Annotation of the results obtained in 2022
A database of 760 geobotanical releves and 147 syntaxa of the rank of associations and subassociations made in the Braun-Blanquet system for the territory of the Republic of Khakassia based on the European Turboveg package has been developed. A generalization of the vegetation classification of Khakassia has been carried out and a geographic information system (GIS) "Vegetation of Khakassia" was formed. It linked into a single information space the database of primary geobotanical data, cartographic materials on the vegetation of the region, as well as additional layers of ecological and geographical information and a series of satellite images of different spatial resolution (Resource – P, World View - 2,3, Sentinel – 2В, Kanopus-В, Landsat - 7, 8, MODIS). GIS was implemented on the basis of ArcGis-10.0 and QGIS packages. New data on the ecological regularities of the spatial differentiation of the vegetation of Khakassia at the topological and regional levels were obtained using modern methods of ordination modeling. The completed set of ordination studies using the Detrended Correspondence Analysis (DCA ordination) method, implemented in the applied packages Decorana, Canoco, made it possible to identify the basic spatial categories of mapping (phyto-coenochories) - ecological series and micro-meso-combinations of plant communities representing the combination of phytocoenoses arranged along gradients of the leading environmental factors and associated with specific geomorphological (and topographic) conditions. The results of ordination were used for the large-scale vegetation mapping at a scale of 1:25,000 on four key polygons representing mainly the primary vegetation of the steppe and mountain-taiga belts of the Minusinskaya intermountain basin, the eastern part of the Kuznetskiy Alatau and the Western Sayan. An ordination model of the mountain-steppe vegetation of the entire territory of Khakassia has been created. An ecological and geographical series of steppe vegetation types oriented from the more humid forest-steppe foothills of the Western Sayan and Kuznetskiy Alatau to the semi-arid central part of the Minusinskaya intermountain basin was revealed. This corresponds to the observed global sub-longitudinal zonal-sectoral and sub-zonal replacement of steppe vegetation categories in the mountains of Southern Siberia according to the climatic effects of the "rain barrier" and "rain shadow". The bioclimatic interpretation of large ecological and geographical categories of steppe vegetation of Khakassia is given and the possibility of using them at the highest levels of the hierarchy of the Khakassia vegetation map legend is substantiated. The differentiation of ecological types of steppes in relation to the type of the substrate of the bedrocks was revealed. The results of ordination demonstrated the presence of ecologically corresponding petrophytic series of steppe communities equally oriented within each of the bioclimatic categories. Based on the results of the ordination analysis, a hierarchy of units of mountain-steppe vegetation was constructed for the medium-scale vegetation map of Khakassia. Based on the results of vegetation classification, new data on its ecological-geographical and spatial regularities of formation as well as on the basis of interpretation of high-resolution satellite images, large-scale maps (1:25000) were compiled for the territories of four key polygons, which reflect the systems of phyto-coenomeres and phytocoenochores of topological dimension. Using the principles of V.B. Sochava made it possible to develop a hierarchical system of higher categories of the legend of the vegetation map of Khakassia. The content of the legend represents the results of the typification of the altitudinal and bioclimatic divisions of vegetation which make up the highest levels of the hierarchy of the spatial organization of the