Subregional Classification Research Articles

BreasTDLUSeg: A coarse-to-fine framework for segmentation of breast terminal duct lobular units on histopathological whole-slide images

Automatic segmentation of breast terminal duct lobular units (TDLUs) on histopathological whole-slide images (WSIs) is crucial for the quantitative evaluation of TDLUs in the diagnostic and prognostic analysis of breast cancer. However, TDLU segmentation remains a great challenge due to its highly heterogeneous sizes, structures, and morphologies as well as the small areas on WSIs. In this study, we propose BreasTDLUSeg, an efficient coarse-to-fine two-stage framework based on multi-scale attention to achieve localization and precise segmentation of TDLUs on hematoxylin and eosin (H&E)-stained WSIs. BreasTDLUSeg consists of two networks: a superpatch-based patch-level classification network (SPPC-Net) and a patch-based pixel-level segmentation network (PPS-Net). SPPC-Net takes a superpatch as input and adopts a sub-region classification head to classify each patch within the superpatch as TDLU positive or negative. PPS-Net takes the TDLU positive patches derived from SPPC-Net as input. PPS-Net deploys a multi-scale CNN-Transformer as an encoder to learn enhanced multi-scale morphological representations and an upsampler to generate pixel-wise segmentation masks for the TDLU positive patches. We also constructed two breast cancer TDLU datasets containing a total of 530 superpatch images with patch-level annotations and 2,322 patch images with pixel-level annotations to enable the development of TDLU segmentation methods. Experiments on the two datasets demonstrate that BreasTDLUSeg outperforms other state-of-the-art methods with the highest Dice similarity coefficients of 79.97% and 92.93%, respectively. The proposed method shows great potential to assist pathologists in the pathological analysis of breast cancer. An open-source implementation of our approach can be found at https://github.com/Dian-kai/BreasTDLUSeg.

The use of ecological niche models improves biogeographic regionalization of the Balsas Depression, Mexico

AbstractAimBiogeographic regionalization classifies zones in terms of their biotas and contributes to understanding the ecological and historical factors that affect the distribution of species. We use Ecological Niche Modeling (ENM) to complement missing information on species distribution and thus improve the accuracy of biogeographic boundaries.LocationBalsas Depression Floristic Province, Mexico.MethodsBased on parameters documented in herbarium collections and environmental variables, ENM was carried out to determine the most suitable environmental conditions for a species to thrive (i.e., the species' ecological niche). The ENM and spatial analysis were used to obtain the biogeographic regionalization of the seasonally dry tropical forest (SDTF) in the Balsas Depression (BD), Mexico, through spatial analysis. Using the Maxent algorithm, we constructed ecological niche models (ENMs) of 134 flowering plant species distributed preferentially in the SDTF (characteristic species), most of them endemic to the BD. Subsequently, we obtained an incidence matrix based on the information from the 134 ENMs, which was used to analyze the turnover of species in Biodiverse software. The turnover matrix was used for Non‐metric Multidimensional Scaling (NMDS) ordination and clustering analyses. Finally, the environmental predictors most related to species turnover were identified using the relative environmental turnover method.ResultsThe clustering analysis divided the SDTF in the BD into four floristic districts — two located in its western part and two in the eastern region. The NMDS differentiated, in the first component, two districts in the western region and one in the eastern. Seven environmental variables contributed significantly to explaining the turnover of species in these districts; the most significant were the elevation, pH, and precipitation of the coldest quarter.Main ConclusionsThe use of ENM for the regionalization of areas with high species richness allows for a more detailed classification of subregions and the distribution patterns of the species that define their limits. This provides a more solid theoretical basis for the investigation of biogeographic patterns.

Machine Learning Model Stability for Sub-Regional Classification of Barossa Valley Shiraz Wine Using A-TEEM Spectroscopy.

