser resolutions. CHL was identified as the most influential variable, showing high importance in both the SHAP distribution and the cumulative area under the curve contribution. Predicted suitable habitats were mainly located in the northern and central-southern South China Sea, with the latter covering a broader area. This study is the first to systematically evaluate the impact of spatial resolution on environmental variable selection in machine learning models, integrating SHAP-based interpretability with MaxEnt modeling to achieve reliable habitat suitability prediction, offering valuable insights for fishery forecasting in the South China Sea. Full article

The common leopard (Panthera pardus) in Azad Jammu and Kashmir (AJ and K), Pakistan, is increasingly threatened by habitat fragmentation and climate change. This study employs a dual-model approach, integrating Maximum Entropy (MaxEnt) and Random Forest algorithms with multi-source remote sensing data to evaluate leopard habitat suitability. Our analysis identifies land cover (LC), fractional vegetation cover (FVC), elevation, temperature seasonality (bio4), and distance to roads (Dist_road) as the most influential habitat predictors. Leopards exhibit a strong preference for mixed forests at elevations between 1000 and 3000 m, with a suitability index of 0.83. The study identifies several unsuitable conditions including: road proximity (<0.08 km), low elevation zones (<1000 m), areas with high temperature seasonality (bio4 > 8 °C), and non-forested land cover types. MaxEnt demonstrated superior habitat prediction accuracy over Random Forest (AUC = 0.912 vs. 0.827). The results highlight a distinct north-to-south suitability gradient, with optimal habitats concentrated in the northern districts (Muzaffarabad, Hattian, Neelum, Bagh, Haveli, Poonch, Sudhnutti) and declining suitability in human-dominated southern areas. Based on these findings, this study underscores the urgency of targeted conservation efforts in the northern districts of AJ and K, where optimal leopard habitats are identified. The findings emphasize the need for habitat connectivity and protection measures to mitigate the impacts of habitat fragmentation and climate change. Future conservation strategies should prioritize the preservation of mixed forests and the establishment of buffer zones around roads to ensure the long-term survival of the common leopard in this region. Full article

The northward expansion of Spartina alterniflora (S. alterniflora) poses a profound ecological threat to coastal ecosystems and biodiversity along China’s coastline. This invasive species exhibits strong adaptability to colder climates, facilitating its potential spread into northern regions and underscoring the urgent need for a nuanced understanding of its spatial distribution and invasion risks to inform evidence-based ecosystem management strategies. This study employed multi-temporal Sentinel-1/2 imagery (2016–2022) to map and predict the spread of S. alterniflora in Bohai Bay. An object-based random forest classification achieved an overall accuracy above 92% (κ = 0.978). Over the six-year period, the S. alterniflora distribution decreased from 46.60 km² in 2016 to 12.56 km² in 2022, reflecting an annual reduction of approximately 5.67 km². This decline primarily resulted from targeted eradication efforts, including physical removal, chemical treatments, and biological competition strategies. Despite this local reduction, MaxEnt modeling suggests that climate trends and habitat suitability continue to support potential northward expansion, particularly in high-risk areas such as the Binhai New District, the Shandong Yellow River Delta, and the Laizhou Bay tributary estuary. Key environmental drivers of S. alterniflora distribution include the maximum temperature of the warmest month, mean temperature of the wettest quarter, isothermality, sea surface temperature, mean temperature of the warmest quarter, and soil type. High-risk invasion zones, covering about 95.65 km². These findings illuminate the spatial dynamics of S. alterniflora and offer scientific guidance for evidence-based restoration and management strategies, ensuring the protection of coastal ecosystems and fostering sustainable development. Full article