phyto-coenochories of the mountain system. For the higher subdivisions of the Vegetation Map of Khakassia, a four-level system of the spatial structure of the vegetation cover was proposed. According to the adopted methodology, the top first level of the hierarchy reflects the patterns of vertical zonality: alpine, forest, forest-steppe and steppe belts. The second and third levels of spatial organization of vegetation reflect a specific combination of regional and global climatic factors. The study of the post-fire dynamics of forest vegetation using satellite data resulted in the generation of time series of annual burned areas in the Republic of Khakassia. We have found that the annual burned areas in the region were characterized by significant variability associated with weather conditions. The assessment of the degree of pyrogenic disturbance of vegetation in the fire-disturbed areas was performed using the dNBR index (delta Normalized Burn Ratio). Since only a small part of fires in the region occur on forest lands, the mean proportion of severely disturbed areas (where dNBR was higher than 0.44) was about 0.4% of the mean annual burned area. Temporal analysis of the burned areas in the study region showed a pronounced seasonal dynamics of the burned area. For instance, April and May account for 65% of the entire burned area within the fire season. Regression analysis showed a statistically significant linear relationship between the degree of pyrogenic disturbance of vegetation, assessed using the dNBR index, and the proportion of the area where post-fire tree mortality was observed. For different dominant forest species the coefficient of determination ranged from 0.78 for pine forests to 0.91 in forests dominated by deciduous stands (p < 0.05). The locations of 12 endangered plant species of Khakassia were confirmed and established. The data obtained on the phytocenotic confinement of habitats of 33 species were summarized. Most of the rare and endangered plant species of Khakasia were found in petrophytic variants of typical grasslands and solonetzic steppes. Monitoring of coeno-populations for 3–5 years showed that left-sided ontogenetic spectra are formed in coeno-populations located on the southwestern and northeastern slopes and in hollows, as well as in some years when there is a sufficient amount of precipitation. Evaluation of coeno-populations according to the ontogenetic structure and demographic indicators revealed that most of the coenopopulations of Oxytropis stenofoliola, O. nuda, O. includens, Hedysarum minussinense, Phlox sibirica, Erodium tataricum, Papaver chakassicum, Nitraria sibirica, Lilium pumilum are in a stable state. On the contrary, in some coeno-populations of O. includens, Astragalus arkalycensis, A. ionae, Hedysarum minussinense, Erodium tataricum, Phlox sibirica, Nitraria sibirica and Spiraea trilobata, unfavorable conditions for the growth of individuals of the species were formed. The reason for this is the anthropogenic influence, eroded soil and changes in species diversity in plant communities. The software-algorithmic tools were developed and tested for the analysis of multispectral satellite images of medium and high spatial resolution (2 m and better). They are based on the original hierarchical ensemble methods for multispectral image segmentation and the of neural networks technology. The developed new data structure was implemented for hierarchical ensemble clustering algorithms: HCA, ECCA and HECA. These algorithms allow obtaining hierarchical representation of the segmentation results, which greatly simplifies image analysis and makes it possible to identify hierarchies of spatial units of vegetation cover. The HCA, ECCA and HECA algorithms were included in the developed software-algorithmic toolkit "ECCA-Pack 12", designed for automatic segmentation and interpretation of multispectral satellite images of medium and high spatial resolution.

