Abstract: The transition from armed conflicts or oppressive regimes to contexts of peace and democracy requires a thorough investigation of what happened during the time of violence. This crucial responsibility of clarifying a past period of abuses has been undertaken by truth commissions. However, the reconstruction of this truth cannot forget the inequalities women had to endure historically and the different impacts they suffer during wartime. In that sense, the truth-seeking work that truth commissions performed needs to incorporate a gender perspective, in order to fight the discrimination against women, and to include their point of view in the version of a country’s truth. Thus, two of the strongest truth commissions in the world have developed particular measures regarding women: Peru and East Timor. For that reason, this research identifies and analyse such actions. This with the aim of recognising the learnings of these experiences in gender issues and, based on them, build a proposal on how a truth commission should incorporate the gender perspective in its truth-seeking work.

Abstract: Chain elongation fermentation in open cultures systems can tolerate complex organic waste and convert it into useful biochemicals (commodities) such as caproic acid. In previous chain elongation research, fatty alcohols such as hexanol have been also documented. Hexanol has a potential application as biofuel, is less corrosive, watersoluble, and toxic than caproic acid. Moreover, the hexanol market size seems more attractive than the one of caproic acid. This research aims to produce hexanol via a chain elongation process in an upflow anaerobic sludge blanket (UASB) system. Granular sludge formation was also considered due to its advantages such as high biomass concentration and settleability. The highest selectivity (based on total carbon products in the effluent) and productivity of caproate was 82.49 % on day 35 and 10.25 g L-1 d -1 on day 33, whereas for hexanol was 1.66 % and 0.19 g L-1 d -1 , both on day 77. CO2 supply had a positive effect on chain elongation activity and hexanol formation. Selectivity and productivity of hexanol was not negatively affected despite methane formation. Throughout the whole study, suspended biomass was the dominant form. A thick biofilm layer on the inner wall of the reactor was present from day 19 to 76 and black aggregates became more visible from day 76 onwards, although not significantly. The process was severely inhibited by substrate (ethanol) toxicity during the last phase (36.77 ± 3.23 g L-1 ).

Abstract: The conditional autoregressive model (CAR model) is the most popular distribution for jointly modeling the \textit{a priori} uncertainty over spatially correlated data. In general, it is used in hierarchical spatial models where it models the uncertainty about random spatial effects. A limitation of the CAR model is its inability to produce high correlations between neighboring areas. We propose a robust model for area data that alleviates this problem. We represent the map by an undirected graph where nodes represent areas and edges connect neighboring nodes on the map. We assign distinct and random weights to the edges. The model is based on a spatially structured $t-$Student multivariate distribution, in which the precision matrix is indirectly constructed assuming a multivariate distribution for the random weights of the edges. Such t- Student distribution spatially correlates the edge weights and induces another $t$-Student model for the spatial effects of the areas that correlates them and is able to accommodate outliers and heavy tail behavior for these effects. More importantly, the proposed model can produce a higher marginal correlation between spatial effects than the CAR model, overcoming one of the main limitations of this model. We adjusted the proposed model to map the incidence of some types of cancer in southern Brazil and compared its performance with several alternative models proposed in the literature. The results show that the proposed model is competitive and provides similar and, in some cases, better results than those obtained by fitting models commonly used to analyze this type of data. In the second proposal, we approach the problem of dimensionality reduction in regression models. One of the most used methods to avoid overfitting and to select relevant variables in regression models with many predictors is the penalized regression technique. Under suchapproaches, variable selection is performed in a non-probabilistic way using some optimization criterion. Bayesian approaches to penalized regression have been proposed assuming an \textit{a priori} distribution for the regression coefficients that plays a role similar to the penalty term in classical statistics: compressing towards zero non-significant coefficients and putting a probability mass significant in