Homography-based ground plane detection using a single on-board camera

This study presents a robust method for ground plane detection in vision-based systems with a non-stationary camera. The proposed method is based on the reliable estimation of the homography between ground planes in successive images. This homography is computed using a feature matching approach, which in contrast to classical approaches to on-board motion estimation does not require explicit ego-motion calculation. As opposed to it, a novel homography calculation method based on a linear estimation framework is presented. This framework provides predictions of the ground plane transformation matrix that are dynamically updated with new measurements. The method is specially suited for challenging environments, in particular traffic scenarios, in which the information is scarce and the homography computed from the images is usually inaccurate or erroneous. The proposed estimation framework is able to remove erroneous measurements and to correct those that are inaccurate, hence producing a reliable homography estimate at each instant. It is based on the evaluation of the difference between the predicted and the observed transformations, measured according to the spectral norm of the associated matrix of differences. Moreover, an example is provided on how to use the information extracted from ground plane estimation to achieve object detection and tracking. The method has been successfully demonstrated for the detection of moving vehicles in traffic environments.

Use of camera trapping in determining Iberian lynx population parameters: The use area and its limitations

Below are the results of the survey of the Iberian lynx obtained with camera-trapping between 2000 and 2007 in Sierra Morena. Two very important aspects of camera-trapping concerning its efficiency are also analyzed. The first is the evolution along years according to the camera-trapping type used of two efficiency indicators. The results obtained demonstrate that the most efficient lure is rabbit, though it is the less proven (92 trap-nights), followed by camera-trapping in the most frequent marking places (latrines). And, we propose as a novel the concept of use area as a spatial reference unit for the camera-trapping monitoring of non radio-marked animals is proposed, and its validity discussed.

Vision-based vehicle detection and tracking with a mobile camera using a statistical framework