With a view to maintaining the reputation of wine-producing regions among consumers, minimising economic losses caused by wine fraud, and achieving the purpose of data-driven terroir classification, the use of an absorbance-transmission and fluorescence excitation-emission matrix (A-TEEM) technique has shown great potential based on the molecular fingerprinting of a sample. The effects of changes in wine composition due to ageing and the stability of A-TEEM models over time had not been addressed, however, and the classification of wine blends required investigation. Thus, A-TEEM data were combined with an extreme gradient boosting discriminant analysis (XGBDA) algorithm to build classification models based on a range of Shiraz research wines (n = 217) from five Barossa Valley sub-regions over four vintages that had aged in bottle for several years. This spectral fingerprinting and machine learning approach revealed a 100% class prediction accuracy based on cross-validation (CV) model results for vintage year and 98.8% for unknown sample prediction accuracy when splitting the wine samples into training and test sets to obtain the classification models. The modelling and prediction of sub-regional production area showed a class CV prediction accuracy of 99.5% and an unknown sample prediction accuracy of 93.8% when modelling with the split dataset. Inputting a sub-set of the current A-TEEM data into the models generated previously for these Barossa sub-region wines yielded a 100% accurate prediction of vintage year for 2018-2020 wines, 92% accuracy for sub-region for 2018 wines, and 91% accuracy for sub-region using 2021 wine spectral data that were not included in the original modelling. Satisfactory results were also obtained from the modelling and prediction of blended samples for the vintages and sub-regions, which is of significance when considering the practice of wine blending.

A STUDY ON THE SPATIAL DIFFERENTIATION OF THE SOCIO-ECONOMIC AND TOURIST SITUATION OF SELECTED MUNICIPALITIES OF THE WEST POMERANIAN VOIVODESHIP

Aim: The aim of the article was to develop an original classification of municipalities in the West Pomeranian Voivodeship, Poland, with particular emphasis on measuring local development, taking into account the de­velopment of health tourism. Methods: The research procedure consisted of the following stages: collection of data and assessment of their completeness for the West Pomeranian region, calculating the values of the indicators, constructing rankings of municipalities in a one-dimensional space of characteristics-indicators and distinguishing typological groups of municipalities similar in terms of the degree of development of the socio-economic and tourist situation. The classification of sub-regions was carried out using discriminant analysis. For this purpose, the indicators characterizing the socio-economic and tourist situation in the selected West Pomeranian Voivodeship municipalities were used, taking into account the division into municipalities: urban (1), rural (2), and urban-rural (3). The preliminary classification of the objects into groups, and thus the selection of the grouping variable, was carried out using the k-means method. The study used secondary data from the Local Data Base – GUS. Results: As a result of the survey, the analyzed local government units were divided into three groups. However, the presented results of the classification do not indicate the dependence of the improvement of socio-economic characteristics on the provision of tourism services in the municipality (including health tourism). Conclusions: Empirical studies show that the placement of munici­palities in the West Pomeranian Voivodeship reveals clear relationships between the territorial distribution of health tourism units and the territorial distribution of natural assets and access to the Baltic Sea.

Mapping of bamboo forest bright and shadow areas using optical and SAR satellite data in Google Earth Engine

Bamboo groves predominantly thrive in tropical or subtropical regions. Assessing the efficacy of remote sensing data of various types in extracting bamboo forest information from bright and shadow areas is a critical issue for achieving precise identification of bamboo forests in complex terrain. In this study, 34 features were obtained from Sentinel-1 SAR and Sentinel-2 optical images using the Google Earth Engine platform. The normalized shaded vegetation index (NSVI) was then employed to segment the bright and shadow woodlands. Different features from diverse data sources were evaluated to extract bamboo forest information in the bright and shadow areas, then use the random forest (RF) classification algorithm to extract bamboo forest. The results showed that (1) the red-edge and short-wave infrared bands of Sentinel-2 optical images and their corresponding vegetation indices are significant in bamboo forest information extraction. (2) The dissimilarity and homogeneity of Sentinel-2 texture features in the bright area and dissimilarity in the shadow area, the Sentinel-1 backscatter features in the bright area and the VV and VH in the bright area and VV-VH in the shadow area have some variability between bamboo and nonbamboo forests, which can be used as effective features for bamboo forest extraction. (3) The combination of spectral, texture and backscatter features yields the highest overall classification accuracy and Kappa coefficient, at 87.96% and 0.7435, respectively. This study has the potential for remote sensing refinement of bamboo forest identification in complex terrain areas by utilizing subregion classification methods combined with optical and radar image features.

