Reed beds, often referred to as dense, nearly monotonous extensive stands of common reed (Phragmites australis), are the most productive vegetation form of inland waters in Central Asia and exhibit great potential for biomass production in such a dryland setting. With its vast delta regions, Kazakhstan has the most extensive reed stands globally, providing a valuable case for studying the potential of reed beds for the bioeconomy. However, accurate and up-to-date figures on available reed biomass remain poorly documented due to data inadequacies in national statistics and challenges in measuring and monitoring it over large and remote areas. To address this gap in knowledge, in this study, the biomass resource characteristics of common reed were estimated for one of the significant reed bed areas of Kazakhstan, the Syr Darya Delta, using ground-truth field-sampled data as the dependent variable and high-resolution Sentinel-2 spectral bands and computed spectral indices as independent variables in multiple Random Forest (RF) regression models. An analysis of the spatially detailed yield map obtained for Phragmites australis-dominated wetlands revealed an area of 58,935 ha under dense non-submerged and submerged reed beds (with a standing biomass of >10.5 t ha−1) and an estimated 1,240,789 tons of reed biomass resources within the Syr Darya Delta wetlands. Our findings indicate that submerged dense reed exhibited the highest biomass at 28.21 t ha−1, followed by dense non-submerged reed at 15.24 t ha−1 and open reed at 4.36 t ha−1. The RF regression models demonstrated robust performance during both calibration and validation phases, as evaluated by statistical accuracy metrics using ten-fold cross-validation. Out of the 48 RF models developed, those utilizing the Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI) as key predictors yielded the best standing reed biomass estimation results, achieving a predictive accuracy of R2 = 0.93, Root Mean Square Error (RMSE) = 2.74 t ha−1 during the calibration, and R2 = 0.83, RMSE = 3.71 t ha−1 in the validation, respectively. This study highlights the considerable biomass potential of reed in the region’s wetlands and demonstrates the effectiveness of the RF regression modeling and high-resolution Sentinel-2 data for mapping and quantifying above-ground and above-water biomass of Phragmites australis-dominated wetlands over a large extent. The results provide critical insights for managing and conserving wetland ecosystems and facilitate the sustainable use of Phragmites australis resources in the region.

The oil palm sector is widely known for its strong impact on biodiversity and human well-being. The United Nations Sustainable Development Goals (SDG) framework calls for a sustainable biodiversity management approach to oil palm plantations and thereby balancing palm oil production and biodiversity conservation. Heterogeneous agricultural landscapes, comprising different vegetation types in a complex spatial pattern, provide habitat to a variety of species and support high biodiversity. Biodiversity can thus offer pest control ecosystem services to agricultural systems. However, so far it remains unclear whether landscape structure and diversity foster or mitigate pest occurrences in plantations. Our case study is the oil palm plantation of Mapirip ' an in Colombia, which integrates local ecosystems such as secondary forests, relict old-grown forests, riparian forests, and water body remnants in between and around the palm stands of the plantation. With this nature-improved design, we pound pose pound two main questions: which landscape structural properties characterize the land-scape of the oil palm plantation?, and how do specific landscape structural properties foster or mitigate the presence of two main oil palm pests (the butterfly larvae split-banded owlet (Opsiphanes cassina) and the red weevil (Rhynchophorus palmarum))? We first characterize the plantation landscape with a multivariate cluster analysis which led us to identify up to nine different landscape structural types for the Macondo plantation. A mosaic of different landscape types ranging from homogeneous and simple landscapes to heterogeneous, diversified, and connected landscape types. Second, we performed an NMDS ordination to show similarities among the landscape structural types, land cover, and pest occurrence. Our results show that the integration of local ecosystems such as forests and water bodies improves landscape connectivity and suppresses the abundance of the two main oil palm pests.