coefficients that can be grouped. Generally, such a priori distributions, called shrinkage priors, assume independence between the effects of the covariates, which may not bean appropriate assumption in many cases. In this work, we focus on the dimensionality reduction of categorical variables with many levels. These variables are included in the model through variables dummy inducing sparsity in the design matrix, which can generate overfitting and difficulties in interpreting the results. The effect of the levels of these categorical variables are naturally correlated. To deal with this problem, we propose two a priori shrinkage distributions for the coefficients associated with the levels of categorical variables, correlating them. The proposed distributions are properand, in addition to sparsity, they have the property of grouping similar effects. We illustrate the use of these distributions by applying them to dimensionality reduction in a linear regression. Their performances are analyzed and compared to pre-existing methods through simulated data studies and considering housing price data available on Airbnb. Resumen: El modelo autorregresivo condicional (modelo CAR) es la distribución más popular para modelar conjuntamente la incertidumbre \ textit {apriori} sobre datos correlacionados espacialmente. En general, se utiliza en modelos espaciales jerárquicos donde modela la incertidumbre sobre los efectos espaciales aleatorios. Una limitación del modelo CAR es su incapacidad para producir altas correlaciones entre áreas vecinas. Proponemos un modelo robusto para datos de área que alivia este problema. Representamos el mapa mediante un gráfico no dirigido donde los nodos representan áreas y los bordes conectan los nodos vecinos en el mapa. Asignamos pesos distintos y aleatorios a los bordes. El modelo se basa en una distribución multivariante $ t- $ Student estructurada espacialmente, en la que la matriz de precisión se construye indirectamente asumiendo una distribución multivariante para los pesos aleatorios de los bordes. Tal distribución t-Student correlaciona espacialmente los pesos de los bordes e induce otro modelo $ t $ -Student para los efectos espaciales de las áreas que los correlaciona y es capaz de acomodar valores atípicos y un comportamiento de cola pesada para estos efectos. Más importante aún, el modelo propuesto puede producir una mayor correlación marginal entre los efectos espaciales que el modelo CAR, superando una de las principales limitaciones de este modelo. Ajustamos el modelo propuesto para mapear la incidencia de algunos tipos de cáncer en el sur de Brasil y comparamos su desempeño con varios modelos alternativos propuestos en la literatura. Los resultados muestran que el modelo propuesto es competitivo y proporciona resultados similares y, en algunos casos, mejores que los obtenidos mediante el ajuste de modelos comúnmente utilizados para analizar este tipo de datos. En la segunda propuesta abordamos el problema de la reducción de dimensionalidad en modelos de regresión. Uno de los métodos más utilizados para evitar el sobreajuste y seleccionar variables relevantes en modelos de regresión con muchos predictores es la técnica de regresión penalizada. Bajo tales enfoques, la selección de variables se realiza de forma no probabilística utilizando algún criterio de optimización. Se han propuesto enfoques bayesianos para la regresión penalizada asumiendo una distribución \ textit {a priori} para los coeficientes de regresión que juega un papel similar al término de penalización en la estadística clásica: comprimir hacia cero coeficientes no significativos y poner una masa de probabilidad significativa en coeficientes que se pueden agrupar. Generalmente, tales distribuciones a priori, llamadas a priori de contracción, asumen independencia entre los efectos de las covariables, lo que puede no ser una suposición apropiada en muchos casos. En este trabajo, nos enfocamos en la reducción de dimensionalidad de variables categóricas con muchos niveles. Estas variables se incluyen en el modelo a través de variables ficticias que inducen escasez en la matriz de diseño, lo que puede generar sobreajustes y dificultades en la interpretación de los resultados. El efecto de los niveles de estas variables categóricas está naturalmente correlacionado. Para abordar este problema, proponemos dos distribuciones de contracción a priori para los coeficientes asociados a los niveles de variables categóricas, correlacionándolos. Las distribuciones propuestas son adecuadas y, además de escasa, tienen la propiedad de agrupar efectos similares. Ilustramos el uso de estas distribuciones aplicándolas a la reducción de dimensionalidad en una regresión lineal. Sus resultados se analizan y comparan con métodos preexistentes a través de estudios de datos simulados y teniendo en cuenta los datos de precios de la vivienda disponibles en Airbnb.

