Savanna Physiognomies Research Articles

Complexity of the Male Perfume of Eulaema nigrita from Forest and Woody Physiognomies of the Brazilian Savanna: Is There a Relationship with Body Size and Wing Wear?

Male orchid bees collect volatile and semi-volatile compounds from the environment for storage and accumulation in specialized hind legs. Later, these compounds form a perfume blend used during courtship to lure conspecific females for mating. Male orchid bees perfume has been suggested to play an important role as a sexual signaling trait involved in pre-mating isolation of species, functioning as an indicator of male genetic quality. Eulaema nigrita Lepeletier (Apidae: Euglossini) is a common species in both forested and woody savanna (Cerrado stricto sensu) physiognomies of the Brazilian savanna biome. By identifying the chemical composition of male E. nigrita perfume, we tested for differences in the bouquet chemical profile in populations from remnants of seasonal semideciduous forest and woody savanna. In addition, we assessed the relation between perfume complexity and morphological traits associated with size and age of males. Our analysis showed a low effect of physiognomies on differences in the perfume chemical profile of sampled males. Nevertheless, we observed significant differences in the chemical profile of individuals from two seasonal semideciduous forest remnants, which suggests an environmental effect in individual bouquet. Wing wear measurements were positively related to perfume complexity, consistent with the premise that perfumes from older individuals are indicators of survival capacity in male orchid bees.

Woody species distribution across a savanna-dry forest soil gradient in the Brazilian Cerrado.

Although richness and distribution of woody species in the Cerrado physiognomies have been extensively studied, the shifts of woody species from savanna physiognomies to dry forests have not yet been addressed. Here, we investigate the effect of soil physical-chemical traits on the woody species turnover between adjacent cerrado stricto sensu and dry forest physiognomies. Woody species were surveyed, and soil and topographic variables measured, in 30 10×40 m plots systematically distributed, with 15 plots in each physiognomy. We found a spatially structured distribution of woody species, and differences of soil traits between cerrado stricto sensu and dry forest areas, mainly related to the aluminum saturation, base saturation, and available phosphorus. Aluminum saturation increased toward the savanna area, while base saturation increased toward the dry forest. Most woody species predominated in one physiognomy, such as Callisthene major in the cerrado stricto sensu and Anadenanthera colubrina in the dry forest. Only 20% of the species were widely distributed across both physiognomies or, not often, restricted to the intermediary values of the soil gradient. General results indicate that contrasting soil traits between cerrado stricto sensu and dry forest produce a strongly spatially organized and sharp transition in terms of species distribution between these physiognomies.

First record of phytolith assemblages from the Salto Ander Egg Formation (Late Pleistocene). Palaeoenvironment inferences in the Last Interglacial Stage in southwestern Mesopotamia (Entre Ríos, Argentina)

This contribution aims to describe the first palaeobotanical record of the Late Pleistocene in the Salto Ander Egg Formation (SAEF) through the analysis of phytolith assemblages. The microremains association allowed us to characterise the palaeoenvironmental conditions for these deposits and to complete the palaeobotanical record for the stratigraphic sequence of the Late Pleistocene in southwestern Entre Ríos province, Argentina. The Salto Ander Egg Formation is composed of fluvial deposits with a well-defined floodplain and channel facies dated at ages ranging from 120 to 60 ky. The phytolith content of the unit was analysed along two stratigraphic sections in outcrops exposed on the cliffs of the fluvial valleys located in the drainage basin of La Ensenada creek. The vegetation composition was interpreted principally from the proportions of very frequent and frequent morphotypes in the phytolith assemblages and was grouped in plant functional group affinities: “non-GSSCPs” (diagnostic grass phytoliths); “GSSCPs” (grass silica short cell phytoliths); “Woody dicots” and “Palms”. The phytolith assemblages describe a palaeoflora composed of a continuous herbaceous stratum of C3/C4 grasses, although C3 might have been more significant accompanied by isolated trees and palms in an open savanna physiognomy, probably with a temperate/warm climate with low rainfall and average to high temperatures.