En esta tesis se aborda la detección y el seguimiento automático de vehículos mediante técnicas de visión artificial con una cámara monocular embarcada. Este problema ha suscitado un gran interés por parte de la industria automovilística y de la comunidad científica ya que supone el primer paso en aras de la ayuda a la conducción, la prevención de accidentes y, en última instancia, la conducción automática. A pesar de que se le ha dedicado mucho esfuerzo en los últimos años, de momento no se ha encontrado ninguna solución completamente satisfactoria y por lo tanto continúa siendo un tema de investigación abierto. Los principales problemas que plantean la detección y seguimiento mediante visión artificial son la gran variabilidad entre vehículos, un fondo que cambia dinámicamente debido al movimiento de la cámara, y la necesidad de operar en tiempo real. En este contexto, esta tesis propone un marco unificado para la detección y seguimiento de vehículos que afronta los problemas descritos mediante un enfoque estadístico. El marco se compone de tres grandes bloques, i.e., generación de hipótesis, verificación de hipótesis, y seguimiento de vehículos, que se llevan a cabo de manera secuencial. No obstante, se potencia el intercambio de información entre los diferentes bloques con objeto de obtener el máximo grado posible de adaptación a cambios en el entorno y de reducir el coste computacional. Para abordar la primera tarea de generación de hipótesis, se proponen dos métodos complementarios basados respectivamente en el análisis de la apariencia y la geometría de la escena. Para ello resulta especialmente interesante el uso de un dominio transformado en el que se elimina la perspectiva de la imagen original, puesto que este dominio permite una búsqueda rápida dentro de la imagen y por tanto una generación eficiente de hipótesis de localización de los vehículos. Los candidatos finales se obtienen por medio de un marco colaborativo entre el dominio original y el dominio transformado. Para la verificación de hipótesis se adopta un método de aprendizaje supervisado. Así, se evalúan algunos de los métodos de extracción de características más populares y se proponen nuevos descriptores con arreglo al conocimiento de la apariencia de los vehículos. Para evaluar la efectividad en la tarea de clasificación de estos descriptores, y dado que no existen bases de datos públicas que se adapten al problema descrito, se ha generado una nueva base de datos sobre la que se han realizado pruebas masivas. Finalmente, se presenta una metodología para la fusión de los diferentes clasificadores y se plantea una discusión sobre las combinaciones que ofrecen los mejores resultados. El núcleo del marco propuesto está constituido por un método Bayesiano de seguimiento basado en filtros de partículas. Se plantean contribuciones en los tres elementos fundamentales de estos filtros: el algoritmo de inferencia, el modelo dinámico y el modelo de observación. En concreto, se propone el uso de un método de muestreo basado en MCMC que evita el elevado coste computacional de los filtros de partículas tradicionales y por consiguiente permite que el modelado conjunto de múltiples vehículos sea computacionalmente viable. Por otra parte, el dominio transformado mencionado anteriormente permite la definición de un modelo dinámico de velocidad constante ya que se preserva el movimiento suave de los vehículos en autopistas. Por último, se propone un modelo de observación que integra diferentes características. En particular, además de la apariencia de los vehículos, el modelo tiene en cuenta también toda la información recibida de los bloques de procesamiento previos. El método propuesto se ejecuta en tiempo real en un ordenador de propósito general y da unos resultados sobresalientes en comparación con los métodos tradicionales. ABSTRACT This thesis addresses on-road vehicle detection and tracking with a monocular vision system. This problem has attracted the attention of the automotive industry and the research community as it is the first step for driver assistance and collision avoidance systems and for eventual autonomous driving. Although many effort has been devoted to address it in recent years, no satisfactory solution has yet been devised and thus it is an active research issue. The main challenges for vision-based vehicle detection and tracking are the high variability among vehicles, the dynamically changing background due to camera motion and the real-time processing requirement. In this thesis, a unified approach using statistical methods is presented for vehicle detection and tracking that tackles these issues. The approach is divided into three primary tasks, i.e., vehicle hypothesis generation, hypothesis verification, and vehicle tracking, which are performed sequentially. Nevertheless, the exchange of information between processing blocks is fostered so that the maximum degree of adaptation to changes in the environment can be achieved and the computational cost is alleviated. Two complementary strategies are proposed to address the first task, i.e., hypothesis generation, based respectively on appearance and geometry analysis. To this end, the use of a rectified domain in which the perspective is removed from the original image is especially interesting, as it allows for fast image scanning and coarse hypothesis generation. The final vehicle candidates are produced using a collaborative framework between the original and the rectified domains. A supervised classification strategy is adopted for the verification of the hypothesized vehicle locations. In particular, state-of-the-art methods for feature extraction are evaluated and new descriptors are proposed by exploiting the knowledge on vehicle appearance. Due to the lack of appropriate public databases, a new database is generated and the classification performance of the descriptors is extensively tested on it. Finally, a methodology for the fusion of the different classifiers is presented and the best combinations are discussed. The core of the proposed approach is a Bayesian tracking framework using particle filters. Contributions are made on its three key elements: the inference algorithm, the dynamic model and the observation model. In particular, the use of a Markov chain Monte Carlo method is proposed for sampling, which circumvents the exponential complexity increase of traditional particle filters thus making joint multiple vehicle tracking affordable. On the other hand, the aforementioned rectified domain allows for the definition of a constant-velocity dynamic model since it preserves the smooth motion of vehicles in highways. Finally, a multiple-cue observation model is proposed that not only accounts for vehicle appearance but also integrates the available information from the analysis in the previous blocks. The proposed approach is proven to run near real-time in a general purpose PC and to deliver outstanding results compared to traditional methods.

Simultaneous 3D object tracking and camera parameter estimation by Bayesian methods and transdimensional MCMC sampling

Multi-camera 3D tracking systems with overlapping cameras represent a powerful mean for scene analysis, as they potentially allow greater robustness than monocular systems and provide useful 3D information about object location and movement. However, their performance relies on accurately calibrated camera networks, which is not a realistic assumption in real surveillance environments. Here, we introduce a multi-camera system for tracking the 3D position of a varying number of objects and simultaneously refin-ing the calibration of the network of overlapping cameras. Therefore, we introduce a Bayesian framework that combines Particle Filtering for tracking with recursive Bayesian estimation methods by means of adapted transdimensional MCMC sampling. Addi-tionally, the system has been designed to work on simple motion detection masks, making it suitable for camera networks with low transmission capabilities. Tests show that our approach allows a successful performance even when starting from clearly inaccurate camera calibrations, which would ruin conventional approaches.

