Random Forests Classification Tree Research Articles

Suitable climatic habitat changes for Mexican conifers along altitudinal gradients under climatic change scenarios.

The high biodiversity of the Mexican montane forests is concentrated on the Trans-Mexican Volcanic Belt, where several Protected Natural Areas exist. Our study examines the projected changes in suitable climatic habitat for five conifer species that dominate these forests. The species are distributed sequentially in overlapping altitudinal bands: Pinus hartwegii at the upper timberline, followed by Abies religiosa, the overwintering host of the Monarch butterfly at the Monarch Butterfly Biosphere Reserve, P. pseudostrobus, the most important in economic terms, and P. devoniana and P. oocarpa, which are important for resin production and occupy low altitudes where montane conifers merge with tropical dry forests. We fit a bioclimatic model to presence-absence observations for each species using the Random Forests classification tree with ground plot data. The models are driven by normal climatic variables from 1961 to 1990, which represents the reference period for climate-induced vegetation changes. Climate data from an ensemble of 17 general circulation models were run through the classification tree to project current distributions under climates described by the RCP 6.0watts/m2 scenario for the decades centered on years 2030, 2060 and 2090. The results suggest that, by 2060, the climate niche of each species will occur at elevations that are between 300 to 500m higher than at present. By 2060, habitat loss could amount to 46-77%, mostly affecting the lower limits of distribution. The two species at the highest elevation, P. hartwegii and A. religiosa, would suffer the greatest losses while, at the lower elevations, P. oocarpa would gain the most niche space. Our results suggest that conifers will require human assistance to migrate altitudinally upward in order to recouple populations with the climates to which they are adapted. Traditional insitu conservation measures are likely to be equivalent to inaction and will therefore be incapable of maintaining current forest compositions.

AN OBJECT-BASED METHOD FOR CHINESE LANDFORM TYPES CLASSIFICATION

Landform classification is a necessary task for various fields of landscape and regional planning, for example for landscape evaluation, erosion studies, hazard prediction, et al. This study proposes an improved object-based classification for Chinese landform types using the factor importance analysis of random forest and the gray-level co-occurrence matrix (GLCM). In this research, based on 1km DEM of China, the combination of the terrain factors extracted from DEM are selected by correlation analysis and Sheffield's entropy method. Random forest classification tree is applied to evaluate the importance of the terrain factors, which are used as multi-scale segmentation thresholds. Then the GLCM is conducted for the knowledge base of classification. The classification result was checked by using the 1:4,000,000 Chinese Geomorphological Map as reference. And the overall classification accuracy of the proposed method is 5.7% higher than ISODATA unsupervised classification, and 15.7% higher than the traditional object-based classification method.

AN OBJECT-BASED METHOD FOR CHINESE LANDFORM TYPES CLASSIFICATION

Abstract. Landform classification is a necessary task for various fields of landscape and regional planning, for example for landscape evaluation, erosion studies, hazard prediction, et al. This study proposes an improved object-based classification for Chinese landform types using the factor importance analysis of random forest and the gray-level co-occurrence matrix (GLCM). In this research, based on 1km DEM of China, the combination of the terrain factors extracted from DEM are selected by correlation analysis and Sheffield's entropy method. Random forest classification tree is applied to evaluate the importance of the terrain factors, which are used as multi-scale segmentation thresholds. Then the GLCM is conducted for the knowledge base of classification. The classification result was checked by using the 1:4,000,000 Chinese Geomorphological Map as reference. And the overall classification accuracy of the proposed method is 5.7% higher than ISODATA unsupervised classification, and 15.7% higher than the traditional object-based classification method.

