Boosted Regression Tree Models Research Articles

Mitigation of reactive nitrogen loss from arable soils through microbial inoculant application: A meta-analysis

Excessive N fertilizer application contributes to a surge in reactive N (Nr) emissions from agroecosystems, consequently threatening global ecosystems’ quality and security. Adding functional microorganisms to agricultural soils, namely microbial inoculants (MI), is a promising method for ensuring sustainable agriculture, but general assessment of microorganism-driven N losses and mediating factors (such as climate conditions, soil properties, and field management practices) remains unclear. Therefore, a comprehensive meta-analysis on the effect of MI addition on NH3 volatilization, N2O emissions, and N leaching in soil-crop systems was conducted by integrating 50 published articles (188 paired observations). The MI application significantly reduced N2O emissions and NH3 volatilization by 29.8% and 24.9%, respectively. The boosted regression tree model showed that the soil organic carbon (SOC) and soil pH significantly affected the efficacy of MI application on both NH3 and N2O loss mitigation. The effects of MI application on N leaching were not clear due to limited data points. Additionally, MI addition significantly increased the abundance of denitrification functional genes (nirK, nirS, and nosZ). In conclusion, MI application presented a high potential to reduce Nr from croplands, future adoption of MI techniques should consider the effects of soil properties and field practices on its efficacy in order to achieving sustainable N management in crop production systems.

Urban spontaneous plant richness in response to the 2D/3D building and green space patterns in a highly urbanized area

The impact of landscape patterns on urban plant diversity has received significant attention; however, previous studies have primarily focused on two-dimensional (2D) patterns. Limited investigations have been conducted to explore the effects of three-dimensional (3D) landscape patterns on urban plant diversity. Spontaneous plants are ideal objects for studying the response of plant diversity to urban environments due to their independence from intentional human inputs. Buildings, as a crucial symbol of urbanization, exhibit strong 3D characteristics in urban landscapes. Green spaces, the most important source of plant propagules in cities, are closely related to spontaneous plant diversity in urban areas. Therefore, we examined how 2D/3D building and green space patterns influence urban spontaneous plant species richness and compared the responses between different categories (native and non-native). We surveyed spontaneous plants in the built-up areas of Shenzhen, a highly urbanized region situated along China’s southern coast. Our survey recorded 278 species belonging to 208 genera and 77 families. Results from boosted regression tree models (BRT) indicated significant correlations between spontaneous plant species richness and various metrics, including building coverage ratio (BCR), mean building height (MAH), green space coverage ratio (GCR), edge density of green spaces (ED_G), spatial congestion degree (SCD), and floor area ratio (FAR). These metrics exhibited complex nonlinear correlations with the species richness of spontaneous plants. Furthermore, native and non-native species richness responded differently to building and green space patterns; while native species richness was equally affected by green space and building patterns, non-native species richness was more strongly impacted by building patterns than green space ones. Our study quantified the nonlinear relationships between urban spontaneous plant species richness and 2D/3D green space and building patterns, demonstrating that optimizing green space and building patterns can be a practical approach for managing and promoting spontaneous plant diversity in highly urbanized areas.

Developing machine learning models with multiple environmental data to predict stand biomass in natural coniferous-broad leaved mixed forests in Jilin Province of China

Forest biomass is influenced by multiple environmental factors. Multi-source data on site-specific soil, climate, and stand factors are essential for stand biomass prediction. While previous studies concerning planted forest biomass at stand level highlighted the importance of environmental factors, few have been conducted on the effects of climate and soil on stand biomass estimation, especially for natural mixed forest with complicated stand structure. Machine learning (ML) algorithms provide new tools for examining the effects, but the applications are still limited. Therefore, the objectives of this paper were to develop ML models for stand biomass estimation with stand and multiple environmental predictors, and quantify the relative importance of environmental variables. Tree above- and below-ground biomass at stand level was set as output variable with the random forest (RF) and boosted regression tree (BRT) regression methods. The data were from 286 sample plots from the National Forest Inventory (NFI) in Jilin Province, northeast China. The results showed that models with the inclusion of stand, soil, and climate variables explained > 89% of forest biomass variations. BA (stand basal area) and H (stand average height) were the most important stand variables, DD18 (degree-days above 18 °C) was the most important climate variable, and bdod (bulk density) and pH were the most important soil variables. Relative importance of climate and soil factors was 0.5% to 5.1% and 1.2% to 4.1%, respectively, which were weak but important predictors of stand biomass. In the RF model, climate variables were more important than soil variables, but the opposite in the BRT models. We concluded that above forest stand, climate and soil factors were significant input variables for regional forest biomass mapping using ML.

