Ana Isabel Rosa Cabral

About 80% of Angola’s forest surface is covered by Miombo and Mopane woodlands, which are explored for diverse activities such as fuelwood and food. This study aimed to assess the recovery dynamics of Miombo and Mopane woodlands after the selective cutting of tree species for charcoal production. For that, the structure and composition of plant communities in 37 plots, located in southwestern Angola, were characterized in fallows of different ages. Results showed that the diameter at breast height, basal area, biomass, and biovolume of trees all rose as the age of the fallow increased, and there were no significant differences in richness, diversity, or dominance of trees between adult–young classes or recent–older fallows. In Mopane, fallows took longer to regenerate, were more affected by environmental and anthropogenic factors, and also presented a higher species adaptation to disturbance. There were more sprouter and seeder trees in Miombo, and new kilns were more distant from roads and villages. Moreover, the selective removal of species deeply altered the community structure and dynamics, despite not directly affecting tree diversity. Thus, new management strategies are needed to ensure the survival of these woodlands such as expanding protected areas and increasing systematic research.

Guinea-Bissau remains among the African countries most vulnerable to climate change due to its flat topography and large meandering coastal area invaded by the tides.We present a case study of the island-village of Djobel, showing the dramatic consequences of socio-environmental change. The inhabitants' attempts to mitigate the impact of extreme weather events and sea-level rise and adapt to the new circumstances have been hampered by the seizure of their upland territory by a neighboring village. In the context of a disconnect between state and society, Djobel's inhabitants want the state to guarantee their rights to productive land, water, and safe housing conditions. This case shows the need for legal protection of the rights of poor, marginalized, and underrepresented populations at risk of becoming climate refugees in the fragile context of politically unstable, failed, and often authoritarian states.

Summary: The Brazilian Legal Amazon contains important reservoirs of forest that are threatened by stakeholders’ behaviour. The Cadastro Ambiental Rural (CAR; Rural Environmental Registry) database was used to define the limits and classes of landowner according to private property size. For each class, we identified the annual forest-cover/cover-loss profile at 2-year intervals from 2000 to 2020 based on Brazilian Annual Land Use and Land Cover Mapping Project (MapBiomas) and Program for Deforestation Monitoring (PRODES) data. The analysis revealed that very large landowners dominate the CAR-registered area and that deforestation is influenced by landowner type. The cumulative contributions to deforestation were 2 916 245.96, 1 234 216.79, 2 871 400.36, 2 805 058.62 and 2 637 485.60 ha for very large landowners, large landowners, medium landowners, small landowners and very small landowners, respectively. Very large landowners (1.7% of the total number of properties) had more forest on their properties but caused the greatest amount of deforestation in total, often associated with agribusiness. Small and very small landowners were more numerous (21.0% and 68.3% of the total number of properties, respectively), but they owned a small total area and contributed less to total deforestation. Property size and landownership asymmetry should be considered in deforestation control policies and commitments to the Sustainable Development Goals agenda.

One of the main applications of machine learning (ML) in remote sensing (RS) is the pixel-level classification of satellite images into land cover types. Although classes with different spectral signatures can be easily separated, e.g. aquatic and terrestrial land cover types, others have similar spectral signatures and are hard to separate using only the information within a single pixel. This work focused on the separation of two cover types with similar spectral signatures, cocoa agroforest and forest, over an area in Pará, Brazil. For this, we study the training and application of several ML algorithms on datasets obtained from a single composite image, a time-series (TS) composite obtained from the same location and by preprocessing the TS composite using simple TS preprocessing techniques. As expected, when ML algorithms are applied to a dataset obtained from a composite image, the median producer’s accuracy (PA) and user’s accuracy (UA) in those two classes are significantly lower than the median overall accuracy (OA) for all classes. The second dataset allows the ML models to learn the evolution of the spectral signatures over 5 months. Compared to the first dataset, the results indicate that ML models generalize better using TS data, even if the series are short and without any preprocessing. This generalization is further improved in the last dataset. The ML models are subsequently applied to an area with different geographical bounds. These last results indicate that, out of seven classifiers, the popular random forest (RF) algorithm ranked fourth, while XGBoost (×GB) obtained the best results. The best OA, as well as the best PA/UA balance, were obtained by performing feature construction using the M3GP algorithm and then applying XGB to the new extended dataset.