Pinus leiophylla suitable habitat for 1961-1990 and future climate

Our objectives were to predict and map the climatic niche for Pinus leiophylla for a period of normalization (years 1961-1990) and future (2030, 2060 and 2090) climates, and to suggest management strategies to accommodate climate changes, and discuss implications for conservation. A bioclimate model predicting the presence or absence of P. leiophylla (lumped with its putative variety P. leiophylla var. chihuahuana ) was developed by using the Random Forests classification tree on Mexican and Unites States of America forest inventory data. The bioclimatic model had an average error of prediction of 4.6 %. The model used six predictor variables, dominated by precipitation variables. Projecting the 1961-1990 climate niche into future climates provided by three general circulation models and two greenhouse-effect gas emission scenarios, suggested that the area occupied by the niche should diminish rapidly over the course of the century: a decrease of 35 % by the decade surrounding 2030, 50 % for 2060, and 76 % for 2090. The most serious habitat reduction occurs at both latitudinal extremes of the species distribution: Chiricagua Mountains, Arizona, Unites States of America in the northern extreme, and at Oaxaca State, Mexico, in the southernmost extreme. There is no indication at all of expansion of suitable climatic habitat northwards. We urge establishing seed banks encompassing seed from provenances sampled from the largest part possible of the natural distribution, and start assisted migration tests, to realign the natural populations with the climate for which they are adapted and that will occur at higher altitudes.

Adapting forest management to climate change using bioclimate models with topographic drivers

Bioclimate models incorporating topographic predictors as surrogates for microclimate effects are developed for Populus tremuloides and Picea engelmannii to provide the fine-grained specificity to local terrain required for adapting management of three Colorado (USA) national forests (1.28 million ha) and their periphery to climate change. Models were built with the Random Forests classification tree using presence–absence observations obtained by overlaying species distribution maps on data points gridded at ∼225 m within the forests and from ground plot observations from adjacent areas. Topographic effects derived from 90-m elevation grids were expressed by weighting aspect by slope angle. Climate estimates were obtained from spline surfaces. Out-of-bag errors were ∼17 per cent, and classification errors for an independent sample from within the forest were ∼13 per cent. Topographic variables were second in importance to climate variables for predicting species distributions; their inclusion captured well-known topographic effects on vegetation in mountainous terrain. Predictions made for future climates described by three General Circulation Models and three emissions scenarios were used to map on 90-m grids the habitat expected to be lost, threatened, persistent or emergent. The habitat categories are used to identify those areas where treatments should have highest likelihood of success.

Potentialities of aberrantly methylated circulating DNA for diagnostics and post-treatment follow-up of lung cancer patients

To date, aberrant DNA methylation has been shown to be one of the most common and early causes of malignant cell transformation and tumors of different localizations, including lung cancer. Cancer cell-specific methylated DNA has been found in the blood of cancer patients, indicating that cell-free DNA circulating in the blood (cirDNA) is a convenient tumor-associated DNA marker that can be used as a minimally invasive diagnostic test. In the current study, we investigated the methylation status in blood samples of 32 healthy donors and 60 lung cancer patients before and after treatment with neoadjuvant chemotherapy followed by total tumor resection. Using quantitative methylation-specific PCR, we found that the index of methylation (IM), calculated as IM=100×[copy number of methylated/(copy number of methylated+unmethylated gene)], for the RASSF1A and RARB2 genes in the cirDNA isolated from blood plasma and cell-surface-bound cirDNA was elevated 2- to 3-fold in lung cancer patients compared with healthy donors. Random forest classification tree model based on these variables combined (RARB2 and RASSF1A IM in both plasma and cell-surface-bound cirDNA) lead to NSCLC patients’ and healthy subjects’ differentiation with 87% sensitivity and 75% specificity. An association of increased IM values with an advanced stage of non-small-cell lung cancer was found for RARB2 but not for RASSF1A. Chemotherapy and total tumor resection resulted in a significant decrease in the IM for RARB2 and RASSF1A, in both cirDNA fractions, comparable to the IM level of healthy subjects. Importantly, a rise in the IM for RARB2 was detected in patients within the follow-up period, which manifested in disease relapse at 9 months, confirmed with instrumental and pathologic methods. Our data indicate that quantitative analysis of the methylation status of the RARB2 and RASSF1A tumor suppressor genes in both cirDNA fractions is a useful tool for lung cancer diagnostics, evaluation of cancer treatment efficiency and post-treatment monitoring.