The Influence of the Board Bending and Recoil Effect on Circus Performance in Korean Teeterboard

The purpose of this study was to examine the recoil effect in jumping on a Korean teeterboard, a circus equipment resembling a seesaw. To our knowledge, our study is the first to include detailed measurements of bending movement of the teeterboard per section. Two elite acrobats performed 120 jumps while a motion capture system collected kinematic data from both acrobats and the teeterboard. Board bending angles and timing were analyzed with a Boosted Regression Tree (BRT) model to identify the most important teeterboard variables associated to jump height. The BRT model showed that both the board recoil effect and the landing of the opposite acrobat influenced 46% and 37% of the jump height, respectively. The recoil timing was found to be synchronized with the last contact at take-off. Coaches should encourage acrobats to take advantage of the recoil effect to increase jump height in Korean teeterboard.

Global pattern and drivers of stable residue size from decomposing leaf litter

Stable residue from decomposing leaf litter is related to humus buildup in the forest floor and thus to the accumulation of carbon in the soil organic layer (OLC). However, the global pattern and controls of stable residue size remains poorly understood. Herein, the spatial pattern of stable residue size and its association with a series of potential drivers were evaluated by synthesizing available data worldwide. The results showed that stable residue size had a significant and positive correlation with the litter initial nitrogen (N) and lignin concentrations, and a negative correlation with the initial manganese concentration. Larger stable residues were found from decomposing broadleaf litter vs. coniferous litter. The decomposition of green leaf litter produces an equal amount of stable residue as that of brown litter. Stable residue size correlated quadratically with mean annual temperature and latitude, and linearly with mean annual precipitation. Methodological factor also matters, because stable residue size decreased as field incubation time increased. The boosted regression tree model indicated that litter initial traits have the most explanative ability for the variance in stable residue size. We highlighted that litter initial traits exert a predominant role over climate in shaping the stable residue size globally. Our findings are beneficial for a more accurate prediction of global-scale OLC accumulation.

Estimation of forest canopy closure in northwest Yunnan based on multi-source remote sensing data colla-boration.

Forest canopy closure (FCC) is an important parameter to evaluate forest resources and biodiversity. Using multi-source remote sensing collaborative means to achieve regional forest canopy closure inversion with low cost and high-precision is a research hotspot. Taking ICESat-2/ATLAS data as the main information source and combined with data of 54 measured plots, we estimated FCC value by the Bayesian optimization (BO) algorithm improved random forest (RF), K-nearest neighbor (KNN), and gradient boosting regression tree (GBRT) model at footprint-scale. Combined with multi-source remote sensing image Sentinel-1/2 and terrain factors, we estimated the regional-scale FCC value of Shangri-La in the northwest Yunnan based on deep neural network (DNN) optimized by BO algorithm. The results showed that six characteristic parameters (percentage of tree canopy, standard deviation of relative height of photons at the top of the canopy, minimum canopy height, difference between 98% canopy height and median canopy height in the segment, number of top canopy photons, apparent surface reflectance) out of the 50 parameters that were extracted from ATLAS lidar footprint had higher contribution rate after RF characteristic variable optimization, which could be used as model variable for footprint-scale remote sensing estimation. Among BO-RF, BO-KNN, and BO-GBRT models, the FCC results estimated by the BO-GBRT model were the best at footprint-scale. The coefficient of determination (R2) was 0.65, the root mean square error (RMSE) was 0.10, the mean absolute residual (RS) was 0.079, and the prediction accuracy (P) was 0.792 for leave-one-out cross validation. It could be used as the FCC estimation model of 74808 ATLAS footprints for forest in the study area. We used the ATLAS footprint-scale FCC value of forest as the large sample data of the regional-scale BO-DNN model and combined with multi-source remote sensing factors to estimate FCC in the study area, the accuracy of the 10-fold cross-validation BO-DNN model was R2=0.47, RMSE=0.22, P=0.558. The mean values of FCC in the study area estimated by BO-DNN model and ordinary Kriging (OK) interpolation were 0.46 and 0.52, respectively, and the values mainly distributed in 0.3-0.6, accounting for 77.8% and 81.4%, respectively. The FCC efficiency obtained directly by the OK interpolation method was higher (R2=0.26), but the prediction accuracy was significantly lower than the BO-DNN model (R2=0.49). The FCC high value was distributed from northwest to southeast in the study area, and the northern and southeastern regions were the main distribution areas of high and low FCC values, respectively. It had certain advantages to estimate mountain area FCC based on ICESat-2/ATLAS high-density footprint, and the estimation results of small sample data at footprint-scale could be used as large sample data of deep learning model at region-scale, which would provide a reference for the low-cost and high-precision to FCC estimation on the footprint-scale up to the extrapolated regional-scale.