Herb–subshrub diversity in open savanna sites with distinct fire regimes in the Jalapão region, Brazil

AbstractThe fire regime is essential in creating a mosaic of plant structure and diversity in South American open savannas, especially favouring herbs. However, studies investigating diversity patterns in Neotropical savannas rarely focus on the herb–subshrub layer. This study investigated the variation of the herb–subshrub layer under contrasting fire regimes in the most conserved site within the Cerrado Domain, the Jalapão region, Brazil. We selected four sites of open savanna physiognomy with similar topographic, climatic and edaphic features: three burned every 2 years, while the fourth site has remained unburned for at least the last 10 years. We randomly distributed 15 plots of 4 m2 in each site and identified all herbs and subshrubs in each plot to estimate density, richness, alpha diversity and species composition. The unburned site had lower herb–subshrub density, richness and diversity than the frequently burned sites. Species composition varied between frequently burned and unburned sites, partially explained by the fire frequency across sites. Although other ecological factors may explain the patterns detected, we cannot rule out the importance of fire in structuring plant communities in the Jalapão region. As in other savannas, our study in the Cerrado Domain reinforces the essential role of the fire regimes in modifying and maintaining the diversity of herbaceous plants at the landscape scale.

On the combined use of phenological metrics derived from different PlanetScope vegetation indices for classifying savannas in Brazil

Mapping of savannas in Brazil is challenging since there is no consensus on the best remote sensing strategy to deal with the spatial variability of some physiognomies and the spectral similarity of others. In this study, we evaluated the performance of 12 land surface phenology (LSP) metrics calculated from 70 cloud-free PlanetScope (PS) satellite images and three vegetation indices (VIs) for Random Forest (RF) classification of eight savanna physiognomies. The 12 LSP metrics were: the start (SOS), end (EOS), length (LOS), and mean (MGS) of greening season; the mean spring (MSP) and mean autumn (MAU); the VI peak (PEAK) and trough (TRG); the positions of the peak (POP) and trough (POT); and the rates of spring green-up (RSP) and autumn senescence (RAU). These metrics were calculated from the Green-Red Normalized Difference (GRND), Enhanced vegetation Index (EVI), and Normalized Difference Vegetation Index (NDVI). At the protected Ecological Station of Águas Emendadas (ESAE) in central Brazil, we compared the LSP classification in the 2017–2018 seasonal cycle against the VI classification in the 2017 dry season using an existent reference vegetation map for accuracy assessment. Furthermore, we analyzed the performance of the individual and combined sets of VIs and their derived LSP metrics for RF classification of the savanna physiognomies. The results showed that LSP added gains of 19.3% (EVI), 13.1% (NDVI), and 5.4% (GRND) to dry-season VI classification. The overall accuracies of the individual and combined sets of VIs and their retrieved LSP metrics generated gains of 22.8% and 28.1% in relation to the dry-season EVI. In the classification combining LSP metrics, the most important ranked predictors originated from the NDVI and EVI (e.g., TRG, PEAK, MSP, MGS, and RSP). Our findings highlight the importance of the combined use of high spatial and temporal resolution data of the Planet's satellite constellation for the classification of Brazilian savannas leveraging the information retrieved from vegetation phenology. However, when dense time series of a given sensor are not available for retrieving the phenological metrics, an alternative is to use combinedly different VIs calculated in the dry season, when the frequency of cloud cover is reduced over Brazilian savanna areas.

The response of sub‐adult savanna trees to six successive annual fires: An experimental field study on the role of fire season

Abstract In mesic savannas worldwide, trees experience frequent fires, almost all set by humans. Management fires are set to reduce or enhance tree cover. Success depends greatly on responses of sub‐adult trees to such fires. To date, the number of successive years that sub‐adult trees can resprout nor the number of years that they must resist being top‐killed by successive fires, nor the requisite height, have been reported. In a 6‐year experimental field study in Guinean savannas of West Africa, we monitored annually the heights and responses of 1,765 permanently tagged sub‐adult trees under annual fires set in three different periods of the long dry season: early‐dry season (EDS), mid‐dry season (MDS) and late‐dry season (LDS). Annual MDS fires are the common local management protocols of Guinean savannas, although EDS fires are common in some of the savannas. Results showed that overall, the proportion of sub‐adults that resisted being top‐killed differed across fire seasons. Furthermore, resisting one fire gave a better chance of resisting the next. Only sub‐adults that were able to resist direct damage for three successive EDS and MDS fires reached sufficient height to be recruited to the adult stage. Resistance height (avoiding topkill) was ∼1 m for EDS and ∼2 m for both MDS and LDS fires. Recruitment height (threshold for transition to adult stage) was ∼3 m for EDS and ∼ 3.3 m for MDS fires. No height was great enough for sub‐adult trees to be recruited to adult stages in LDS fire. Synthesis and applications. The results of this novel field study showed clearly that successive early‐ and mi‐dry season fires can enhance tree density and that successive late‐dry season fires alone reduce tree density in Guinean savannas due to the effects of successive fires on sub‐adult trees. The results suggest that a planned regime of these seasons of fire could be used to maintain the desired tree density in Guinean savannas and may inform fire management in other mesic savannas where goals are to increase or decrease tree densities. It also provides relevant information for comparative studies on the mechanisms of recruitment of sub‐adult trees to an adult stage in all mesic savannas, a process that ultimately determines savanna physiognomy.