Abstract: Because there is nothing intelligent to say about a massacre. Everybody is supposed to be dead, to never say anything or want anything ever again. Everything is supposed to be very quiet after a massacre, and it always is, except for the birds.” All wars are atrocious, and the task of incorporating such inglorious events into a nation’s collective identity becomes understandably hard. But it needs to be done, otherwise, the memory of the war might be trivialized and forgotten and, hence, are more prone to be repeated. War is bound to replicate itself without confrontation and awareness. Mosul city was established on the western bank of the Tigris River in the 6th Century, just across the water from the ruins of the Assyrian city of Nineveh. The city flourished during the ancient era establishing itself as an epicentre of the Islamic world, therefore it is rich with ancient ruins and cultural heritage. The war which erupted in Iraq didn’t aim to destroy only its defence, but also its identity. Entire urban blocks of Mosul’s city were damaged so severely, that when it comes to reconstruction it cannot simply restore the former fabrics, either of space or architecture. The war has changed everything, and most critically the structure of society itself. The spatial reconstruction will mostly require more than just physical rehabilitation and restoration of the cities devastated infrastructure and housing. In order to provide sensations for the need of diversity and unity, more urgent concern should be given to restoring public facilities and daily utilities. Architectural and landscape design can be judiciously used in healing the social structure. How can common spaces be rearranged to create conditions for a peaceful and creative coexistence among citizens which survived such horrors? How can the location and the design of such utilities be done in a way as to meet the needs for intimacy and peace? This thesis study intends to recommend an urban and architectural regeneration project, to be taken upon during the redevelopment of the old city of Mosul. Furthermore, a set of strategies that may ensure the preservation of the urban identity of the area, while redeveloping and coping with the environmental, socioeconomic and political changes have been identified for reclaiming the urban identity and promoting the sense of belonging.

Abstract: The adequate provision of service delivery is essential for social welfare; however, providers performance varies, tending to underperform in developing countries. Therefore, whether providers responsiveness is sensitive to citizen oversight remains a relevant question for public governance. To this extent, Understanding the dynamics of accountability as a strategy to foster access, quality, and affordability of services delivery is vital to ensure development. This thesis establishes a quantitative assessment of the relationship between accountability and development. First, I systematically reviewed the empirical literature evaluating how accountability interventions affect service delivery in developing countries between 1990-2020, finding twenty-one papers from nine developing countries. Then, I conducted a meta-analysis based on an RVE model with over 90 outcomes to quantitatively synthesise the effectiveness of accountability in triggering citizen oversight of service delivery and providers responsiveness. The findings suggest that although accountability increases citizen knowledge of civic and political means to demand service delivery reforms, factors such as context, accountability structure and interventions length influence its effectiveness. While the results provide evidence to inform policy-making about aspects that effectively ignite citizen participation, they also suggest future research avenues to further conceptualise this relationship in an era of widely accessible information and new approaches to democracy.

Abstract: Floods are disastrous natural severe weather events that cause large damage worldwide each year, inducing loss of human lives, destruction of infrastructure and economical losses. Consequently, forecasting this type of events through hydrological modelling is of great importance from a civil protection point of view since it allows institutions to reduce the generated hydrological risk by means of early warning systems. Nevertheless, ungauged basins where there is lack of direct measurements of meteorological information to force the models is one encountered problematic affecting the forecasts. In the present study is evaluated the possibility of using meteorological predictions coming from MOLOCH model to force the FEST-WB hydrological model to perform discharge forecasting. This, under the hypothesis that the committed error of the prediction is negligible when using forecasts up to 24 h. The study is done in a well-known area such as the Seveso-Olona-Lambro river basins located in northern Italy. Thus, the main hydro-meteorological variables are analysed by carrying out a comparison between spatialized observed meteorological data coming from ARPA and meteonetwork weather stations and meteorological predictions. Moreover, a sensitivity analysis following the one-factor-at-time methodology is accomplished with the aim of defining which forcing mostly affects flowrate forecasts. It is shown that discreet correspondence of information with an underestimation trend – particularly for large values - is verified for discharge