Global change influences species diversity patterns. Compared with seed plants, ferns are more sensitive to temperature and humidity changes and are an ideal group for studying species diversity patterns under future climate changes. Polypodiaceae, which has important ecological and application value, such as medicinal and ornamental value, is one of the most widely distributed fern families, with rich species diversity. Here, we explore the changes in the species diversity patterns of Polypodiaceae and their influencing factors. We collected more than 300,000 data points on the distribution of Polypodiaceae to map actual current species diversity patterns. We used Maxent to establish current and future potential species distribution models using 20 predictors and determined the current species diversity patterns using the actual current species diversity patterns and current potential species distribution model method. Multiple linear regression and random forest models were used to evaluate the effects of climate factors on the species diversity patterns of Polypodiaceae. We evaluated the effects of future climate changes on the species diversity of Polypodiaceae. The species diversity of Polypodiaceae increased gradually from higher to lower latitudes and the centers were concentrated in the low latitudes of tropical rainforests. There were four distribution centers across the world for Polypodiaceae: central America, central Africa, southern Asia, and northern Oceania. The species diversity of Polypodiaceae was greatly affected by precipitation factors rather than temperature factors. Under future climate change scenarios, species diversity is expected to shift and accumulate toward the equator in mid-to-low latitudes. Species diversity is projected to remain concentrated in low-latitude regions but will tend to aggregate towards higher altitude areas as global temperatures rise, with precipitation during the warmest season identified as the most influential factor. Full article

Forest fires are the result of poor land management and climate change. Depending on the type of the affected eco-system, they can cause significant biodiversity losses. This study was conducted in the Amazonas department in Peru. Binary data obtained from the MODIS satellite on the occurrence of fires between 2010 and 2022 were used to build the risk models. To avoid multicollinearity, 12 variables that trigger fires were selected (Pearson ≤ 0.90) and grouped into four factors: (i) topographic, (ii) social, (iii) climatic, and (iv) biological. The program Rstudio and three types of machine learning were applied: MaxENT, Support Vector Machine (SVM), and Random Forest (RF). The results show that the RF model has the highest accuracy (AUC = 0.91), followed by MaxENT (AUC = 0.87) and SVM (AUC = 0.84). In the fire risk map elaborated with the RF model, 38.8% of the Amazonas region possesses a very low risk of fire occurrence, and 21.8% represents very high-risk level zones. This research will allow decision-makers to improve forest management in the Amazon region and to prioritize prospective management strategies such as the installation of water reservoirs in areas with a very high-risk level zone. In addition, it can support awareness-raising actions among inhabitants in the areas at greatest risk so that they will be prepared to mitigate and control risk and generate solutions in the event of forest fires occurring under different scenarios. Full article

Salurnis marginella causes agricultural and forest damage in various Asian environments. However, considering the environmental adaptability of pests and the active international trade, it may invade other regions in the future. As the damage to local communities caused by pests becomes difficult to control after invasion, it is essential to establish measures to minimize losses through pre-emptive monitoring and identification of high-risk areas, which can be achieved through model-based predictions. The aim of this study was to evaluate the potential distribution of S. marginella by developing multiple species distribution modeling (SDM) algorithms. Specifically, we developed the CLIMEX model and three machine learning-based models (MaxEnt, random forest, and multi-layer perceptron), integrated them to conservatively assess pest occurrence under current and future climates, and overlaid the host distribution with climatically suitable areas of S. marginella to identify high-risk areas vulnerable to the spread and invasion of the pest. The developed model, demonstrating a true skill statistic >0.8, predicted the potential continuous distribution of the species across the southeastern United States, South America, and Central Africa. This distribution currently covers approximately 9.53% of the global land area; however, the model predicted this distribution would decrease to 6.85%. Possible areas of spread were identified in Asia and the southwestern United States, considering the host distribution. This study provides data for the proactive monitoring of pests by identifying areas where S. marginella can spread. Full article