Climatic niche, ecological genetics, and impact of climate change on eastern white pine (Pinus strobus L.): Guidelines for land managers

Approximately, 123,500 forest inventory and ecological ground plots representing eastern North America were used to predict the contemporary distribution of eastern white pine (Pinus strobus L.) from climate. The random forests classification tree procedures produced an 8-variable algorithm that had an 8% overall error rate. Erroneous predictions of presence, or errors of commission, were 13%, while falsely predicting absence, or errors of omission were 1%. Climate-based multiple regression models were developed to describe patterns of genetic variation among 112 populations representing the range of P. strobus in Ontario, Canada east of Lake Superior. Degree days >5°C was the best predictor of variation in phenology and growth potential, with 5-year height providing the greatest resolution of inter-population variation (R2=0.68). Cold hardiness in the fall was most closely associated with mean minimum temperature (R2=0.26). Height growth data from four disparate provenance test series that together included a total of 354 provenances corroborated the range-wide applicability of the regional genetic models. Although variation in growth potential in the central Appalachian Mountains was most closely associated with mean minimum temperature, degree days >5°C remained the best predictor of range-wide variation in growth potential (R2=0.41). The contemporary distribution and inter-population genetic variation were projected into future climates predicted by three General Circulation Models, two scenarios, and three time steps. All projections indicate early and sustained deterioration in the contemporary habitat. Concurrence among projections regarding the redistribution of suitable habitat to the north of the contemporary distribution identifies geographic locations with the highest probability of supporting vigorous stands of P. strobus. Concurrences among genetic projections clarify the intraspecific redistribution required to conserve adaptive variation. The projections have direct relevance in developing management strategies for accommodating the changing climate.

Abies religiosa habitat prediction in climatic change scenarios and implications for monarch butterfly conservation in Mexico

Abies religiosa (HBK) Schl. & Cham. (oyamel fir) is distributed in conifer-dominated mountain forests at high altitudes along the Trans-Mexican Volcanic Belt. This fir is the preferred host for overwintering monarch butterfly (Danaus plexippus) migratory populations which habitually congregate within a few stands now located inside a Monarch Butterfly Biosphere Reserve. Our objectives were to predict and map the climatic niche for A. religiosa for contemporary and future (2030, 2060 and 2090) climates, suggest management strategies to accommodate climate changes, and discuss implications for conservation of monarch butterfly overwintering sites in Mexico. A bioclimate model predicting the presence or absence of A. religiosa was developed by using the Random Forests classification tree on forest inventory data. The model used six predictor variables and was driven primarily by the mean temperature of the warmest month, an interaction between summer precipitation to and winter temperatures, and the ratio of summer to annual precipitation. Projecting the contemporary climate niche into future climates provided by three General Circulation Models and two scenarios suggested that the area occupied by the niche should diminish rapidly over the course of the century: a decrease of 69.2% by the decade surrounding 2030, 87.6% for that surrounding 2060, and 96.5% for 2090. We discuss assisted migration of A. religiosa upwards in altitude by 275m so that populations of 2030 would occupy the same climates as today. The projections also show that by the end of the century, suitable habitat for the monarch butterfly may no longer occur inside the Biosphere Reserve. We therefore discuss management options and associated research programs necessary for assuring perpetuation of future butterfly habitat.

North American vegetation model for land‐use planning in a changing climate: a solution to large classification problems

Data points intensively sampling 46 North American biomes were used to predict the geographic distribution of biomes from climate variables using the Random Forests classification tree. Techniques were incorporated to accommodate a large number of classes and to predict the future occurrence of climates beyond the contemporary climatic range of the biomes. Errors of prediction from the statistical model averaged 3.7%, but for individual biomes, ranged from 0% to 21.5%. In validating the ability of the model to identify climates without analogs, 78% of 1528 locations outside North America and 81% of land area of the Caribbean Islands were predicted to have no analogs among the 46 biomes. Biome climates were projected into the future according to low and high greenhouse gas emission scenarios of three General Circulation Models for three periods, the decades surrounding 2030, 2060, and 2090. Prominent in the projections were (1) expansion of climates suitable for the tropical dry deciduous forests of Mexico, (2) expansion of climates typifying desertscrub biomes of western USA and northern Mexico, (3) stability of climates typifying the evergreen-deciduous forests of eastern USA, and (4) northward expansion of climates suited to temperate forests, Great Plains grasslands, and montane forests to the detriment of taiga and tundra climates. Maps indicating either poor agreement among projections or climates without contemporary analogs identify geographic areas where land management programs would be most equivocal. Concentrating efforts and resources where projections are more certain can assure land managers a greater likelihood of success.