Aims: Under global climate and land use change, the vegetation layer of steep high mountain slopes is prone to increasing disturbance via erosion and mass-wasting events. The ecological restoration and stabilization of an intact vegetation layer on steep slopes with the soil seed bank is therefore an important factor for functional high mountain ecosystems. We analyze the relationships between the soil seed bank composition, topography, and land use and compare the plant species composition of the aboveground vegetation to the soil seed bank. Study area: Stepantsminda, Georgia, Greater Caucasus (1,800-2,500 m a.s.l.). Methods: 74 grassland vegetation releves (5 x 5 m) on steep high mountain slopes were compared to the respective persistent soil seed bank (05 cm depth), assessed via the seedling emergence method. Non-metric multidimensional scaling and vector fitting uncovered the relationships between topography, land use, and soil on the seed bank and aboveground vegetation composition. The similarity between both datasets was assessed with a Mantel test. Vegetation types were described with indicator species analysis for both aboveground and belowground plant species composition. Results: Exposure to the north and incoming solar radiation relate to seed bank composition, as well as the belowground species diversity, whereas the effect of land use on the soil seed bank remains unclear. The similarity between the aboveground vegetation and the soil seed bank is generally low (18% shared species, Mantel test r = 0.21, no common indicator species). Conclusions: The higher seed density and belowground species richness on intensively illuminated and pastured south-exposed slopes imply a higher restoration potential in the case of disturbance events such as mass-wasting. However, the overall moderate diversity within the soil seed bank challenges the restoration of the aboveground grassland, especially on north-exposed slopes.

AimsThe encroachment of tree and shrub species in high mountains is an increasing worldwide phenomenon, which is expected to dramatically alter high-mountain ecosystems and their functioning. Moreover it indicates in some cases a reforestation process, which will result in important ecological and social benefits, including carbon sequestration and protection against landslides. We therefore examined the spatial extent of forest growth and shrub encroachment mainly of birch (Betula litwinowii) in the sub-alpine belt of the Central Greater Caucasus between 1987 and 2010 and its relation to topographic site conditions. LocationKazbegi district, Central Greater Caucasus, Georgia. MethodsWe analysed 155 vegetation releves sampled in 2009, 2011 and 2015, mainly derived from the Caucasus Vegetation Database, to obtain information about topographic site conditions and structure of B.litwinowii stands. B.litwinowii forest growth was assessed by digitizing the forest outlines from aerial and space-borne imagery (1987, 2005 and 2010). To identify areas of B.litwinowii encroachment as an indicator for different encroachment stages, we modelled the tree and shrub cover using the Random Forest algorithm. ResultsWe found four types of B.litwinowii stands, characterized by different tree and shrub coverage (initial Bromus variegatus-Betula litwinowii encroachment indicating the first stage of succession, Aconitum nasutum-Betula litwinowii forest, Rubus idaeus-Betula litwinowii forest and Rhododendron caucasicum-Betula litwinowii tree line scrubs). B.litwinowii forest increased 25% compared to 1987 mainly in an uphill direction. Furthermore the modelled tree and shrub cover (R-2=.69) could be related to the four vegetation types. ConclusionsOur results indicate a recent trend towards shrub encroachment and consequently reforestation in the Kazbegi region.

Plant functional groups—in our case grass, herbs, and legumes—and their spatial distribution can provide information on key ecosystem functions such as species richness, nitrogen fixation, and erosion control. Knowledge about the spatial distribution of plant functional groups provides valuable information for grassland management. This study described and mapped the distribution of grass, herb, and legume coverage of the subalpine grassland in the high-mountain Kazbegi region, Greater Caucasus, Georgia. To test the applicability of new sensors, we compared the predictive power of simulated hyperspectral canopy reflectance, simulated multispectral reflectance, simulated vegetation indices, and topographic variables for modeling plant functional groups. The tested grassland showed characteristic differences in species richness; in grass, herb, and legume coverage; and in connected structural properties such as yield. Grass (Hordeum brevisubulatum) was dominant in biomass-rich hay meadows. Herb-rich grassland featured the highest species richness and evenness, whereas legume-rich grassland was accompanied by a high coverage of open soil and showed dominance of a single species, Astragalus captiosus. The best model fits were achieved with a combination of reflectance, vegetation indices, and topographic variables as predictors. Random forest models for grass, herb, and legume coverage explained 36%, 25%, and 37% of the respective variance, and their root mean square errors varied between 12–15%. Hyperspectral and multispectral reflectance as predictors resulted in similar models. Because multispectral data are more easily available and often have a higher spatial resolution, we suggest using multispectral parameters enhanced by vegetation indices and topographic parameters for modeling grass, herb, and legume coverage. However, overall model fits were merely moderate, and further testing, including stronger gradients and the addition of shortwave infrared wavelengths, is needed.