at hourly and daily scale, and that underestimation of precipitation - especially in summer - and overestimation of solar radiation are the main reasons of this. Hence, to improve the predictability when coupling the two models, some actions should be evaluated to enhance the correspondence of the meteorological forcings. Resumen: Las inundaciones son desastres naturales que causan una gran cantidad de daño alrededor del mundo cada año, induciendo pérdidas de vidas humanas, destrucción de infraestructura y pérdidas económicas. En consecuencia, la predicción de este tipo de eventos a través de modelos hidrológicos es de suma importancia desde el punto de vista de la protección civil ya que permite a las instituciones reducir el riesgo hidrológico creado, por medio de sistemas de alerta temprana. Sin embargo, cuencas hidrográficas sin instrumentación en las que existe una falta de mediciones directas de información meteorológica para inicializar los modelos, es una problemática encontrada que afecta a las predicciones. En el presente estudio se evalúa la posibilidad de usar predicciones meteorológicas obtenidas con el modelo MOLOCH para inicializar el modelo hidrológico FEST-WB y así realizar predicciones de caudal. Lo anterior se realiza hipotetizando que el error cometido en la predicción de caudal resulta ser insignificante cuando se utilizan predicciones meteorológicas hasta las primeras 24 h. El estudio se realiza en un área ampliamente estudiada como la cuenca de los ríos Seveso-Olona-Lambro localizados en el norte de Italia. De este modo, las principales variables hidro-meteorológicas son analizadas llevando a cabo una comparación entre información meteorológica observada y espacializada, obtenida a partir de las estaciones climáticas de ARPA y meteonetwork, y las predicciones meteorológicas. Adicionalmente, se efectúa un análisis de sensibilidad con la metodología “un factor a la vez” (one-factor-at-time) con el objetivo de definir cuál de las forzantes afecta mayoritariamente las predicciones de caudal. Se evidencia que para los caudales a escalas horaria y diaria existe una correspondencia de información discreta con una tendencia a la subestimación, particularmente para valores elevados, y que, la subestimación de la precipitación (especialmente en verano) y la sobreestimación de la radiación solar son las razones principales de este comportamiento. Por consiguiente, para mejorar la predictibilidad cuando se acoplan los dos modelos mencionados, acciones adicionales deben ser evaluadas para realzar la correspondencia de las forzantes meteorológicas.

Abstract: Floods are disastrous natural severe weather events that cause large damage worldwide each year, inducing loss of human lives, destruction of infrastructure and economical losses. Consequently, forecasting this type of events through hydrological modelling is of great importance from a civil protection point of view since it allows institutions to reduce the generated hydrological risk by means of early warning systems. Nevertheless, ungauged basins where there is lack of direct measurements of meteorological information to force the models is one encountered problematic affecting the forecasts. In the present study is evaluated the possibility of using meteorological predictions coming from MOLOCH model to force the FEST-WB hydrological model to perform discharge forecasting. This, under the hypothesis that the committed error of the prediction is negligible when using forecasts up to 24 h. The study is done in a well-known area such as the Seveso-Olona-Lambro river basins located in northern Italy. Thus, the main hydro-meteorological variables are analysed by carrying out a comparison between spatialized observed meteorological data coming from ARPA and meteonetwork weather stations and meteorological predictions. Moreover, a sensitivity analysis following the one-factor-at-time methodology is accomplished with the aim of defining which forcing mostly affects flowrate forecasts. It is shown that discreet correspondence of information with an underestimation trend – particularly for large values - is verified for discharge at hourly and daily scale, and that underestimation of precipitation - especially in summer - and overestimation of solar radiation are the main reasons of this. Hence, to improve the predictability when coupling the two models, some actions should be evaluated to enhance the correspondence of the meteorological forcings. Resumen: Las inundaciones son desastres naturales que causan una gran cantidad de daño alrededor del mundo cada año, induciendo pérdidas de vidas humanas, destrucción de infraestructura y pérdidas económicas. En consecuencia, la predicción de este tipo de eventos a través de modelos hidrológicos es de suma importancia desde el punto de vista de la protección civil ya que permite a las instituciones reducir el riesgo hidrológico creado, por medio de sistemas de alerta temprana. Sin embargo, cuencas hidrográficas sin instrumentación en las que existe una falta de mediciones directas de información meteorológica para inicializar los modelos, es una problemática encontrada que afecta a las predicciones. En el presente estudio se evalúa la posibilidad de usar predicciones meteorológicas obtenidas con el modelo MOLOCH para inicializar el modelo hidrológico FEST-WB y así realizar predicciones de caudal. Lo anterior se realiza hipotetizando que el error cometido en la predicción de caudal resulta ser insignificante