Greater loss and fragmentation of savannas than forests over the last three decades in Yunnan Province, China

Yunnan Province, southwest China, has a monsoonal climate suitable for a mix of fire-driven savannas and fire-averse forests as alternate stable states, and has vast areas with savanna physiognomy. Presently, savannas are only formally recognised in the dry valleys of the region, and a no-fire policy has been enforced nationwide since the 1980s. Misidentification of savannas as forests may have contributed to their low protection level and fire-suppression may be contributing to vegetation change towards forest states through woody encroachment. Here, we present an analysis of vegetation and land-use change in Yunnan for years 1986, 1996, 2006, and 2016 by classifying Landsat imagery using a hybrid of unsupervised and supervised classification. We assessed how much savanna area had changed over the 3 decades (area loss, fragmentation), and of this how much was due to direct human intervention versus vegetation transition. We also assessed how climate (mean annual temperature, aridity), landscape accessibility (slope, distance to roads), and fire had altered transition rates. Our classification yielded accuracy values of 77.89%, 82.16%, 94.93%, and 86.84% for our four maps, respectively. In 1986, savannas had the greatest area of any vegetation type in Yunnan at 40.30%, whereas forest cover was 30.78%. Savanna coverage declined across the decades mainly due to a drop in open parkland savannas, while forest cover remained stable. Savannas experienced greater fragmentation than forests. Savannas suffered direct loss of coverage to human uses and to woody encroachment. Savannas in more humid environments switched to denser vegetation at a higher rate. Fire slowed the rate of conversion away from savanna states and promoted conversion towards them. We identified remaining savannas in Yunnan that can be considered when drafting future protected areas. Our results can inform more inclusive policy-making that considers Yunnan’s forests and savannas as distinct vegetation types with different management needs.

Hierarchical mapping of Brazilian Savanna (Cerrado) physiognomies based on deep learning

The Brazilian Savanna, also known as Cerrado, is considered a global hotspot for biodiversity conservation. The detailed mapping of vegetation types, called physiognomies, is still a challenge due to their high spectral similarity and spatial variability. There are three major ecosystem groups (forest, savanna, and grassland), which can be hierarchically subdivided into 25 detailed physiognomies, according to a well-known classification system. We used an adapted U-net architecture to process a WorldView-2 image with 2-m spatial resolution to hierarchically classify the physiognomies of a Cerrado protected area based on deep learning techniques. Several spectral channels were tested as input datasets to classify the three major ecosystem groups (first level of classification). The dataset composed of RGB bands plus 2-band enhanced vegetation index (EVI2) achieved the best performance and was used to perform the hierarchical classification. In the first level of classification, the overall accuracy was 92.8%. On the other hand, for the savanna and grassland detailed physiognomies (second level of classification), 86.1% and 85.0% were reached, respectively. As the first work that intended to classify Cerrado physiognomies in this level of detail using deep learning, our accuracy rates outperformed others that applied traditional machine learning algorithms for this task.

Savannas of the Brazilian semiarid region: what do we learn from floristics?