Mountain regions cover one quarter of the Earth's terrestrial surface, and are both valuable and vulnerable areas with complex human-environmental interrelationships. In this coupled system, land-use changes induced by political or socio-economic transformations generate consequences for ecological landscape functions like soil productivity and species richness, and integrative land-use concepts provide the potential of a sustainable land development. In the Kazbegi region in the central Greater Caucasus of Georgia, these transformations further lead to landscape-structure change and population marginalization. Hence, we developed three agricultural land-use scenarios that meet Agenda 2030 Sustainable Development Goals to ensure a sustainable rural land development and the conservation of mountain ecosystems. Our normative scenario approach integrates quantitative and qualitative findings of empirical research in landscape ecology, soil science, vegetation ecology as well as agronomics and socio-economics. According to the examined environmental and socio-economic resources, we defined various scenario logics and normative assumptions that combine optimized livestock production (in dairy cow keeping and cattle fattening) with ecological limitations to maintain the functioning of mountain ecosystems. The rule-based scenarios achieved measurably increased outputs in biomass yields, livestock production and related revenues at the regional scale. Further, GIS generated scenario maps demonstrate the related land-use patterns spatially explicit and in high resolution, and visualize the alternative future from local to the regional scale. In conclusion, scenario development helps to determine region-specific and integrated land-use options to provide a sound base for land users and decision makers. Based on research on multiple landscape functions, this approach can assist sustainable land development in a mountain region.

High mountain grasslands offer multiple goods and services to society but are severely threatened by improper land use practices such as abandonment or rapid intensification. In order to reduce abandonment and strengthen the common extensive agricultural practice a sustainable land use management of high mountain grasslands is needed. A spatially detailed yield assessment helps to identify possible meadows or, on the contrary, areas with a low carrying capacity in a region, making it easier to manage these sites. Such assessments are rarely available for remote and inaccessible areas. Remotely sensed vegetation indices are able to provide valuable information on grassland properties. These indices tend, however, to saturate for high biomass. This affects their applicability to assessments of high-yield grasslands.The main aim of this study was to model a spatially explicit grassland yield map and to test whether saturation issues can be tackled by consideration of plant species composition in the modelling process. The high mountain grassland of the subalpine belt (1800 - 2500 m a.s.l.) in the Kazbegi region, Greater Caucasus, Georgia, was chosen as test site for its strong species composition and yield gradients.We first modelled the species composition of the grassland described as metrically scaled gradients in the form of ordination axes by random forest regression. We then derived vegetation indices from Rapid Eye imagery, and topographic variables from a digital elevation model, which we used together with the multispectral bands as predictive variables. For comparison, we performed two yield models, one excluding the species composition maps and one including the species composition map as predictors. Moreover, we performed a third individual model, with species composition as predictors and a split dataset, to produce the final yield map.Three main grassland types were found in the vegetation analysis: Hordeum violaceum-meadows, Gentianella caucasea-grassland and Astragalus captiosus-grassland. The three random forest regression models for the ordination axes explained 64%, 33% and 46% of the variance in species composition. Independent validation of modelled ordination scores against a validation data set resulted in an R-2 of 0.64, 0.32 and 0.46 for the first, second and third axes, respectively. The model based on species composition resulted in a R-2 = 0.55, whereas the benchmark model showed weaker relationships between yield and the multispectral reflectance, vegetation indices, and topographical parameters (R-2 = 0.42). The final random forest yield model used to derive the yield map resulted in 62% variance explained and an R-2 = 0.64 between predicted and observed biomass. The results further indicate that high yields are generally difficult to predict with both models.The benefit of including a species composition map as a predictor variable for grassland yield lies in the preservation of ecologically meaningful features, especially the occurrence of high yielding vegetation type of Hordeum violaceum meadows is depicted accurately in the map. Even though we used a gradient based design, sharp boundaries or immediate changes in productivity were visible, especially in small structures such as arable fields or roads (Fig. 6b), making it a valuable tool for sustainable land use management. The saturation effect however, was mitigated by using species composition as predictor variables but is still present at high yields. (C) 2017 Elsevier Ltd. All rights reserved.