La implementación de esquemas de pago por servicios ambientales (PSA) en áreas públicas protegidas plantea varias preguntas en los ámbitos económico, social y ambiental. En este documento, analizamos el caso del Programa Bolsa Floresta (PBF) de Brasil, establecido en las Unidades de Conservación de la Naturaleza del estado de Amazonas y considerado uno de los programas más grandes del mundo en términos de área de cobertura. Primero, verificamos si el programa realmente cumple con los requisitos de un PSA típico: condicionalidad, adicionalidad y voluntariedad. En segundo lugar, buscamos evaluar los dos requisitos clave para que un PSA funcione bien: efectividad económica y sostenibilidad ecológica. Al estudiar el caso de la Reserva de Desarrollo Sostenible de Uatumã, determinamos que los pagos realizados no cubren los costos de oportunidad del proveedor-receptor, lo que resulta en la no entrega de los servicios ambientales contratados. El área total de bosque secundario (SF) disponible para ser incorporada en sucesivos ciclos de cultivo puede no ser lo suficientemente grande como para permitir períodos de barbecho de una duración lo suficientemente larga para que la fertilidad del suelo se recupere completamente. La intensificación de las actividades agrícolas en las áreas de SF puede conducir a una degradación del sistema agrícola tradicional y la consecuente pérdida de los servicios ecosistémicos que este sistema puede proporcionar potencialmente. En resumen, estos impactos negativos indican que, al menos en el DEG de Uatumã, el BFP podría no ser efectivo a medio o largo plazo.

Genetic programming (GP) is a powerful machine learning (ML) algorithm that can produce readable white-box models. Although successfully used for solving an array of problems in different scientific areas, GP is still not well known in the field of remote sensing. The M3GP algorithm, a variant of the standard GP algorithm, performs feature construction by evolving hyperfeatures from the original ones. In this work, we use the M3GP algorithm on several sets of satellite images over different countries to create hyperfeatures from satellite bands to improve the classification of land cover types. We add the evolved hyperfeatures to the reference datasets and observe a significant improvement of the performance of three state-of-the-art ML algorithms (decision trees, random forests, and XGBoost) on multiclass classifications and no significant effect on the binary classifications. We show that adding the M3GP hyperfeatures to the reference datasets brings better results than adding the well-known spectral indices NDVI, NDWI, and NBR. We also compare the performance of the M3GP hyperfeatures in the binary classification problems with those created by other feature construction methods such as FFX and EFS.

Genetic Programming (GP) is a powerful Machine Learning (ML) algorithm that can produce readable white-box models. Although successfully used for solving an array of problems in different scientific areas, GP is still not well known in Remote Sensing. The M3GP algorithm, a variant of the standard GP algorithm, performs Feature Construction by evolving hyper-features from the original ones. In this work, we use the M3GP algorithm on several sets of satellite images over different countries to create hyper-feature from satellite bands to improve the classification of land cover types. We add the evolved hyper-features to the reference datasets and observe a significant improvement of the performance of three state-of-the-art ML algorithms (Decision Trees, Random Forests and XGBoost) on multiclass classifications and no significant effect on the binary classifications. We show that adding the M3GP hyper-features to the reference datasets brings better results than adding the well-known spectral indices NDVI, NDWI and NBR. We also compare the performance of the M3GP hyper-features in the binary classification problems with those created by other Feature Construction methods like FFX and EFS.

Genetic Programming (GP) is a powerful Machine Learning (ML) algorithm that can produce readable white-box models. Although successfully used for solving an array of problems in different scientific areas, GP is still not well known in Remote Sensing. The M3GP algorithm, a variant of the standard GP algorithm, performs Feature Construction by evolving hyper-features from the original ones. In this work, we use the M3GP algorithm on several sets of satellite images over different countries to create hyper-feature from satellite bands to improve the classification of land cover types. We add the evolved hyper-features to the reference datasets and observe a significant improvement of the performance of three state-of-the-art ML algorithms (Decision Trees, Random Forests and XGBoost) on multiclass classifications and no significant effect on the binary classifications. We show that adding the M3GP hyper-features to the reference datasets brings better results than adding the well-known spectral indices NDVI, NDWI and NBR. We also compare the performance of the M3GP hyper-features in the binary classification problems with those created by other Feature Construction methods like FFX and EFS.