A method for Kinematic Calibration of a Parallel Robot by using one camera in hand and a spherical object

The main purpose of robot calibration is the correction of the possible errors in the robot parameters. This paper presents a method for a kinematic calibration of a parallel robot that is equipped with one camera in hand. In order to preserve the mechanical configuration of the robot, the camera is utilized to acquire incremental positions of the end effector from a spherical object that is fixed in the word reference frame. The positions of the end effector are related to incremental positions of resolvers of the motors of the robot, and a kinematic model of the robot is used to find a new group of parameters which minimizes errors in the kinematic equations. Additionally, properties of the spherical object and intrinsic camera parameters are utilized to model the projection of the object in the image and improving spatial measurements. Finally, the robotic system is designed to carry out tracking tasks and the calibration of the robot is validated by means of integrating the errors of the visual controller.

Scalable software architecture for on-line multi-camera video processing

In this paper we present a scalable software architecture for on-line multi-camera video processing, that guarantees a good trade off between computational power, scalability and flexibility. The software system is modular and its main blocks are the Processing Units (PUs), and the Central Unit. The Central Unit works as a supervisor of the running PUs and each PU manages the acquisition phase and the processing phase. Furthermore, an approach to easily parallelize the desired processing application has been presented. In this paper, as case study, we apply the proposed software architecture to a multi-camera system in order to efficiently manage multiple 2D object detection modules in a real-time scenario. System performance has been evaluated under different load conditions such as number of cameras and image sizes. The results show that the software architecture scales well with the number of camera and can easily works with different image formats respecting the real time constraints. Moreover, the parallelization approach can be used in order to speed up the processing tasks with a low level of overhead

Adaptive Multi-Camera System for Real Time Object Detection

In this paper we present an adaptive multi-camera system for real time object detection able to efficiently adjust the computational requirements of video processing blocks to the available processing power and the activity of the scene. The system is based on a two level adaptation strategy that works at local and at global level. Object detection is based on a Gaussian mixtures model background subtraction algorithm. Results show that the system can efficiently adapt the algorithm parameters without a significant loss in the detection accuracy.

Comparative Analysis of Tolerances Allocation in Mechanical Assemblies based on Cost-tolerance Curves

El presente trabajo propone un método para la determinación de los valores de las tolerancias individuales de las piezas que forman un conjunto ensamblado a partir de valores de tolerancias especificados en el conjunto final, optimizando el coste total de fabricación de las piezas individuales a partir de funciones de coste-tolerancia basadas en el proceso de fabricación de cada una de ellas. Para ello se parte de los principales trabajos desarrollados en la línea de asignación de tolerancias y se realiza la propuesta del modelo de trabajo, basado en la optimización de costes a partir de la aplicación del método de los multiplicadores de Lagrange a diversas curvas de coste-tolerancia

IDRA (IDeal Resource Allocation): Computing ideal speedups in parallel logic programming

We present a technique to estimate accurate speedups for parallel logic programs with relative independence from characteristics of a given implementation or underlying parallel hardware. The proposed technique is based on gathering accurate data describing one execution at run-time, which is fed to a simulator. Alternative schedulings are then simulated and estimates computed for the corresponding speedups. A tool implementing the aforementioned techniques is presented, and its predictions are compared to the performance of real systems, showing good correlation.

IDRA (IDeal Resource Allocation): A tool for computing ideal speedups

Performance studies of actual parallel systems usually tend to concéntrate on the effectiveness of a given implementation. This is often done in the absolute, without quantitave reference to the potential parallelism contained in the programs from the point of view of the execution paradigm. We feel that studying the parallelism inherent to the programs is interesting, as it gives information about the best possible behavior of any implementation and thus allows contrasting the results obtained. We propose a method for obtaining ideal speedups for programs through a combination of sequential or parallel execution and simulation, and the algorithms that allow implementing the method. Our approach is novel and, we argüe, more accurate than previously proposed methods, in that a crucial part of the data - the execution times of tasks - is obtained from actual executions, while speedup is computed by simulation. This allows obtaining speedup (and other) data under controlled and ideal assumptions regarding issues such as number of processor, scheduling algorithm and overheads, etc. The results obtained can be used for example to evalúate the ideal parallelism that a program contains for a given model of execution and to compare such "perfect" parallelism to that obtained by a given implementation of that model. We also present a tool, IDRA, which implements the proposed method, and results obtained with IDRA for benchmark programs, which are then compared with those obtained in actual executions on real parallel systems.