Questions: Shrub encroachment has been observed in many alpine and arctic environments and is expected to significantly alter these ecosystems. Mapping these processes with remote sensing is a powerful tool for monitoring purposes. Thus, we test the distinctiveness of the reflectance signature of target species relative to their co-occurring shrub species using uni- and multivariate analyses for an alpine ecosystem. We ask: (i) is it possible to differentiate shrub species with a unique growth form by their reflectance signature; (ii) which of the tested multispectral sensors produces the best separation; and (iii) how are the results affected by the timing of data acquisition in the vegetation period?Location: Kazbegi district, Central Greater Caucasus, Georgia.Method: We analysed three shrub (Betula litwinowii, Rhododendron caucasicum, Hippophae rhamnoides) and one tall forb (Veratrum lobelianum) species occurring in the sub-alpine to alpine belt The vegetation of 52 releves was analysed using non-metric multidimensional scaling and indicator species analysis. From field spectrometric data we simulated multispectral sensor bands (IKONOS, Quickbird 2, RapidEye, WorldView-2) directly taken from the target species. We analysed the reflectance signature in RapidEye data from June and September. For all data sets we calculated the Jeffries-Matusita distance (JMD) as a separation measure and tested the reflectance signature of the single bands for differences.Results: Betula litwinowii and V. lobelianum always co-occurred in our data. A high abundance of B. litwinowii could also be found in the Rhododendron cluster and vice versa, whereas the Hippophae cluster was more homogeneous. Simulated bands showed good overall separation (JMD 1.58-2.00) of the target species. The separation increased with the increase of number of bands and inclusion of the red edge band. There was a general trend in which the reflectance from satellite images produced a lower separation (JMD 1.20-1.55) than the simulations, with the best separation in the late vegetation period.Conclusion: Our results showed the possibility to spectrally separate encroaching shrub species with a unique growth form in a high-mountain environment using simulated multispectral data and satellite imagery.

In the Georgian Caucasus, unregulated grazing has damaged grassland vegetation cover and caused erosion. Methods for monitoring and control of affected territories are urgently needed. Focusing on the high-montane and subalpine grasslands of the upper Aragvi Valley, we sampled grassland for soil, rock, and vegetation cover to test the applicability of a site-specific remote-sensing approach to observing grassland degradation. We used random-forest regression to separately estimate vegetation cover from 2 vegetation indices, the Normalized Difference Vegetation Index (NDVI) and the Modified Soil Adjusted Vegetation Index (MSAVI2), derived from multispectral WorldView-2 data (1.8 m). The good model fit of R2 = 0.79 indicates the great potential of a remote-sensing approach for the observation of grassland cover. We used the modeled relationship to produce a vegetation cover map, which showed large areas of grassland degradation.

Mountainous grassland landscapes increasingly experience modified management and land use, which are often triggered by socio-economic and climate change. Remote-sensing based mapping approaches are needed to monitor gradual changes in grassland composition and biodiversity of remote and inaccessible areas. Hyperspectral remote sensing in combination with regression and ordination techniques is a promising approach to map continuous representations of prominent floristic gradients at the landscape scale. This approach has, however, to date not been tested in alpine environments with difficult terrain. In the present study we tested whether hyperspectral data allows for the differentiation of species-rich grassland types of a subalpine pasture landscape in Georgia, Greater Caucasus. Due to the unavailability of space- or airborne hyperspectral data, we used field-spectrometric in situ data to investigate this potential.We sampled the vegetation of four grassland types (as classified by cluster analysis) using the Braun-Blanquet approach. For these 60 plots we made four measurements of their hyperspectral canopy reflectance (325-1075 nm) on two dates within a seven-day period at biomass peak. Floristic gradients were derived from Non-metric Multi Dimensional Scaling (NMDS) ordination. These gradients were subsequently subjected to Partial Least Square Regression (PLSR) to predict NMDS scores from the corresponding canopy reflectance.Cross validated Pearson R-2 values for the PLSR models derived from the four measurements ranged from 0.60 to 0.83. Reflectance data sampled on the same day yielded similar results in the respective models, while temporal transferability within the seven day period was limited.The reflectance values of the significant wavelengths in the best models were subjected to Principal Component Analysis (PCA) to display the spectral similarity of plots. The congruency between the similarity of species composition (NMDS) and spectral similarity (PCA) was satisfactorily assessed by Procrustes rotation.The floristic gradients in our study were directly related to pronounced biophysical and environmental gradients, which are responsible for differences in the spectral signatures. Our results indicate that hyperspectral data has the potential to differentiate the composition of Caucasian subalpine grassland types and offers multiple opportunities for very detailed vegetation mapping in difficult terrain. (c) 2013 Elsevier B.V. All rights reserved.