While geobotanists have long used plant occurrence to locate subsurface resources, none have utilised floristic surveys as evidence in models of mineral potential. Here, we combine plant species distributions with terrain metrics to produce predictive models showing the probability of bauxite presence. We identified nineteen taxa with statistically significant associations with known bauxite deposits and identified eleven terrain metrics from previous studies. We grouped variables into three variable sets (floristic, topographic, and topo-flora) and produced mineral potential models for each using four algorithms or approaches: (a) a generalised linear model (GLM); (b) random forest (RF); (c) maxent (ME); and (d) a heterogenous stacking ensemble (GLM-RF-ME). Overall, the random forest model outperformed all algorithms including the ensemble based on the area under the curve (AUC) metric. The floristic set of variables outperformed the topographic set (AUC: 0.86 v 0.82). However, together they had the greatest predictive capacity (AUC: 0.89). Six taxa, including Banksia grandis, Leucopogon verticillatus, and Persoonia longifolia, were indicators of bauxite presence, while five other taxa, including Xanthorrhoea preissii and Hypocalymma angustifolium, were associated with bauxite absence. Important topographic variables were topographic wetness, landscape position, and valley depth, which characterised bauxite locations as being well drained, in the upper slope positions of subdued hills, and at some distance from valleys. The addition of floristic surveys provides a new line of evidence about the overlying botanical life that tolerates, accumulates, or avoids bauxite or associated minerals. As opposed to drilling, both datasets can be collected and interrogated at low cost and without impact to the surrounding environment. These data are valuable additions to future applications of mineral potential modelling. Full article

Studying the current range of species presence is crucial for ecologists and related scientists to understand potential habitats and the influence of environmental factors on species distribution. In this study, we used species distribution modeling (SDM) to look into where the yellow-bellied gecko, also known as the northern house gecko (Hemidactylus flaviviridis Rüppell, 1835), lives in Iran. We achieved this by combining four machine learning algorithms: Random Forest (RF), the Support Vector Machine (SVM), Maximum Entropy (Maxent), and the Generalized Linear Model (GLM). We utilized 19 historical bioclimatic variables, the Digital Elevation Model (DEM), slope, aspect, and the Normalized Difference Vegetation Index (NDVI). After calculating their correlations, we selected variables for modeling with a variance inflation factor (VIF) of less than 10. The findings indicate that the variables “Precipitation of the Coldest Quarter” (BIO19) and “Mean Temperature of Wettest Quarter” (BIO8) have the most significant influence on the species’ distribution. The gecko primarily inhabits low elevations and slopes, particularly those below 400 m above sea level with slopes less than 8 degrees, primarily in southern Iran. Additionally, we found that the NDVI had a minimal impact on the distribution of the species. Therefore, we identify the provinces of Khuzestan, Bushehr, Hormozgan, and Fars, along with parts of the coastal strip of Sistan and Baluchistan, as suitable areas for the current presence of this species. Full article

Climate change will have various impacts on the survival and development of species, and it is important to explore whether plants can adapt to future climate conditions. Dipteronia dyeriana is an endangered species with a narrow distribution in Yunnan, characterized by a small population size. However, studies on its current distribution and the impact of climate change on its future survival and distribution are very limited. In this study, the current and future (2050 and 2090) potential habitats under the SSP1-2.6, SSP3-7.0, and SSP5-8.5 scenarios were predicted using both maximum entropy (MaxEnt) and random forest (RF) models based on the current range points of D. dyeriana, soil, topographical, land cover, and climate data. The results showed that the RF model demonstrated significantly higher AUC, TSS, and Kappa scores than the MaxEnt model, suggesting high accuracy of the RF model. Isothermality (bio_3), minimum temperature of the coldest month (bio_6), and annual precipitation (bio_12) are the main environmental factors affecting the distribution of D. dyeriana. At present, the high suitability area of D. dyeriana is mainly concentrated in the eastern part of Yunnan Province and part of southern Tibet, covering an area of 3.53 × 10⁴ km². Under future climate change scenarios, the total area suitable for D. dyeriana is expected to increase. Although, the highly suitable area has a tendency to decrease. With regards to land use, more than 77.53% of the suitable land area (29.67 × 10⁴ km²) could be used for D. dyeriana planting under different SSP scenarios. In 2090, the centroid shifts of the two models exhibit a consistent trend. Under the SSP1-2.6 scenario, the centroids transfer to the southeast. However, under the SSP3-7.0 and SSP5-8.5 scenarios, the centroids of high suitability areas migrate toward the northwest. In summary, this study enhances our understanding of the influence of climate change on the geographic range of D. dyeriana and provides essential theoretical backing for efforts in its conservation and cultivation. Full article