The Cerrado represents the largest extension of savanna in South America. It occupies large stretches of central Brazil, being fragmented towards the Northeast, Southeast, and South regions of the country. Examples of disjunct patches of vegetation with savanna physiognomy within the Caatinga can be found in the Chapada Diamantina, the Chapada do Araripe, in small areas of southern Ceará State, and also in the coastal plains. This study recorded the floristic composition of four savannas within the Caatinga in northern Ceará State and evaluated the floristic relationships between these and other savannas, Cerrado and Caatinga sites. Periodic floristic collections recorded 247 species distributed among 162 genera and 55 families. Fifty-seven percent of the recorded species were of herbaceous or sub-shrubby habit, while the majority of the flora was of the therophytic life-form. Biogeographic analyses revealed that the study sites differ from typical Cerrado in flora and life-form spectra and have closer floristic relationships with Caatinga vegetation. The presence of floristic elements from the Cerrado, together with species from the Caatinga, in the study sites allows us to conclude that these savanna enclaves in the Caatinga are composed of a mixed flora with typical elements of these two Brazilian biomes.

Aboveground biomass estimates over Brazilian savannas using hyperspectral metrics and machine learning models: experiences with Hyperion/EO-1

ABSTRACT We investigated the potential of hyperspectral remote sensing to estimate aboveground biomass (AGB) over the Brazilian savannas (Cerrado), the second-largest source of carbon emissions in Brazil. For this purpose, a Hyperion/Earth Observing-1 (EO-1) image was collected in the dry season at the Ecological Station of Águas Emendadas (ESAE). In order to estimate the AGB, we evaluated the performance of five machine learning models (Classification and Regression Trees – CART; Cubist – CB, Partial Least Squares Regression – PLS; Random Forest – RF; and Support Vector Machine – SVM) and four sets of metrics (reflectance, narrowband vegetation indices – VIs; absorption band parameters; and the combination of these attributes). The lowest root mean square error (RMSE) was obtained for RF using VIs (29%) and a combination of metrics (28%). For VIs, RF differed from CUB, PLS and SVM at 5% significance level. From cross-validation results, the RMSE was 26.36% for grasslands, 35.04% for open savannas, and 24.85% for dense savannas. The RF model with VIs had the most stable predictive performance across the models, as indicated by small variations in RMSE from CART to SVM. The five most important ranked VIs in the RF model were the Normalized Difference Vegetation Index (NDVI), Pigment Specific Simple Ratio (PSSR), Enhanced Vegetation Index (EVI), Red Edge Normalized Difference Vegetation Index (RENDVI) and Structure Insensitive Pigment Index (SIPI). Most of their relationships with AGB were non-linear. The resultant AGB estimates showed consistent results with a vegetation cover map of the ESAE. Areas of the ESAE with AGB lower than 10 Mg.ha−1 were coincident with the occurrence of grassland physiognomies (savanna grasslands and shrub savannas), while areas with AGB higher than 25 Mg.ha−1 matched the occurrence of dense savanna physiognomies (woodland savanna and dense woodland savanna). Grassland areas showed larger values of coefficient of variation (CV) than areas of dense savannas. These first-hand results set a baseline of models and metrics for AGB modeling of savannas during the future transition from current sampling-type hyperspectral missions (< 10 km of swath) to large-coverage hyperspectral satellites (> 100 km of swath).

Linking the spatiotemporal variation of litterfall to standing vegetation biomass in Brazilian savannas

Abstract Aims Litterfall at a global scale is affected by climate, edaphic features and vegetation structure, with litter production increasing from grasslands to forests following the rise in standing biomass. However, at landscape scales, the same relationship between litter production and vegetation structure has rarely been studied and comparisons of litterfall patterns between adjacent, structurally distinct communities are lacking. Here, we use a standardized methodology to describe the structural differences among four savanna physiognomies and analyze their relationship with changes in litterfall across the Cerrado. Methods We evaluated the woody vegetation structure and composition in 48 sites, equally distributed across four physiognomies and monitored the monthly litter production from April 2014 to March 2015. Important Findings Results showed that the density, basal area, cylindrical volume and aboveground biomass of woody vegetation differ among physiognomies, increasing consistently from cerrado ralo, cerrado típico, cerrado denso and cerradão. Indeed, we found a strong and positive relationship between aboveground biomass and annual litter production, with litter yield increasing from 0.9 to 8.4 Mg ha−1 across different physiognomies, following the increment in vegetation structure. Monthly production was seasonal and similar among vegetation types, increasing during the dry season. Leaves comprised the dominant fraction (approx. 85%) and litterfall seasonality primarily resulted from the concentration of leaf shedding during dry months. However, the temporal pattern of litterfall throughout the year showed a gradual reduction in the seasonality from open to closed vegetation types, likely following the decrease of deciduous species abundance in the plant community. Our results showed that changes in vegetation structure may affect spatial and temporal litterfall patterns in different physiognomies, which co-occur across the Cerrado landscape, with potential implications for the overall functioning of this ecosystem. Moreover, these findings highlight the use of standardized methods as essential to correctly compare litterfall patterns among different environments.