Combining Human Parsing with Analytical Feature Extraction and Ranking Schemes for High-Generalization Person Reidentification

Person reidentification (re-ID) has been receiving increasing attention in recent years due to its importance for both science and society. Machine learning (particularly Deep Learning (DL)) has become the main re-ID tool that has allowed to achieve unprecedented accuracy levels on benchmark datasets. However, there is a known problem of poor generalization in respect of DL models. That is, models that are trained to achieve high accuracy on one dataset perform poorly on other ones and require re-training. In order to address this issue, we present a model without trainable parameters. This, in turn, results in a great potential for high generalization. This approach combines a fully analytical feature extraction and similarity ranking scheme with DL-based human parsing wherein human parsing is used to obtain the initial subregion classification. We show that such combination, to a high extent, eliminates the drawbacks of existing analytical methods. In addition, we use interpretable color and texture features that have human-readable similarity measures associated with them. In order to verify the proposed method we conduct experiments on Market1501 and CUHK03 datasets, thus achieving a competitive rank-1 accuracy comparable with that of DL models. Most importantly, we show that our method achieves 63.9% and 93.5% rank-1 cross-domain accuracy when applied to transfer learning tasks, while also being completely re-ID dataset agnostic. We also achieve a cross-domain mean average precision (mAP) that is higher than that of DL models in some experiments. Finally, we discuss the potential ways of adding new features to further improve the model. We also show the advantages of interpretable features for the purposes of constructing human-generated queries from verbal descriptions in order to conduct searches without a query image.

Impact of the COVID-19 Pandemic on International Tourism Income in Tourism Receiving Countries

The purpose of this article is to analyze the impact of the COVID-19 pandemic on international tourism income in tourism reception countries, in which the greatest tourist traffic in the world is observed. The analysis was performed on the basis of data obtained from the UNWTO, which was used to create a single-feature classification of subregions (using Hellwig’s method) and to create a single-feature classification (using a method based on positional measures) of the most important reception countries, i.e., generating the highest tourist traffic. Based on the research, it was found that the critical range of Hellwig occurred between Northeast Asia and Southeast Asia, North America, the Middle East and Sub-Saharan Africa, and between the Caribbean and South Asia. The largest percentage drops in revenues from international tourism in 2020 in comparison to 2019 occurred in Hong Kong, Malaysia, Greece and Spain. These countries recorded values above the upper quartile, which was −76.23%.

Enabling access to quality abortion care: WHO's Abortion Care guideline

Enabling access to quality abortion care: WHO's Abortion Care guideline

Putting Together Pieces of the Lateral Septum: Multifaceted Functions and Its Neural Pathways.

The lateral septum (LS) is implicated as a hub that regulates a variety of affects, such as reward, feeding, anxiety, fear, sociability, and memory. However, it remains unclear how the LS, previously treated as a structure of homogeneity, exhibits such multifaceted functions. Emerging evidence suggests that different functions of the LS are mediated largely by its diverse input and output connections. It has also become clear that the LS is a heterogeneous region, where its dorsal and ventral poles play dissociable and often opposing roles. This functional heterogeneity can often be explained by distinct dorsal and ventral hippocampal inputs along the LS dorsoventral axis, as well as antagonizing connections between LS subregions. Similarly, outputs from LS subregions to respective downstream targets, such as hypothalamic, preoptic, and tegmental areas, also account for this functional heterogeneity. In this review, we provide an updated perspective on LS subregion classification, connectivity, and functions. We also identify key questions that have yet to be addressed in the field.

A Multiparametric MRI-Based Radiomics Analysis to Efficiently Classify Tumor Subregions of Glioblastoma: A Pilot Study in Machine Learning

Glioblastoma multiforme (GBM) carries a poor prognosis and usually presents with heterogenous regions of a necrotic core, solid part, peritumoral tissue, and peritumoral edema. Accurate demarcation on magnetic resonance imaging (MRI) between the active tumor region and perifocal edematous extension is essential for planning stereotactic biopsy, GBM resection, and radiotherapy. We established a set of radiomics features to efficiently classify patients with GBM and retrieved cerebral multiparametric MRI, including contrast-enhanced T1-weighted (T1-CE), T2-weighted, T2-weighted fluid-attenuated inversion recovery, and apparent diffusion coefficient images from local patients with GBM. A total of 1316 features on the raw MR images were selected for each annotated area. A leave-one-out cross-validation was performed on the whole dataset, the different machine learning and deep learning techniques tested; random forest achieved the best performance (average accuracy: 93.6% necrosis, 90.4% solid part, 95.8% peritumoral tissue, and 90.4% peritumoral edema). The features from the enhancing tumor and the tumor shape elongation of peritumoral edema region for high-risk groups from T1-CE. The multiparametric MRI-based radiomics model showed the efficient classification of tumor subregions of GBM and suggests that prognostic radiomic features from a routine MRI exam may also be significantly associated with key biological processes that affect the response to chemotherapy in GBM.