Integrating stress‐related ventricular functional and angiographic data in preventive cardiology: a unified approach implementing a Bayesian network

Identification of key factors associated with the risk of adverse cardiovascular events and quantification of this risk using multivariable prediction algorithms are among the major advances made in preventive cardiology and cardiovascular epidemiology. In the present paper, we examined clinical predictors of adverse cardiovascular events among 228 individuals with symptoms suggestive of coronary artery disease (CAD) undergoing functional (stress echocardiography) and anatomical (coronary angiography) assessment of CAD. Particularly, we evaluate the possibility to integrate simple measures that have known prognostic value and more recently discovered predictors of risk, such as stress-related ventricular function data and angiographic data, in a unique model implementing a Bayesian network (BN). Moreover, we compared the performance of BN and the covariates hierarchy with those obtained from logistic regression model and from a set of alternative tools becoming popular in various clinical settings, including random forest classification tree analysis, artificial neural networks and support vector machine. Network graph and results coming from sensitivity analysis, where variables are ranked according to the gain they provided in variance reduction, seem have an easily intuitive lecture: variables that are measure of ventricular disfunction or of the extent of CAD show a greater impact in predicting event. On the other hand, anamnestic data such as diabetes, dyslipidaemia, hypertension, smoke habits, which are related to the outcome throughout a process of intermediate variables, per se have a small role in outcome prediction. BNs are able to explain a relevant part of variance (70%) and have discrimination ability superior or comparable with those to random forest classification tree analysis, artificial neural networks and support vector machine. Despite the complexity of interactions, model obtained implementing a BN seems to be able to adequately describe the relationships existing among the analysed variables. BN, being able to predict scenarios in which new variables can be incorporated as health process evolves, can measure individual's risks for adverse cardiovascular events, providing a permanent second opinion to the medical practitioner and assisting diagnostic and therapeutic process.

Aspen, climate, and sudden decline in western USA

A bioclimate model predicting the presence or absence of aspen, Populus tremuloides, in western USA from climate variables was developed by using the Random Forests classification tree on Forest Inventory data from about 118,000 permanent sample plots. A reasonably parsimonious model used eight predictors to describe aspen's climate profile. Classification errors averaged 4.5%, most of which were errors of commission. The model was driven primarily by three variables: an annual dryness index, the ratio of summer to annual precipitation, and an interaction of growing season precipitation with the summer–winter temperature differential. Projecting the contemporary climate profile into the future climate provided by three General Circulation Models and two scenarios (SRES A2 and either B1 or B2) suggested that the area occupied by the profile should diminish rapidly over the course of the century, 6–41% by the decade surrounding 2030, 40–75% for that surrounding 2060, and 46–94% for 2090. The relevance of the climate profile to understanding climate-based responses is illustrated by relating trends in climate to the recent incidence of sudden aspen dieback that has plagued portions of the aspen distribution. Of the eight variables in the profile, four reached extreme values during 2000–2003, the period immediately preceding the appearance of damage in aerial surveys.

QUANTIFYING THE ABUNDANCE OF CO-OCCURRING CONIFERS ALONG INLAND NORTHWEST (USA) CLIMATE GRADIENTS

The occurrence and abundance of conifers along climate gradients in the Inland Northwest (USA) was assessed using data from 5082 field plots, 81% of which were forested. Analyses using the Random Forests classification tree revealed that the sequential distribution of species along an altitudinal gradient could be predicted with reasonable accuracy from a single climate variable, a growing-season dryness index, calculated from the ratio of degree-days >5 degrees C that accumulate in the frost-free season to the summer precipitation. While the appearance and departure of species in an ascending altitudinal sequence were closely related to the dryness index, the departure was most easily visualized in relation to negative degree-days (degree-days < 0 degrees C). The results were in close agreement with the works of descriptive ecologists. A Weibull response function was used to predict from climate variables the abundance and occurrence probabilities of each species, using binned data. The fit of the models was excellent, generally accounting for >90% of the variance among 100 classes.