Benggang erosion is one of the most severe geomorphological hazards occurring on deeply weathered crusts in the hilly regions of southern China. Unraveling the susceptibility and pinpointing the risk areas of Benggang erosion are essential for developing effective prevention and management strategies. This study introduced the Maxent model to investigate Benggang erosion susceptibility (BES) and compared the evaluation results with the widely used Random Forest (RF) model. The findings are as follows: (1) the incidence of Benggang erosion is rising initially with an increase in elevation, slope, topographic wetness index, rainfall erosivity, and fractional vegetation cover, followed by a subsequent decline, highlighting its distinct characteristics compared to typical types of gully erosion; (2) the AUC values from the ROC curves for the Maxent and RF models are 0.885 and 0.927, respectively. Both models converge on elevation, fractional vegetation cover, rainfall erosivity, Lithology, and topographic wetness index as the most impactful variables; (3) both models adeptly identified regions prone to potential Benggang erosion. However, the Maxent model demonstrated superior spatial correlation in its susceptibility assessment, contrasting with the RF model, which tended to overestimate the BES in certain regions; (4) the Maxent model’s advantages include no need for absence samples, direct handling of categorical data, and more convincing results, suggesting its potential for widespread application in the BES assessment. This research contributes empirical evidence to study the Benggang erosion developing conditions in the hilly regions of southern China and provides an important consideration for the sustainability of the regional ecological environment and human society. Full article

Quantifying how climatic change affects wheat production, and accurately predicting its potential distributions in the face of future climate, are highly important for ensuring food security in Ethiopia. This study leverages advanced machine learning algorithms including Random Forest, Maxent, Boosted Regression Tree, and Generalised Linear Model alongside an ensemble approach to accurately predict shifts in wheat habitat suitability in the Central Ethiopia Region over the upcoming decades. An extensive dataset consisting of 19 bioclimatic variables (Bio1–Bio19), elevation, solar radiation, and topographic positioning index was refined by excluding collinear predictors to increase model accuracy. The analysis revealed that the precipitation of the wettest month, minimum temperature of the coldest month, temperature seasonality, and precipitation of the coldest quarter are the most influential factors, which collectively account for a significant proportion of habitat suitability changes. The future projections revealed that up to 100% of the regions currently classified as moderately or highly suitable for wheat could become unsuitable by 2050, 2070, and 2090, illustrating a dramatic potential decline in wheat production. Generally, the future of wheat cultivation will depend heavily on developing varieties that can thrive under altered conditions; thus, immediate and informed action is needed to safeguard the food security of the region. Full article

With the global energy crisis and the decline of fossil fuel resources, biofuels are gaining attention as alternative energy sources. China, as a major developing country, has long depended on coal and is now looking to biofuels to diversify its energy structure and ensure sustainable development. However, due to its large population and limited arable land, it cannot widely use corn or sugarcane as raw materials for bioenergy. Instead, the Chinese government encourages the planting of non-food crops on marginal lands to safeguard food security and support the biofuel sector. The Southern China Karst Region, with its typical karst landscape and fragile ecological environment, offers a wealth of potential marginal land resources that are suitable for planting non-food energy crops. This area is also one of the most impoverished rural regions in China, confronting a variety of challenges, such as harsh natural conditions, scarcity of land, and ecological deterioration. Idesia polycarpa, as a fast-growing tree species that is drought-tolerant and can thrive in poor soil, is well adapted to the karst region and has important value for ecological restoration and biodiesel production. By integrating 19 bioclimatic variables and karst landform data, our analysis reveals that the Maximum Entropy (MaxEnt) model surpasses the Random Forest (RF) model in predictive accuracy for Idesia polycarpa’s distribution. The karst areas of Sichuan, Chongqing, Hubei, Hunan, and Guizhou provinces are identified as highly suitable for the species, aligning with regions of ecological vulnerability and poverty. This research provides critical insights into the strategic cultivation of Idesia polycarpa, contributing to ecological restoration, local economic development, and the advancement of China’s biofuel industry. Full article