A dataset of the observations of carbon, water and heat fluxes over an alpine meadow in Haibei (2015–2020)

Covering an area of 5.04×105 km2 on the Qinghai-Tibetan Plateau, Alpine meadow is essential to the plateau ecological barrier function and regional sustainable development. Since May 2014, Haibei National Field Research Station for Alpine Grassland (Haibei Station hereafter) has been accumulating amounts of valuable data by employing eddy covariance techniques to continuously measure carbon and water cycles and energy exchanges between an alpine Graminoid-Kobresia meadow ecosystem and the atmosphere. In the following of data processing such as outlier removal and flux data gaps filled by boosted regression tree model, Haibei Station plans to publish a dataset of the continuous observations of carbon, water, and heat fluxes of the alpine meadow from 2015 to 2010. This dataset consists of the subsets of carbon, water, and heat fluxes data (i.e. net ecosystem CO2 exchange, ecosystem CO2 respiration, gross ecosystem CO2 exchange, latent heat flux, and sensible heat flux) and the subsets of routine meteorological data (i.e. air temperature, air relative humidity, total solar radiation, net radiation, photosynthetically active radiation, precipitation, soil temperature, volumetric soil moisture content). The temporal resolutions of the dataset are half-hourly, daily, monthly, and yearly scales. This dataset can be used to validate the parameters of processes-based ecological models of carbon and water cycles and to evaluate the spatiotemporal patterns and evolution trends in ecological functions of carbon sequestration and water-holding capacity in alpine meadow ecosystems.

Prediction of higher heating value of coal based on gradient boosting regression tree model

Higher heating value, also known as the coal calorific value, is an important indicator of coal quality. Nevertheless, traditional experimental determination of higher heating value is expensive, tedious, and time-consuming. Some regression models have been commonly used to solve this problem. However, the prediction accuracy still needs to be improved. Accordingly, based on the correlation between proximate analysis of coal and higher heating value, a gradient boosting regression tree model was established in this study to predict the higher heating value as a means to a less intensive procedure to estimate coal heating values. This approach used data from the U.S. Geological Survey Coal Quality database and Kentucky Coal Database to evaluate coal quality. The hyperparameters of the regression tree model were tuned by cross-validation. A comparative study was conducted between the model established in this paper and the existing methods. Experimental results indicate that the coefficient of determination of the gradient boosting regression tree model of two databases achieves 0.9862 and 0.9541, respectively. Thus, the prediction capability of the gradient boosting regression tree model outperforms the existing methods.

Assessing the carbon sequestration potential and identifying influential factors of cultivated soils in Northeast China

Cultivated lands play a crucial role in terrestrial carbon cycle, and enhancing soil organic carbon (SOC) sequestration in these areas can help effectively mitigate the rise of atmospheric CO2 concentration. In this study, topsoil (0–20 cm) saturated SOC and its density of cultivated soil in Northeast China were mapped using a boosted regression trees (BRT) model. The distribution of the SOC sequestration potential was also calculated based on the difference between saturated SOC and SOC density in ArcGIS. Nine environmental factors including climate, topography and lengths of cultivation data (LCD) and 197 soil samples are used. A 10-fold cross-validation technique is applied to derive four statistical indices - mean absolute prediction error (MAE), root mean square error (RMSE), coefficient of determination (R2) and Lin's consistent correlation coefficient (LCCC) to verify the model performance. The model explains 81% and 85% of the spatial variation of saturated SOC and SOC density, respectively. Mean annual temperature and mean annual precipitation are key factors controlling SOC density and saturated SOC distribution. In addition, LCD showed a similar spatial pattern to SOC sequestration potential, influencing the distribution of SOC density and saturated SOC in the study area. We recommend LCD as an important factor to consider in saturated SOC and SOC density predictions, especially in the farmland ecosystem with a long reclamation history. Accurate mapping SOC sequestration potential and identifying environmental factors will help manage land use and promote soil quality evaluation and improve soil carbon sequestration in this region.