Breast Cancer Tissue Sub-Region Classification from Second Harmonic Generation Imagery via Machine Learning

Cancer treatment involves complex decision-making processes. A better understanding of recurrence risk at diagnosis, as well as prediction of treatment response (e.g., to choose the most cost-effective treatment path in an informed manner) are needed to produce the best possible outcomes at the patient level. Prior work shows that the forward/backward (F/B) ratio calculated from Second Harmonic Generation (SHG) imagery can be indicative of risk of recurrence if relevant tissue sub-regions are selected. The choice of which sub-regions to image is currently made by human experts, which is subjective and labor intensive. In this paper, we investigate machine learning methods to automatically identify tissue sub-regions that are most relevant to the prediction of breast cancer recurrence. We formulate the task as a multi-class classification problem and use support vector machine (SVM) classifiers as the inference engine. Given the limited amount of data available, we focus on exploring the feature extraction stage. To that end, we evaluate methods leveraging handcrafted features, deep features extracted from pre-trained models, as well as features extracted via transfer learning. The results show a steady trend of improvement on the classification accuracy as the features become more data-driven and customized to the task at hand. This is an indication that having larger amounts of labeled data could be beneficial for improving automated methods of classification. The best results achieved, using features learned via transfer learning from ResNet-101, correspond to 85% accuracy in a 3-class problem and 94% accuracy in a binary classification problem.

Woody plant subregions of the Amazon forest

Abstract The Amazon forest covers 7.5 million km2 in nine countries, hosts 25% of the global biodiversity and is a major contributor to the biogeochemical and climatic functioning of the Earth system. Despite its global importance, a regionalization of the Amazon tree flora is still lacking. Clear and data‐driven delimitation of subregions is important for macroecological studies, to the identification of metacommunities and is a requisite for conservation planning. We aimed at identifying and mapping plant species subregions and investigated their relationships with environmental, historical and human correlates. We provide the first woody plant regionalization of the entire Amazon forest using a data‐driven approach based on assemblage composition patterns. We compiled data on woody species composition from 301 assemblages based on species occurrences. We then used unconstrained ordination, interpolation and clustering techniques to identify and map discrete woody subregions. Hierarchical clustering analysis was conducted in order to investigate the relationships between the identified subregions. We used multinomial logistic regression model and deviance partitioning to investigate the influence of environmental, historical and human factors on subregions distribution. We identified 13 woody subregions in the entire Amazon forest. The hierarchical subregion classification showed a broad Andean‐Cratonic east–west division. Variation in subregions was explained jointly by human factors and spatial structure followed by environmental factors and spatial structure combined. Synthesis. Our woody plant subregions differed from World Wildlife Fund ecoregions and physiognomic‐based maps, highlighting the importance of basing regionalizations on taxon‐specific groups and confirming that vegetation maps should not be used as proxies to plant diversity subregions. Our findings also confirm the need for multiple and extensive protected areas in the Amazon forest. The relevance of current climate factors in our study alerts to a profound impact that climate change could have on the spatial organization of the Amazon flora.

Towards the Increase in Granularity for the Main Hadronic ATLAS Calorimeter: Exploiting Deep Learning Methods

During the second phase upgrade program developed for LHC and its experiments, the main hadronic calorimeter of ATLAS (TileCal) will replace completely its readout electronics, but the optical signal pathway and detector will be kept unchanged. During the R&D studies for the upgrade, initial analyses for improving the calorimeter granularity were made. A granularity improvement could be achieved through the introduction of Multi-Anode Photomultiplier Tubes (MA-PMTs) into the calorimeter readout chain, together with applications of image processing algorithms for identifying sub-regions on calorimeter cells. This paper presents the latest results from using a Generative Adversarial Network (GAN) to generate synthetic images, which simulate real images formed in the MA-PMT. After the classification of cell sub-regions, preliminary results show a classification accuracy of more than 98% on the experimental test set.