cuando se utilizan predicciones meteorológicas hasta las primeras 24 h. El estudio se realiza en un área ampliamente estudiada como la cuenca de los ríos Seveso-Olona-Lambro localizados en el norte de Italia. De este modo, las principales variables hidro-meteorológicas son analizadas llevando a cabo una comparación entre información meteorológica observada y espacializada, obtenida a partir de las estaciones climáticas de ARPA y meteonetwork, y las predicciones meteorológicas. Adicionalmente, se efectúa un análisis de sensibilidad con la metodología “un factor a la vez” (one-factor-at-time) con el objetivo de definir cuál de las forzantes afecta mayoritariamente las predicciones de caudal. Se evidencia que para los caudales a escalas horaria y diaria existe una correspondencia de información discreta con una tendencia a la subestimación, particularmente para valores elevados, y que, la subestimación de la precipitación (especialmente en verano) y la sobreestimación de la radiación solar son las razones principales de este comportamiento. Por consiguiente, para mejorar la predictibilidad cuando se acoplan los dos modelos mencionados, acciones adicionales deben ser evaluadas para realzar la correspondencia de las forzantes meteorológicas.

Abstract: Water is considered a cross-cutting resource for all the ecosystem services (ES) types, namely: for the provision of drinking water and food; regulation through flood control; guarantee of a suitable habitat for fauna and flora; and inspiration for many cultures and their intangible heritage around the planet. The evaluation of water ecosystem services has drawn much attention both in the political arena and the scientific community. Still, there is not a consensus in the approach of evaluating them. Recent notions suggest that the study of ES cannot be assessed from a single discourse but through comprehensive frameworks. Therefore, the main objective of this thesis was to evaluate the water provision as an ecosystem service of the Conservation Area of Guanacaste (ACG, for its acronym in Spanish) by adapting an Ecosystem Services (ES)-based approach with four core elements: (1) effects on human well-being, (2) bio-physical underpinning of service delivery, (3) transdisciplinarity, and (4) assessment of services for decision-making. I have deepened in core element two by assessing all the physically-based processes involved in the hydrologic cycle, which generates the quantity of water available for the study region. As a result, here I present the first calibrated and validated Soil and Water Assessment Tool (SWAT) for the entire ACG, including the land use and cover change (LUCC) impact on water availability. I tested seven different climatic and geospatial information resources to establish the best input data to build the SWAT model. The model showed excellent performance for both the calibration period (r2 = 0.81, NSE = 0.75, and br2 = 0.80) and validation period (r2 = 0.85, NSE = 0.82, and br2 = 0.85). I have found that the ACG has a high potential for the provision of water as an ecosystem service. I also identified the spatial and temporal patterns and changes in the water availability in this region between 1980 - 2019. I deepened in the vegetation-water dynamic understanding by evaluating the land use cover change (LUCC), finding that the recovery of forests, and decrease in grasslands and agricultural territories have a positive effect on the water availability also associated with exceptional events in Costa Rica’s history that put in place the conditions for the recovery of natural resources. I also demonstrated the advantages of using spatially explicit and physically-based models, such as SWAT, to quantify water provision as an ecosystem service. Finally, this work complements the baseline for future applications of ES-based approaches in this region.

Abstract: To ensure a sustainable future, climate and biodiversity emergencies must be addressed together. Photovoltaic energy has emerged in the EU as a very attractive alternative to conventional energy production to achieve the objectives of the United Nations 2030 Agenda. However, clean energy sources can also have significant negative impacts on the environment and conflict with other interests in the territory, such as habitat fragmentation and contamination, loss of productive soil and pressure on agricultural uses. We have created and evaluated two contrasting conflict scenarios between the efficiency of photovoltaic production, agricultural uses and biodiversity by using a sensitivity analysis of hotspots and coldspots of conflict. Our results showed that a low efficiency scenario was the one that reduced the incidence of conflicts the most, but also the one that required slightly larger areas that are also more expensive to reach an exploitation target. We consider that the conflict analysis provides a guarantee of the areas suitable for photovoltaic projects in scenarios of low conflict with land uses and conservation interests and it contributes to the sustainable development of society, since it identifies the conflicts generated by the most important productive activities such as the provision of basic services as energy and food. We emphasize this work does not indicate all the areas of opportunities for the installation of photovoltaics, since this work focused on analyzing two extreme scenarios, it is necessary to evaluate new scenarios considering the complexity of the trade-offs.