Factors influencing the establishment of vascular plants at Sphagnum cultivation sites

Sphagnum cultivation is a type of paludiculture and a way to use formerly drained peatlands productively but under wet and therefore climate-friendly conditions. Where Sphagnum mosses are cultivated other plant species will also establish and possibly compete with the Sphagnum. The aim of this study was to determine which factors influence vascular plant cover as well as plant species numbers at Sphagnum cultivation sites and to derive recommendations for their management. Two cultivation sites were studied in northwest Germany. One of these was established directly after peat extraction while the other was rewetted seven years prior to establishment. Irrigation ditches for water management were installed at both sites. The cover of vascular plants and the number of plant species present were determined in systematically positioned plots. Six variables were tested for their influence on the assessed data by applying boosted regression tree models. The main factors influencing vascular plant cover at the two Sphagnum cultivation sites were the distance to an irrigation ditch (m), the site (location) and Sphagnum cover (%). The number of species per plot was influenced mainly by Sphagnum cover (%), the distance to an irrigation ditch (m) and the donor species used for initiating the cultivation sites. A sufficient supply of nutrient-poor water and optimal Sphagnum growth can reduce vascular plant cover and the number of plant species potentially present at a site. Insufficient water distribution and uneven Sphagnum establishment lead to inhomogeneous site conditions and thus to a higher number of plant species. The number and cover of plant species at a cultivation site are influenced by the vegetation of the sites’ surroundings and the selection of the donor site.

Effect of Spartina alterniflora Invasion on Soil C:N:P Stoichiometry in Coastal Wetland of Hangzhou Bay

The invasion of Spartina alterniflora poses a great threat to coastal wetland ecosystems. In this study, the stoichiometric characteristics of soil carbon, nitrogen, and phosphorus under a Spartina alterniflora invasion were explored using ANOVA in a coastal wetland in Hangzhou Bay, and the driving coupling relationship between soil environmental factors and soil C:N:P stoichiometric characteristics of the coastal wetland were further explored based on the redundancy analysis (RDA), boosted regression tree (BRT), and partial least squares-structural equation (PLS-SEM) model. The results showed that:① after the invasion of Spartina alterniflora, soil N:P and total nitrogen (TN) in the wetland increased significantly, and with the increase in invasion time, TN and N:P decreased significantly, whereas soil organic carbon (SOC), C:N, and C:P increased significantly. ② The RDA model revealed that the main factors affecting the stoichiometric characteristics of topsoil C:N:P were SOC>electrical conductivity (EC)>TN in winter and SOC>bulk density (BD)>TN in summer. ③ The BRT model showed that under the invasion of Spartina alterniflora, TN was the key factor affecting soil C:N and N:P, and SOC was the key factor affecting C:P. ④ The PLS-SEM model showed that clay and water content directly affected SOC, thus affecting C:N and C:P; the clay and EC directly affected total phosphorus (TP), thus affecting N:P and C:P; and the EC directly affected TN, thus affecting C:N and N:P. In conclusion, the invasion of Spartina alterniflora had a significant impact on soil C:N:P stoichiometric characteristics in the study area. Soil physical properties and nutrient content directly or indirectly affected soil C:N:P stoichiometric characteristics to varying degrees.

Trait interactions effects on tropical tree demography depend on the environmental context

Although functional traits are defined based on their impact on demographic parameters, trait-demography relationships are often reported as weak. These weak relationships might be due to disregarding trait interactions and environmental contexts, which should modulate species trait-demography relationships. We applied different models, including boosted regression tree (BRT) models, to investigate changes in the relationship between traits and demographic rates of tropical tree species in plots along an elevational gradient and among time intervals between censuses, analyzing the effect of a strong drought event. Based on a large dataset of 18,000 tree individuals from 133 common species, distributed among twelve 1-ha plots (habitats) in the Atlantic Forest (Brazil), we evaluated how trait interactions and the environmental context influence the demographic rates (growth, mortality, and recruitment). Functional traits, trait-trait, and trait-habitat interactions predicted demography with a good fit through either BRTs or linear mixed-models. Changes in growth rates were best related to size (diameter), and mortality rates to habitats, while changes in recruitment rates were best related to the specific leaf area. Moreover, the influence of traits differed among time intervals, and for demographic parameters, habitat affected growth and mortality by interacting with diameter. Here, we provide evidence that trait-demography relationships can be improved when considering the environmental context (space and time) and trait interactions to cope with the complexity of changes in the demography of tropical tree communities. Thus, to expand predictions of demography based on functional traits, we show that it is useful to fully incorporate the concept of multiple trait-fitness optima, resulting from trait interactions in different habitats and growth conditions.