Climate change has massive global impacts and affects a wide range of species. Threatened species such as the roan antelope (Hippotragus equinus) are particularly vulnerable to these changes because of their ecological requirements. Attempts to address concerns about the roan’s vulnerability have not been well documented in South African protected areas. This study identifies the landscape use and distribution of the roan as well as habitat and forage suitability changes to help inform management decisions for the conservation of roan. We used fine- and broad-scale data from Mokala National Park, South Africa that includes roan occurrence data, vegetation condition indices, vegetation (structure and plant species composition), elevation and temperature differences, and precipitation strata to construct a suitability framework using the Maximum Entropy (Maxent) and Random Forest statistical package. In Mokala National Park, roan occurred in the Schmidtia pappophoroides–Vachellia erioloba sparse woodland, Senegalia mellifera–Vachellia erioloba closed woodland, Senegalia mellifera–Vachellia tortilis open shrubland, Vachellia erioloba–V. tortilis closed woodland and Rhigozum obovatum–Senegalia mellifera open shrubland. The veld (vegetation) condition index (VCI) improved from 2019 (VCI < 50%) to 2021 (VCI > 60%), with the proportion of palatable grass species (Schmidtia pappophoroides and Eragrostis lehmanniana) also increasing. This study identified four key climatic conditions affecting roan distribution, namely annual mean daily temperature range, temperature seasonality, minimum temperatures of the coldest month, and precipitation of the wettest month. These results suggest that the conservation of roan antelope should consider these key variables that affect their survival in preferred habitats and foraging areas in anticipation of changing ecological conditions. Full article

The mountaintops of eastern Brazil harbor the highest rates of plant endemism in South America. However, local biodiversity faces constant threats due to habitat loss and mining activities. About 89 rare and endangered species are exclusive to this region, including the threatened species Arthrocereus glaziovii. This study aims to evaluate the potential distribution of A. glaziovii based on abiotic variables and soil elements and to characterize the distribution of mineral titles that may restrict the species’ occurrence areas. We used the Bioclim, Domain, MaxEnt, GLM, and Random Forest algorithms to model this ecological niche under future climatic scenarios, in addition to modeling the layers of mineral titles corresponding to areas already mined and those slated for future mining projects. Our predictions indicate an expansion in the future distribution of A. glaziovii. Nevertheless, the future predicted occurrence areas of the species are already compromised due to mining. According to our findings, we emphasize the looming threat of the predicted extinction of this species. Therefore, implementing conservation strategies to ensure the survival of A. glaziovii is imperative. Full article

Globally, groundwater-influenced ecosystems (GIEs) are increasingly vulnerable to groundwater extraction and land use practices. Groundwater supports these ecosystems by providing inflow, which can maintain water levels, water temperature, and the chemistry necessary to sustain the biodiversity that they support. Many aquatic systems receive groundwater as a portion of baseflow, and in some systems, the connection with groundwater is significant and important to the system’s integrity and persistence. There is a lack of information about where these systems are found and their relationships with environmental conditions in the surrounding landscape. Additionally, groundwater management for human use often does not address maintaining the ecological functions of GIEs. We used correlative distribution modeling methods (GLM, GAM, MaxEnt, Random Forest) to predict landscape-scale habitat suitability for GIEs in two ecologically distinct ecoregions (EPA Level II ecoregions: Atlantic Highlands and Mixed Wood Plains) in the northeastern United States. We evaluated and combined the predictions to create ensemble models for each ecoregion. The accuracy of the ensemble models was 75% in the Atlantic Highlands and 86% in the Mixed Wood Plains. In the Mixed Wood Plains, hydric soil, surface materials, and soil permeability were the best predictors of GIE presence, whereas hydric soil, topographic wetness index, and elevation were the best predictors of GIE presence in the Atlantic Highlands. Approximately 1% of the total land area in each ecoregion was predicted to be suitable for GIEs, highlighting that there likely is a small proportion of the landscape occupied by these systems. Full article