Quantum and tempo effects of changes during the demographic transition: classification of world sub-regions and selected countries

The development patterns of both historical and contemporary demographic transitions are essential in understanding the outcomes of age structure changes among the countries. Existing literature analyzing demographic transition as a precursor to population ageing does not classify countries or regions worldwide by the scale and dynamics of those processes. The aim of this paper is to identify and describe the differentiation of quantum and tempo effects of changes taking place during the demographic transition through classification of sub-regions and selected countries worldwide. The results of this research, based on the historical vital statistics starting from as far as 1736 and future projections until 2100, feature how sub-regions and selected countries in the world are differentiated in terms of the scale and dynamics of the demographic transition process. The analysis reveals the fact that majority of sub-regions in the world tend(ed) to undergo long transitions lasting over 70 years, though contemporary demographic transitions demonstrate higher intensity and higher maximal rates of natural increase. African sub-regions are expected to experience relatively long processes of declining birth and death rates and estimated to complete their demographic transition in 90 years on average. The fastest processes, however, are observed in South America and Eastern Asia where some countries like China, for instance, completed their transitions in just 50 years. The scale and dynamics of demographic transition processes can serve as the ground for further research of challenges and development opportunities resulting from ageing societies during consequent post-transitional stages.

Classification of border subregions of Eastern Poland according to theIR level of socio-economic development

The paper presents the classification of subregions (NUTS 3) with permanent border crossings, in terms of the level of socio-economic development in 2008-2018. The study covered 9 subregions located in the following provinces: Warminsko-Mazurskie, Podlaskie, Lubelskie and Podkarpackie. Application of Hellwig's measure of development made it possible to group subregions according to three categories of the level of development: high, medium and low. The relatively high level of development characterized two metropolitan subregions: Olsztyński and Białostocki. Łomżyński subregion proved to be the weakest, but it was not completely devoid of any development potential. In comparison with the average values in the designated higher development classes, Łomżyński subregion was characterized by: positive natural increase, low values of indices describing ageing of local society and proportion of investment expenditure of local self-government units at a level similar to the leaders of the ranking. However, there is a concern that the relatively low level of spending of the EU funds and the persistent trend of population migration outflow, due to global trends of population ageing and depopulation of rural areas, may negatively affect the socio-economic situation of the subregion in the future.

A New Agenda to Study the Guatemala–Mexico/Mexico–Guatemala Border(Lands) Region

ABSTRACT This paper introduces a proposal to reconsider the way the Mexico–Guatemala border(lands) region is conceived. The border(lands) expanse is usually portrayed as a uniform entity. Generalizations are made from a single locality, thus obscuring the diversity of the area. Also, the Mexican viewpoint has prevailed in the reading of this border(lands) region, instead of adopting a compound perspective that takes into account the Guatemalan angle. These and other issues are discussed at length along with ideas to redress existing shortcomings. The use of a regional approach is advanced as an analytical tool and a planning policy instrument to overcome this problematic. A sub-regional classification will reveal, for instance, the heterogeneous conditions that prevail across the international boundary. This typology is meant as a starting point to plan and implement further collaborative research agendas and integrated policy initiatives that respond to the idiosyncrasy of the border(lands) region.

Clustering of multi-parametric functional imaging to identify high-risk subvolumes in non-small cell lung cancer

Background and purposeWe aimed to identify tumour subregions with characteristic phenotypes based on pre-treatment multi-parametric functional imaging and correlate these subregions to treatment outcome. The subregions were created using imaging of metabolic activity (FDG-PET/CT), hypoxia (HX4-PET/CT) and tumour vasculature (DCE-CT). Materials and methods36 non-small cell lung cancer (NSCLC) patients underwent functional imaging prior to radical radiotherapy. Kinetic analysis was performed on DCE-CT scans to acquire blood flow (BF) and volume (BV) maps. HX4-PET/CT and DCE-CT scans were non-rigidly co-registered to the planning FDG-PET/CT. Two clustering steps were performed on multi-parametric images: first to segment each tumour into homogeneous subregions (i.e. supervoxels) and second to group the supervoxels of all tumours into phenotypic clusters. Patients were split based on the absolute or relative volume of supervoxels in each cluster; overall survival was compared using a log-rank test. ResultsUnsupervised clustering of supervoxels yielded four independent clusters. One cluster (high hypoxia, high FDG, intermediate BF/BV) related to a high-risk tumour type: patients assigned to this cluster had significantly worse survival compared to patients not in this cluster (p = 0.035). ConclusionsWe designed a subregional analysis for multi-parametric imaging in NSCLC, and showed the potential of subregion classification as a biomarker for prognosis. This methodology allows for a comprehensive data-driven analysis of multi-parametric functional images.