Quantifying the nonlinear response of vegetation greening to driving factors in Longnan of China based on machine learning algorithm

The main influencing factors and their nonlinear effects on the changes of vegetation in China’s mountainous areas under the interaction of different factors are not yet clear, and comprehending the evolutionary trends and driving mechanisms of vegetation is crucial to reveal the changes in ecosystem structure and function. In this study, trend analysis (M−K, T-S and EEMD) combined with machine learning algorithm, namely Boosted Regression Tree model (BRT), were used to quantify the trends of nonlinear responses and thresholds for bioclimatic variables, topography, soil properties and anthropogenic factors for vegetation changes in Longnan. The results showed that the trend analysis clearly confirm the increasing trend of vegetation at multiple spatio-temporal scales. The BRT indicated that total precipitation (bio12, 15.22%), land use (LUCC, 12.68%), elevation (DEM, 11.20%), and population density (Pd, 9.20%) were the more important factors of dominant vegetation greening. Bioclimatic variables were found to revealed the effects of climate with vegetation more clearly. In addition, the BRT revealed that the selected factors have the different nonlinear response relationships to vegetation greening trend and specific thresholds. Among them, increasing of cropland, grassland and forestland can promote vegetation greening. However, GDP, Pd, DEM, bio12, mean diurnal range and temperature seasonality (bio2, bio4) exceed the threshold can significantly inhibit vegetation growth. The BRT combined with trend analysis revealed the nonlinear response relationships and thresholds of the drivers behind the vegetation change patterns, which have obvious effects in exploring the driving mechanisms of vegetation changes in mountainous areas. This study provided an important reference for better revealing the interaction mechanisms between vegetation changes and drivers in the semi-humid zone in East Asia even globally.

Spatial Prediction and Mapping of Gully Erosion Susceptibility Using Machine Learning Techniques in a Degraded Semi-Arid Region of Kenya

This study aimed at (i) developing, evaluating and comparing the performance of support vector machines (SVM), boosted regression trees (BRT), random forest (RF) and logistic regression (LR) models in mapping gully erosion susceptibility, and (ii) determining the important gully erosion conditioning factors (GECFs) in a Kenyan semi-arid landscape. A total of 431 geo-referenced gully erosion points were gathered through a field survey and visual interpretation of high-resolution satellite imagery on Google Earth, while 24 raster-based GECFs were retrieved from the existing geodatabases for spatial modeling and prediction. The resultant models exhibited excellent performance, although the machine learners outperformed the benchmark LR technique. Specifically, the RF and BRT models returned the highest area under the receiver operating characteristic curve (AUC = 0.89 each) and overall accuracy (OA = 80.2%; 79.7%, respectively), followed by the SVM and LR models (AUC = 0.86; 0.85 & OA = 79.1%; 79.6%, respectively). In addition, the importance of the GECFs varied among the models. The best-performing RF model ranked the distance to a stream, drainage density and valley depth as the three most important GECFs in the region. The output gully erosion susceptibility maps can support the efficient allocation of resources for sustainable land management in the area.