Allocation of 14C assimilated in late spring to tissue and biochemical stem components of cork oak (Quercus suber L.) over the seasons

Carbon distribution in the stem of 2-year-old cork oak plants was studied by 14CO2 pulse labeling in late spring in order to trace the allocation of photoassimilates to tissue and biochemical stem components of cork oak. The fate of 14C photoassimilated carbon was followed during two periods: the first 72 h (short-term study) and the first 52 weeks (long-term study) after the 14CO2 photosynthetic assimilation. The results showed that 14C allocation to stem tissues was dependent on the time passed since photoassimilation and on the season of the year. In the first 3 h all 14C was found in the polar extractives. After 3 h, it started to be allocated to other stem fractions. In 1 day, 14C was allocated mostly to vascular cambium and, to a lesser extent, to primary phloem; no presence of 14C was recorded for the periderm. However, translocation of 14C to phellem was observed from 1 week after 14CO2 pulse labeling. The phellogen was not completely active in its entire circumference at labeling, unlike the vascular cambium; this was the tissue that accumulated most photoassimilated 14C at the earliest sampling. The fraction of leaf-assimilated 14C that was used by the stem peaked at 57% 1 week after 14CO2 plant exposure. The time lag between C photoassimilation and suberin accumulation was ∼8 h, but the most active period for suberin accumulation was between 3 and 7 days. Suberin, which represented only 1.77% of the stem weight, acted as a highly effective sink for the carbon photoassimilated in late spring since suberin specific radioactivity was much higher than for any other stem component as early as only 1 week after 14C plant labeling. This trend was maintained throughout the whole experiment. The examination of microautoradiographs taken over 1 year provided a new method for quantifying xylem growth. Using this approach it was found that there was more secondary xylem growth in late spring than in other times of the year, because the calculated average cell division time was much shorter.

Risk Allocation in Toll Highway Concessions in Spain: Lessons from Economic Recession

Spain has a long tradition of encouraging toll highways by granting concessions to private companies. Concessions in Spain have been characterized by a willingness to transfer considerable risk to the private sector. Traffic demand, acquisition of the right-of-way, and financial risk have often been allocated to the private sector. From 1996 to 2011, 16 toll highway concessions, covering a total distance of 835 km, were awarded by the central government of Spain with this approach. Some of those highways started their operations just before the economic recession began. The recession had negative consequences for Spain's economy. The gross domestic product per capita plummeted, and the unemployment rate increased from 9% to 20% of the working population in just 2 years. The recession also had severe consequences for the economic performance of toll highway concessions. Traffic levels declined at a much greater rate than did the gross domestic product. In addition, the conditions imposed by the financial markets on borrowers became much stricter because of the liquidity crisis. This study analyzes the impact that the economic recession ultimately had on the performance of toll highway concessions in Spain and the actions that the government adopted to avoid the bankruptcy of the concessionaires. It was found that the economic recession helped identify some deficiencies in how risk had been allocated in Spain. The measures that both Spain and the European Union are adopting so as to improve risk allocation are discussed.

Distributed bees algorithm for task allocation in swarm of robots

In this paper, we propose the distributed bees algorithm (DBA) for task allocation in a swarm of robots. In the proposed scenario, task allocation consists in assigning the robots to the found targets in a 2-D arena. The expected distribution is obtained from the targets' qualities that are represented as scalar values. Decision-making mechanism is distributed and robots autonomously choose their assignments taking into account targets' qualities and distances. We tested the scalability of the proposed DBA algorithm in terms of number of robots and number of targets. For that, the experiments were performed in the simulator for various sets of parameters, including number of robots, number of targets, and targets' utilities. Control parameters inherent to DBA were tuned to test how they affect the final robot distribution. The simulation results show that by increasing the robot swarm size, the distribution error decreased.