Savanna canopy trees under fire: long‐term persistence and transient dynamics from a stage‐based matrix population model

AbstractFire is a major disturbance driving the dynamics of the world's savannas. Almost all fires are set by humans who are increasingly altering fire timing and frequency on every continent. The world's largest protected areas of savannas are found in monsoonal northern Australia. These include relatively intact, tall, open forests where traditional indigenous fire regimes have been largely replaced in the past half century by contemporary patterns with trees experiencing fire as often as three out of five years. Eucalypt canopy trees form the basic structure of these savannas and changes to the canopy due to fire regimes cascade to affect other plants and animals. In this study, we used data from nearly three decades of field studies on the effects of fire on individual trees to define eight life‐history stages and to calculate transition rates among stages. We developed a stage‐based matrix population model that explicitly considers how fire season and understory influence growth, survival, and recruitment for each life‐history stage. Long‐term population growth rates and transient population dynamics were calculated under five different fire regimes, each in two understory types, using both deterministic and stochastic simulations of seasonal timing of fires. We found that fire was necessary for long‐term persistence of eucalypt canopy tree populations but, under annual fires, most populations did not survive. Population persistence was highly dependent on fire regime (fire season and frequency) and understory type. A stochastic model tended to yield higher population growth rates than the deterministic model with regular, periodic fires, even under the same long‐term frequency of fires. Transient population dynamics over 100 yr also depended on fire regime and understory, with implications for savanna physiognomy and management. Model predictions were tested in an independent data set from a 21‐yr longitudinal field study in Kakadu National Park. This study is a novel and integrative contribution to our understanding of fire in savanna biomes regarding the potential for long‐term persistence and transient dynamics of savanna canopy tree populations. The model is relatively simple, generalizable, and adaptable for further investigations of the population dynamics of savanna trees under fire.

Angiosperms of dry grasslands and savannahs of Jalapão, the largest conserved Cerrado area in Brazil

Abstract This study provides a checklist of angiosperm species from Jalapão region, Tocantins, Brazil, inserted within the Cerrado biome, a global biodiversity hotspot. The region of Jalapão is still well preserved despite current threats to its biodiversity, however its plant diversity is still poorly understood. To generate the present checklist, fieldwork was carried out and relevant herbarium collections were consulted. Angiosperms distributed in the grassland and savanna physiognomies with dry, non-rocky soils were recorded. We detected 550 species within 85 families. The richest families are Leguminosae, Poaceae, Asteraceae, Lamiaceae, Rubiaceae, Myrtaceae, Malpighiaceae and Euphorbiaceae. Among the plant species, ten are listed as rare, ten are threatened, and twelve were identified as new to science. This study increases the number of angiosperm species occurring in these vegetation types in Jalapão almost 2-fold compared to previous inventories. The Cerrado biome and Jalapão region are under heavy threat due to agricultural expansion, and our study contributes to the knowledge of plant biodiversity as a fundamental step towards designing and carrying out conservation actions.

The Year 2017: Megafires and Management in the Cerrado

The year 2017 was a megafire year, when huge areas burned on different continents. In Brazil, a great extension of the Cerrado burned, raising once more the discussion about the “zero-fire” policy. Indeed, most protected areas of the Cerrado adopted a policy of fire exclusion and prevention, leading to periodic megafire events. Last year, 78% of the Chapada dos Veadeiros National Park burned at the end of the dry season, attracting media attention. Furthermore, 85% of the Reserva Natural Serra do Tombador burned as a result of a large accumulation of fuel caused by the zero-fire policy. In 2014, some protected areas started to implement the Integrate Fire Management (IFM) strategy. During 2017, in contrast to other protected areas, the Estação Ecológica Serra Geral do Tocantins experienced no megafire events, suggesting that a few years of IFM implementation led to changes in its fire regime. Therefore, we intended here to compare the total burned area and number of fire scars between the protected areas where IFM was implemented and those where fire exclusion is the adopted policy. The use of fire as a management tool aimed at wildfire prevention and biodiversity preservation should be reconsidered by local managers and environmental authorities for most Cerrado protected areas, especially those where open savanna physiognomies prevail. Changing the paradigm is a hard task, but last year’s events showed the zero-fire policy would bring more damage than benefits to Cerrado protected areas.