 

Publications

---

Annotation of the results obtained in 2023
The primary data base "Vegetation of Khakassia", created on the basis of the Turboveg application package, has been replenished with 370 geobotanical releves and currently has 1300 primary documents. The composition of the geographical information system "Vegetation Cover of Khakassia" has been replenished with 9 new thematic layers: large-scale cartographic models created for key polygons, medium-scale (1:00000, 1:300000) maps of the forest and steppe vegetation belts of the Western Sayan mountain system and the northern part of the Minusinskaya intermountain basin, a climate model for the southern part of Khakassia, high-resolution satellite images Sentinel, Resource P. A system of classification of forest vegetation using the Braun-Blanquet method has been developed. All the phytocenotic diversity of the forests of Khakassia is attributed to 4 classes, 8 orders, 11 alliances and 50 associations. The classification results were used to create a bioclimatic model of higher categories of forest vegetation. The results of ordination modeling have demonstrated the importance of the leading complex gradients of environmental factors - climatic vertical zonality and oceanic-continentality of climate in the formation of zonal and sectoral series of the highest categories of forest cover of Khakassia. The hierarchy of the legend of the Vegetation Map of Khakassia has been developed using reasonable bioclimatic categories of forest vegetation. Large-scale cartographic models of the spatial organization of vegetation at the topological level have been created for a series of 5 key polygons placed in different altitude zones (forest-steppe, forest and high-mountain) of the main orographic structures of Khakassia. Cartographic models are based on ordination modeling (DCA-ordination) of geobotanical releves, ecological correlation analysis, and processing the high-resolution satellite images. Bioclimatic, ecological-topographical and digression-demutation vegetation series have been identified, which were used in the substantiation of spatial units of the rank of meso- and micro-combinations. As a result of the decoding of the Landsat-7,8 spectral satellite images, a medium-scale cartographic vegetation model was created at the scale of 1:100000 for a key polygon in the northern part of the Minusinskaya intermountain basin, as well as a vegetation map on the territory of the mountain taiga belt of the Western Sayan and the adjacent part of the Kuznetsk Alatau at the scale of 1:300000. In the medium-scale mapping of vegetation, the principles of V. B. Sochava were used and five levels of the legend hierarchy were determined. At the highest first level of the hierarchy of phytochores, the altitudinal structure of vegetation is represented. The second level shows the ecological-geographical and bioclimatic sectors formed under the conditions of fundamental interactions of relief and leading climatic processes (Western precipitation transferring and Asian anticyclone), leading to the effects of "rain barrier" and "rain shadow", which in turn cause gradients of important factors for the formation of the spatial structure of vegetation cover – humidity-aridity and oceanic-continentality of climate. The third hierarchical level of the legend is a combination of syntaxa (alliances-sub-alliances) that characterize the altitudinal vegetation belts within each bioclimatic sector. The fourth level contains different regional geographical combinations of vegetation association rank units within each of the sub-altitudinal belts. The fifth basic level represents ecologically determined combinations of specific plant communities, closely related to the features of mesorelief in the mountain system. Using remote sensing data, the start and end dates of the fire season were estimated for each year from 1981 to 2022. It was found that in mixed forest stands with a predominance of deciduous trees, as well as in larch and pine forests, the largest burned areas were observed in May. At the same time, in dark coniferous stands, significant burned areas were also observed during summer months (June, July). The largest burned areas were observed in non-forest areas. The longest fire season duration was also observed in non-forest areas (about 190 days). Forests were characterized by a shorter fire season (about 100–150 days). An increase in the duration of fire season in non-forest areas was also revealed. The start and end dates of the growing season were also assessed. On undisturbed forest lands, between 2001 and 2022, the beginning of the growing season shifted by approximately 4.4 days towards an earlier start. At the same time, in fire disturbed areas, the start date of the season shifted by approximately 5.9 days towards the beginning of the season. During the reporting period, a map of turning points was created – the years of the beginning of statistically significant trends (p < 0.05) in the dynamics of the vegetation index and a map of the corresponding coefficients of determination for 2000–2021; zones of significant positive and negative trends (p < 0.05) were identified for periods before and after turning points; trends in NDVI dynamics and their relationship with the topography were assessed; maps of multiple regressions were calculated to assess the fraction of contributions of climate variables to the explained variance of the dynamics of the vegetation cover of Khakassia. It has been revealed that in 2000–2021, for 35% of the territory of Khakassia, significant NDVI trends were observed in periods after turning points (TP) until 2021 and for 14% – in periods from 2000 to TP. Positive trends prevailed both before and after TP. Positive trends for the periods from the TP year to 2021 dominated mainly in the steppe zone. And it prevailed at elevations of 600–2200 m