Global, regional, and subregional classification of abortions by safety, 2010–14: estimates from a Bayesian hierarchical model

SummaryBackgroundGlobal estimates of unsafe abortions have been produced for 1995, 2003, and 2008. However, reconceptualisation of the framework and methods for estimating abortion safety is needed owing to the increased availability of simple methods for safe abortion (eg, medical abortion), the increasingly widespread use of misoprostol outside formal health systems in contexts where abortion is legally restricted, and the need to account for the multiple factors that affect abortion safety.MethodsWe used all available empirical data on abortion methods, providers, and settings, and factors affecting safety as covariates within a Bayesian hierarchical model to estimate the global, regional, and subregional distributions of abortion by safety categories. We used a three-tiered categorisation based on the WHO definition of unsafe abortion and WHO guidelines on safe abortion to categorise abortions as safe or unsafe and to further divide unsafe abortions into two categories of less safe and least safe.FindingsOf the 55· 7 million abortions that occurred worldwide each year between 2010–14, we estimated that 30·6 million (54·9%, 90% uncertainty interval 49·9–59·4) were safe, 17·1 million (30·7%, 25·5–35·6) were less safe, and 8·0 million (14·4%, 11·5–18·1) were least safe. Thus, 25·1 million (45·1%, 40·6–50·1) abortions each year between 2010 and 2014 were unsafe, with 24·3 million (97%) of these in developing countries. The proportion of unsafe abortions was significantly higher in developing countries than developed countries (49·5% vs 12·5%). When grouped by the legal status of abortion, the proportion of unsafe abortions was significantly higher in countries with highly restrictive abortion laws than in those with less restrictive laws.InterpretationIncreased efforts are needed, especially in developing countries, to ensure access to safe abortion. The paucity of empirical data is a limitation of these findings. Improved in-country data for health services and innovative research to address these gaps are needed to improve future estimates.FundingUNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction; David and Lucile Packard Foundation; UK Aid from the UK Government; Dutch Ministry of Foreign Affairs; Norwegian Agency for Development Cooperation.

Differentiation of Polish subregions due to the risk of poverty

This article aims to show the diversity of the development of sub-regions in Poland in view of the selected features and to characterize the areas with the highest risk of poverty. The Polish sub-regions were described using data for 2013, depicting economic, social and housing conditions of the Polish population as well as its economic activity. These data were standardized and — using discriminant function — sub-regions were grouped in 7 areas. The areas I and II, based on the value of the discriminant function, were classified as areas of the risk of poverty and social exclusion. Then, for each area classification functions were estimated. Each sub-region was assigned to the area for which it has the greatest classification value. It was found that the greatest impact on the final classification of sub-regions have following indicators: the unemployment rate, population per 1 employee and the average size of the apartment.

Spatial climate-dependent growth response of boreal mixedwood forest in western Canada

The western Canadian mixedwood boreal forests were projected to be significantly affected by regional drought. However, drought degrees were spatially different across elevations, longitudes and latitudes, which might cause different tree growth responses to climate change in different sub-regions within western Canada. In this way, regional classification of western Canadian boreal forests and understanding spatial tree growth responses to climate might be necessary for future forest management and monitoring. In this paper, tree-ring chronologies of two dominant tree species, trembling aspen (Populus tremuloides Michx.) and white spruce (Picea glauca (Moench.) Voss), were obtained from mixed forest stands distributed across western Canada to study spatial tree growth response to climate based on three regional classification schemes (a phytogeographic sub-region classification, a natural sub-region classification and non-classification). Phytogeographic sub-region classification was estimated based on tree ring samples we collected in this study, while natural sub-region classification was previously developed based on analysis of regional differences in vegetation, soil, site and climate conditions. Results showed that air temperature did not significantly increase, while drought stress became more severe between 1985 to 2010. Relationships between trembling aspen growth and temperature differed between north and south parts of the study area, resulting from spatial difference in water supply. Trembling aspen growth was influenced by temperature or moisture variables of the previous years. White spruce growth was influenced primarily by moisture variables (current or previous year), and response coefficients between white spruce and drought conditions (represented by drought code) were negative in all phytogeographic sub-regions, suggesting that white spruce was more sensitive to drought stress under climate change. As a late-successional dominant species, increasingly drought stress on white spruce might cause significant alteration in forest composition of western Canadian boreal forest.