Response of vegetation to fire disturbance: short-term dynamics in two savanna physiognomies

Fire is a constitutive ecological force in savanna ecosystems, but few studies have monitored its short-term effects on plant community dynamics. This study investigated changes in plant diversity in the South American savanna (Cerrado) after severe disturbance by fire. We monitored 30 permanent plots (10 m × 5 m) distributed in two Cerrado physiognomies (típico: more forested; ralo: grass-dominated), being 10 plots in the area disturbed by fire, and five in a preserved control area (undisturbed). From August 2010 to June 2011, we evaluated changes in species richness, abundance and composition of savanna vegetation. Monitoring started one week after the fire; disturbed plots were surveyed monthly, while control plots were surveyed every two months. We observed rapid reassembling in both physiognomies: plots affected by fire showed rapid increase in species richness and plant density during the first four months after the disturbance. Concerning species composition, disturbed plots in the cerrado típico tended to converge to control plots after one year, but each local assemblage followed particular temporal trajectories. A different pattern characterized cerrado ralo plots, which showed heterogeneous trajectories and lack of convergence between disturbed and control plots; the structure of these assemblages will likely change in next years. In conclusion, our results showed that fire significantly affected plant diversity in the two savanna physiognomies (cerrado típico and ralo), but also indicated that community reassembling is fast, with different dynamics between Cerrado physiognomies.

Temporal changes in species composition, diversity, and woody vegetation structure of savannas in the Cerrado-Amazon transition zone

ABSTRACT Vegetation in the transition between tropical forest and savanna is hyperdynamic and there is evidence that in the absence of fire, forest advances over savanna. Between 2008 and 2013 we evaluated changes in species composition and diversity and in the structure of the woody vegetation of savanna physiognomies in the transition between the Cerrado and Amazon biomes that were fire free for 11 years. The physiognomies form a gradient from savanna woodland (Typical Cerrado - TC), to low woodland (Dense Cerrado - DC), to woodland (locally called Cerradao - CO). We hypothesise that: 1) the more open physiognomies (TC and DC) are more dynamic compared to the closed physiognomy (CO); and 2) in the absence of fire vegetation tends to become more forested. We found that: 1) TC was more dynamic (e.g. greater increases in richness, diversity, and abundance of plants and basal area) than CO and DC; 2) The three physiognomies experienced an increase in basal area and abundance of individuals, but only certain key species contributed to these increases. These results indicate that the open physiognomies were more dynamic than the closed physiognomies, and in the absence of fire the savanna physiognomies became more forested and accumulated biomass.

Vegetation types of the upper Madeira River in Rondônia, Brazil

We characterize the vegetation types of the upper basin of the Madeira River in the Brazilan state of Rondonia, a biodiverse region with elevated rates of habitat loss. Vegetation and environmental parameters were recorded from 37 observation points distributed along and near a 160 km stretch of the Madeira River and representing the range of regional environments. Analysis of structure and floristic variables, as well as associated edaphic attributes and water table fluctuation, permitted recognition of five main vegetation types and seven subtypes. Open Ombrophilous Forest was the most frequently encountered vegetation type and occurred on well-drained, nutrient-poor soils, whereas Dense Ombrophilous Forest was seldom recorded. Alluvial Ombrophilous Forests (varzea) were found along a narrow strip of land along the banks of the Madeira River on the most fertile soils in the study area. Semideciduous Forests were found on small areas of rocky outcrops with shallow soils and reduced water availability during the dry season. Campinaranas, which range from open savanna physiognomies to closed canopy forests, were found to be a key environment in the lowlands south of the Madeira River on silty hydromorphic soils, where they harbor a peculiar flora tolerant of flooding during the rainy season. Our classification of the main vegetation types in the upper Madeira River illuminates a high degree of floristic and environmental heterogeneity in a highly threatened region. Our results will be useful for designing conservation strategies aimed at protecting the full range of floristic diversity present in the region.