Model-based assessment and mapping of total phosphorus enrichment in rivers with sparse reference data

Water nutrient management efforts are frequently coordinated across thousands of water bodies, leading to a need for spatially extensive information to facilitate decision making. Here we explore potential applications of a machine learning model of river low-flow total phosphorus (TP) concentrations to support landscape nutrient management. The model was trained, validated, and then applied for all rivers of Michigan, USA to identify potential drivers of nutrient variation, predict alteration in nutrient concentrations from minimally disturbed conditions, and explore reach-specific sensitivity to riparian agricultural change. A boosted regression tree model of low-flow TP concentrations trained on natural and anthropogenic landscape predictors accounted for 53 % of variation in cross-validation data, had good accuracy, little bias, and plausible relationships between predictors and response. Percent riparian agricultural cover accounted for the greatest root mean square error reduction in the modeled response (33.2 %), followed by riparian soil permeability (12.9 %), watershed slope (9.6 %), and percent urban cover (9.6 %). An apparent non-linear relationship between TP concentrations and percent riparian agricultural cover suggested steep positive increases in stream TP concentrations between 10 and 30 % upstream riparian agricultural cover. Predicted minimally disturbed TP concentrations were spatially variable and ranged from 7.0 to 48.5 μg/L, with the highest concentrations in watersheds draining low-permeability lake plain soils. Comparison of minimally disturbed predictions to those from the early 2000s suggested that much of northern Michigan existed close to the reference condition, while southern Michigan streams were often substantially enriched. Our predicted values of minimally disturbed condition generally agreed with previous studies but offer greater geographic specificity. Expanded application of machine learning modeling with landscape predictor data have great potential to inform stream nutrient strategy development in settings with sparse reference data.

Prediction of the Potential Distribution of the Endangered Species Meconopsis punicea Maxim under Future Climate Change Based on Four Species Distribution Models.

Climate change increases the extinction risk of species, and studying the impact of climate change on endangered species is of great significance to biodiversity conservation. In this study, the endangered plant Meconopsis punicea Maxim (M. punicea) was selected as the research object. Four species distribution models (SDMs): the generalized linear model, the generalized boosted regression tree model, random forest and flexible discriminant analysis were applied to predict the potential distribution of M. punicea under current and future climates scenarios. Among them, two emission scenarios of sharing socio-economic pathways (SSPs; i.e., SSP2-4.5 and SSP5-8.5) and two global circulation models (GCMs) were considered for future climate conditions. Our results showed that temperature seasonality, mean temperature of coldest quarter, precipitation seasonality and precipitation of warmest quarter were the most important factors shaping the potential distribution of M. punicea. The prediction of the four SDMs consistently indicated that the current potential distribution area of M. punicea is concentrated between 29.02° N-39.06° N and 91.40° E-105.89° E. Under future climate change, the potential distribution of M. punicea will expand from the southeast to the northwest, and the expansion area under SSP5-8.5 would be wider than that under SSP2-4.5. In addition, there were significant differences in the potential distribution of M. punicea predicted by different SDMs, with slight differences caused by GCMs and emission scenarios. Our study suggests using agreement results from different SDMs as the basis for developing conservation strategies to improve reliability.

Contrasting determinants of land surface temperature in three megacities: Implications to cool tropical metropolitan regions

Megacities in tropical regions are among the urban centers most vulnerable to increasingly intense heatwaves. However, the complex interactions between urban characteristics and thermal environments are yet to be fully understood. Here, we investigated the relationship between urban characteristics and land surface temperature (LST) for three megacities in the tropical savannah climate zone, Chennai, Dhaka, and Kolkata. LST values were retrieved from Landsat 8 data, and urban features were constructed using digital surface models, building footprints, and satellite imagery. Model-agnostic approaches were adopted to analyze the impacts of urban characteristics on LST. The boosted regression tree (BRT) model was used to capture the nonlinear effects of six most important factors on LST changes. The dominant influencing factors of LST were normalized difference built-up index, albedo, and normalized difference vegetation index. Building roofs, roads, and other hardscapes with high-albedo materials are preferred in alleviating surface urban heat island effect in Dhaka and Kolkata. The placement of urban parks with woody vegetation in an area larger than 0.6 ha produced effective cooling effect. In Chennai and Kolkata, urban policies to increase sky view factor are recommended. The application of BRT model with integrated urban data provides unique insights for thermal resilience planning.