Resumen: La crisis mundial causada por la pandemia puso sobre la mesa cuestiones acerca de lo que es esencial y no para nosotros como especie humana. Encuestas realizadas durante la pandemia evidenciaban que si bien el consumo de producciones culturales había aumentado a través de plataformas digitales, las personas catalogaban como no esencial el trabajo del artista. Si bien el arte y la cultura estaban más presentes que nunca en el centro de nuestras vidas aun cuando todo había parado, una especie de disonancia en la percepción de la gente se veía reflejada en las encuestas. Con un título inspirado en las canciones más escuchadas en Spotify durante los primeros meses de la pandemia, (Dance Monkey de Tones and I y The Box de Roddy Rich); Monos danzantes en una caja, alude a lo que fuimos durante el confinamiento: Monos danzantes encerrados en cuatro paredes, y hace un recorrido por el papel del arte en tiempos de crisis. Cuando consideramos el artista como una profesión no esencial, la pregunta que surge es, ¿que entendemos por arte? Acaso ¿quiénes componían las canciones que escuchábamos en Spotify, los guiones o las bandas sonoras de las películas que veíamos en Netflix? ¿Qué factores sociales, económicos, políticos, tecnológicos y sociales influencian nuestra percepción y como damos valor al arte y la cultura? Dentro de la civilización occidental (y esto incluye a una Latinoamérica altamente influenciada por la tradición cultural europea) cultura es entendida entre la tensión del concepto antropológico (cultura como forma de vida), y cultura equiparada al arte: el conocimiento de una serie de objetos culturales privilegiados, un sistema creado en el siglo XVIII ligado a disciplinas como historia del arte, estética y mercado. Este ha sido el discurso dominante durante siglos que ha conceptualizado el arte como algo elitista producido por genios y asequible solo a unos cuantos. A esto se contrapone la visión de la antropología y el psicoanálisis que nos muestran como el arte es una dimensión fundamental en el ser humano. La separación del arte en una esfera aparte de la vida cotidiana, hace además que constantemente se tenga que abogar por el valor de este. En Reino Unido anualmente se realizan decenas de investigaciones acerca del valor del arte y la cultura en un esfuerzo por instrumentalizarlos y para justificar la adjudicación presupuestal. El arte entonces se utiliza para satisfacer intereses externos generalmente políticos como bienestar de la comunidad, diplomacia cultural (soft power), creación de identidad nacional o como una herramienta de transformación social. Otros de los factores que han influenciado nuestra relación con el arte, es por un lado la apropiación de la cultura por parte del capitalismo, quien la utiliza para transmitir sus ideologías y por otro lado la forma en la que consumimos, en donde las producciones culturales son simplemente un medio para sumar clicks y reproducciones, permitiendo que mucho del contenido se vuelva “gratis”, e invisibilizando y trayendo a menos el trabajo de los creadores y artistas. (Recordemos que el verdadero producto es el consumidor y su atención). Después de exponer este panorama general, hago un llamado a volver a entender el arte desde su esencia particular como vector fundamental en el ser humano. ¿Que pasaría en un mundo sin arte? Geertz nos da luces: “No ocurriría nada especialmente destacable (...) De hecho, la sociedad no se desmoronaría; algunas de las cosas que experimentaron simplemente no podrían expresadas -y quizás, después de algún tiempo, tampoco podrían ser experimentadas y, por tanto, la vida sería más gris”. El arte nos devuelve la humanidad cuando solo somos monos danzantes en una caja, experimentando la vida a través de luces enceguecedoras. (pantallas de ordenadores, teléfonos móviles, televisores e Ipads).