Potential of hyperspectral metrics and classifiers for mapping Brazilian savannas in the rainy and dry seasons

Land cover mapping of savannas in Brazil, a world's hotspot of biodiversity, is still challenging due to the tree cover gradient and the spectral similarity between some vegetation physiognomies. Here, we evaluated the potential of four classifiers (Decision Tree (DT), Random Forest (RF), Spectral Angle Mapper (SAM) and Support Vector Machine (SVM)) for discriminating eight savanna physiognomies in the rainy and dry seasons of the Brasília National Park (BNP). Five sets of Hyperion/Earth Observing One (EO-1) metrics (reflectance, first-order derivative of reflectance; narrow-band vegetation indices (VIs); absorption band parameters; and the combination of these attributes) were tested as input data for each classifier. Before classification, the Correlation-based Feature Selection (CFS) algorithm was applied to reduce data dimensionality. To evaluate the agreement between the classifications of the different techniques, we calculated the Shannon entropy. Finally, Monte Carlo simulation was applied to determine the presence of statistical differences in classifiers and metrics between seasons. The results showed that the greater spectral confusion between the savannas generally observed in the rainy season was compensated by the selection of a greater number of hyperspectral metrics for classification. SVM and RF had the highest overall classification accuracy (OA) and Kappa values in the rainy and dry seasons for each set of metrics. The reflectance and VIs presented better discrimination capability than the absorption band parameters and first-order derivative data. When all metrics were considered in the analysis, gains of 6% and 8% in OA were obtained over the first ranked classifiers and metrics (SVM with reflectance in the rainy season; RF with VIs in the dry season). The lowest Shannon entropy values in the rainy and dry seasons were observed for VIs and reflectance and for physiognomies with larger vegetation cover, while the largest uncertainties were noted in savanna grassland/shrub areas. From the Monte Carlo simulations, differences in Kappa between seasons were not statistically significant for most of the metrics and classifiers at the 99% confidence level. Variations in brightness and VIs, associated with canopy structure, biochemistry and physiology, were therefore more important for classification than variations in spectral features, spectra shape and absorption bands.

ASSESSMENT OF TEXTURE FEATURES FOR BRAZILIAN SAVANNA CLASSIFICATION: A CASE STUDY IN BRASÍLIA NATIONAL PARK

Distinguishing Brazilian savanna physiognomies is an essential task to better evaluate carbon storage and potential emissions of greenhouse gases. In this study, we propose to evaluate the potential of texture features to improve the discrimination among fi ve physiognomies in the Brazilian savanna: Open Grasslands, Shrubby Grassland, Shrubby Savanna, Savanna Woodland and Gallery Forest. Texture features extracted from RapidEye images and also from Spectral Linear Mixture Model components and Vegetation Index were evaluated. Results showed that texture features based on GLCM can reduce misclassifi cation for Open Grasslands, Shrubby Grasslands and Shrubby Savanna classes.

No evidence of intrinsic spatial processes driving Neotropical savanna vegetation on different substrates

AbstractThe relationships between floristic patterns and environmental variation in tropical savannas have been the focus of many studies worldwide. However, important aspects of these relationships, such as the role of geographic distance in structuring plant communities, have received little attention. We investigated the individual and combined influences of substrate, climatic, and spatial factors on the floristic‐structural dissimilarity between two savanna physiognomies in the core region of Brazilian savannas: one on plain relief with deep soils and another on steep relief with shallow rocky soils. Ten 1‐ha plots were sampled in each physiognomy. We modeled species abundance using multiple linear models and variance partitioning. Our results indicated that spatial processes that are intrinsically related to species variation have negligible effects on floristic variation. The most important predictors in our models were related to soil characteristics (mainly nutrient availability) and topography (relief and elevation). Consequently, the substrate component exhibited the greatest power (14%) in explaining the floristic‐structural variation in the overall variance partitioning. Our results provide the first demonstration of the individual and combined contributions of substrate, climatic, and spatial factors to the occurrence and abundance of woody species in the most diverse and threatened savanna in the world. We also provide evidence that neutral processes might not be strong predictors of vegetation structure where savanna substrates differ greatly; instead, community structure may be primarily regulated by environmental filters.