a.s.l. primarily on the western slopes, and negative trends were typical at elevations up to 1100 m a.s.l. (~75%) on the south-eastern slopes. The most fraction of the explained variance in vegetation dynamics corresponded to summer precipitation (median ~27%), soil moisture (~29%), and soil temperature (~27%). According to the results of the study of the reaction of the Tortuous birch (Betula tortuosa) in the ecotone of the mountain forest tundra of the Kuznetskiy Alatau to climatic changes, it was revealed that the rate of advance of the birch distribution boundary in the alpine zone was ~ 3.9 m/year. It is established that the beginning of the birch's advance into the mountain tundra zone is synchronized with the beginning of the warming period. A positive correlation was recorded between the radial growth of birch and the temperature of June (r = 0.63) during the period of active warming (1985-2005). The obtained data allow us to estimate the rate of advance of the upper boundary of woody vegetation in the ecotone of the mountain forest tundra of the Kuznetsk Alatau against the background of warming, which can be used to predict the response of vegetation to ongoing climatic changes. The locations of 10 valuable plant species in Khakasia, including relict, endemic and included in the Red Book of the Republic of Khakasia, were confirmed and established, 44 coenopopulations were studied. The habitats of the species in Khakasia are confined to large- and small-grass true steppes and their petrophytic variants, meadow steppes, steppe, salt marsh and true meadows, and birch herbaceous forests. 12 coenopopulations of the species Oxytropis includens, O. stenofoliola, O. reverdattoi, Viola selkirkii and Corydalis bracteata are small (the number of individuals in them does not exceed 15), unstable and require annual monitoring. The 32 studied coenopopulations of the species Oxytropis includens, O. stenofoliola, Astragalus arkalycensis, Oxytropis reverdattoi, Phlox sibirica, Limonium gmelinii, Viola selkirkii, Erythronium sibiricum and Anemone jenisseensis are in a stable state. Depending on the biological and ecological-phytocenotic conditions, their ontogenetic structure has different spectral options (left-sided, centered, bimodal) and differs in demographic indicators. Some specific ontogenetic spectra of coenopopulations differ from the characteristic one, which is associated with growing conditions (lack of available substrate, high phytocenotic competition, amount of precipitation) and features of ontogenesis (features of seed regeneration, omission of ontogenetic states). The locations of 16 relict, endemic and included in the list of the Red Book of the Republic of Khakassia plant species have been confirmed and established. In different habitats, the dimensional and morphological polyvariance of ontogenesis was found, expressed in changes in the biometric indicators of shoots and the number of individuals. In disturbed plant communities, an accelerated rate of development of individuals (Oxytropis stenofoliola, O. nuda, O. includes, Coluria geoides) was revealed. It was found that in steppe communities of coeno-populations, normal incomplete, with high elimination of young individuals, have low values of recovery and replacement indices. In meadow and forest communities, the replacement of a mature individual is carried out by two young ones. According to the type of coeno-population, mature, mature, less often – transitional, young. Centered types of spectra are formed, less often – left–sided, in disturbed plant communities - right-sided, bimodal. Monitoring of coeno-populations showed fluctuations in the number of age groups associated with uneven seed replenishment and elimination of young individuals, which is determined by the method of self-maintenance, frequency and intensity of inspiration, edaphic-coenotic and weather conditions. The analysis of the stability of populations in plant communities showed a complex relationship of several factors (species biology, rate and stage of development, ecological and phyto-coenotic conditions, anthropogenic load, etc.). Thus, in steppe plant communities, the predominance of generative individuals was noted, with minor participation of young and old. This is due to the difficulty of seed reproduction and the biology of the species (Oxytropis stenofoliola, O. includens, Astragalus arkalycensis, Zygophyllum pinnatum), fires (Oxytropis stenofoliola, O. nuda, O. includes, Coluria geoides) and absolute conservation (Oxytropis includes, Astragalus ionae). In meadow and forest plant communities, fluctuations in the number of young individuals are determined by the biology of the species (Erantis sibirica, Viola selkirkii), grazing and haymaking (Anemone caerulea, Corydalis bracteata, Adonis vernalis). Coeno-populations are in a definitive state, most are stable. In the developed software package for segmentation and decryption of multispectral satellite images "ECCA-Pack 12", partial learning approaches have been implemented, which make it possible to effectively take into account the small volume of the training sample (up to individual points) in the process of segmentation by spectral-textural features. This approach was implemented for the hierarchical method of segmentation of multispectral images by spectral characteristics of HCA. It has been shown that the use of convolutional neural networks is an effective solution for the task of automated decoding of the composition of stands on RGB images of ultra-high spatial resolution. The U-Net-M architecture was proposed, which is a modification of U-Net with Inception blocks. Experimental comparisons of various convolutional neural network architectures have shown that the best quality of semantic segmentation of aspen and birch stands is provided by the DeepLabV3+ architecture and the proposed U-Net-M.

 

Publications

---