Quantifying blue carbon stocks and the role of protected areas to conserve coastal wetlands

Vegetated coastal ecosystems, in particular mangroves, tidal marshes and seagrasses are highly efficient at sequestering and storing carbon, making them valuable assets for climate change mitigation and adaptation. The state of Queensland, in northeastern Australia, contains almost half of the total area of these blue carbon ecosystems in the country, yet there are few detailed regional or state-wide assessments of their total sedimentary organic carbon (SOC) stocks. We compiled existing SOC data and used boosted regression tree models to evaluate the influence of environmental variables in explaining the variability in SOC stocks, and to produce spatially explicit blue carbon estimates. The final models explained 75 % (for mangroves and tidal marshes) and 65 % (for seagrasses) of the variability in SOC stocks. Total SOC stocks in the state of Queensland were estimated at 569 ± 98 Tg C (173 ± 32 Tg C, 232 ± 50 Tg C, and 164 ± 16 Tg C from mangroves, tidal marshes and seagrasses, respectively). Regional predictions for each of Queensland's eleven Natural Resource Management regions revealed that 60 % of the state's SOC stocks occurred within three regions (Cape York, Torres Strait and Southern Gulf Natural Resource Management regions) due to a combination of high values of SOC stocks and large areas of coastal wetlands. Protected areas in Queensland play an important role in conserving SOC assets in Queensland's coastal wetlands. For example, ~19 Tg C within terrestrial protected areas, ~27 Tg C within marine protected areas and ~ 40 Tg C within areas of matters of State Environmental Significance. Using multi-decadal (1987–2020) mapped distributions of mangroves in Queensland; we found that mangrove area increased by approximately 30,000 ha from 1987 to 2020, which led to temporal fluctuations in mangrove plant and SOC stocks. We estimated that plant stocks decreased from ~45 Tg C in 1987 to ~34.2 Tg C in 2020, while SOC stocks remained relatively constant from ~107.9 Tg C in 1987 to 108.0 Tg C in 2020. Considering the level of current protection, emissions from mangrove deforestation are potentially very low; therefore, representing minor opportunities for mangrove blue carbon projects in the region. Our study provides much needed information on current trends in carbon stocks and their conservation in Queensland's coastal wetlands, while also contributing to guide future management actions, including blue carbon restoration projects.

Using Machine Learning to Predict the Performance of a Cross-Flow Ultrafiltration Membrane in Xylose Reductase Separation

This study provides a new perspective for xylose reductase enzyme separation from the reaction mixtures—obtained in the production of xylitol—by means of machine learning technique for large-scale production. Two types of machine learning models, including an adaptive neuro-fuzzy inference system based on grid partitioning of the input space and a boosted regression tree were developed, validated, and tested. The models’ inputs were cross-flow velocity, transmembrane pressure, and filtration time, whereas the membrane permeability (called membrane flux) and xylitol concentration were considered as the outputs. According to the results, the boosted regression tree model demonstrated the highest predictive performance in forecasting the membrane flux and the amount of xylitol produced with a coefficient of determination of 0.994 and 0.967, respectively, against 0.985 and 0.946 for the grid partitioning-based adaptive neuro-fuzzy inference system, 0.865 and 0.820 for the best nonlinear regression picked from among 143 different equations, and 0.815 and 0.752 for the linear regression. The boosted regression tree modeling approach demonstrated a superior capability of predictive accuracy of the critical separation performances in the enzymatic-based cross-flow ultrafiltration membrane for xylitol synthesis.

Temporal variations of precipitation driven by local meteorological parameters in southwest China: Insights from 9 years of continuous hydro-meteorological and isotope observations

Study regionNorthwest of Guangxi Province, southwest China. Study focusKnowing the role of isotopes in precipitation can improve understanding of climate change and eco-hydrological processes in complex critical-zone systems. However, knowledge of the influence of local meteorological factors on variations of precipitation isotopes remains limited as high-resolution rainwater sampling is lacking. New hydrological insights for the regionThe temporal characteristics (intra-annual, seasonal, and monthly scales) of precipitation were analyzed using daily precipitation isotopes data from 2011 to 2020 and event-resolving data from 2019 to 2020 at Huanjiang in southwest China. The boosted regression tree model was used to explore the relationship between local meteorological factors and δ18O values. On an intra-annual scale, δ18O, δD, and d-excess values exhibited significant sinusoidal fluctuation, being lower and higher during the wet and dry seasons respectively. Temperature is the most sensitive meteorological factor to monsoon climate response. On the seasonal scale, ET0 explains 37.5% of isotopic variations in the dry season, thus indicating that the influence of evaporation on the rainwater isotopes variation should be considered. The “amount effect” of rainwater isotopes was clearly shown on the monthly scale. However, precipitation isotopes did not follow the “amount effect” on an inter-event scale, which can be explained by large-scale convection activity. These findings can improve interpretations of isotopic variations in precipitation in humid subtropical climate zones.