Using clustering techniques for intelligent camera-based user interfaces

The area of Human-Machine Interface is growing fast due to its high importance in all technological systems. The basic idea behind designing human-machine interfaces is to enrich the communication with the technology in a natural and easy way. Gesture interfaces are a good example of transparent interfaces. Such interfaces must identify properly the action the user wants to perform, so the proper gesture recognition is of the highest importance. However, most of the systems based on gesture recognition use complex methods requiring high-resource devices. In this work, we propose to model gestures capturing their temporal properties, which significantly reduce storage requirements, and use clustering techniques, namely self-organizing maps and unsupervised genetic algorithm, for their classification. We further propose to train a certain number of algorithms with different parameters and combine their decision using majority voting in order to decrease the false positive rate. The main advantage of the approach is its simplicity, which enables the implementation using devices with limited resources, and therefore low cost. The testing results demonstrate its high potential.

Carbon allocation in north-western Amazon forests (Colombia)

Los estudios sobre la asignación del carbono en los ecosistemas forestales proporcionan información esencial para la comprensión de las diferencias espaciales y temporales en el ciclo del carbono de tal forma que pueden aportar información a los modelos y, así predecir las posibles respuestas de los bosques a los cambios en el clima. Dentro de este contexto, los bosques Amazónicos desempeñan un papel particularmente importante en el balance global del carbono; no obstante, existen grandes incertidumbres en cuanto a los controles abióticos en las tasas de la producción primaria neta (PPN), la asignación de los productos de la fotosíntesis a los diferentes componentes o compartimentos del ecosistema (aéreo y subterráneo) y, cómo estos componentes de la asignación del carbono responden a eventos climáticos extremos. El objetivo general de esta tesis es analizar los componentes de la asignación del carbono en bosques tropicales maduros sobre suelos contrastantes, que crecen bajo condiciones climáticas similares en dos sitios ubicados en la Amazonia noroccidental (Colombia): el Parque Natural Nacional Amacayacu y la Estación Biológica Zafire. Con este objetivo, realicé mediciones de los componentes de la asignación del carbono (biomasa, productividad primaria neta, y su fraccionamiento) a nivel ecosistémico y de la dinámica forestal (tasas anuales de mortalidad y reclutamiento), a lo largo de ocho años (20042012) en seis parcelas permanentes de 1 hectárea establecidas en cinco tipos de bosques sobre suelos diferentes (arcilloso, franco-arcilloso, franco-arcilloso-arenoso, franco-arenoso y arena-francosa). Toda esta información me permitió abordar preguntas específicas que detallo a continuación. En el Capítulo 2 evalúe la hipótesis de que a medida que aumenta la fertilidad del suelo disminuye la cantidad del carbono asignado a la producción subterránea (raíces finas con diámetro <2 mm). Y para esto, realicé mediciones de la masa y la producción de raíces finas usando dos métodos: (1) el de los cilindros de crecimiento y, (2) el de los cilindros de extracción secuencial. El monitoreo se realizó durante 2.2 años en los bosques con suelos más contrastantes: arcilla y arena-francosa. Encontré diferencias significativas en la masa de raíces finas y su producción entre los bosques y, también con respecto a la profundidad del suelo (010 y 1020 cm). El bosque sobre arena-francosa asignó más carbono a las raíces finas que el bosque sobre arcillas. La producción de raíces finas en el bosque sobre arena-francosa fue dos veces más alta (media ± error estándar = 2.98 ± 0.36 y 3.33 ± 0.69 Mg C ha1 año1, con el método 1 y 2, respectivamente), que para el bosque sobre arcillas, el suelo más fértil (1.51 ± 0.14, método 1, y desde 1.03 ± 0.31 a 1.36 ± 0.23 Mg C ha1 año1, método 2). Del mismo modo, el promedio de la masa de raíces finas fue tres veces mayor en el bosque sobre arena-francosa (5.47 ± 0.17 Mg C ha1) que en el suelo más fértil (de 1.52 ± 0.08 a 1.82 ± 0.09 Mg C ha1). La masa de las raíces finas también mostró un patrón temporal relacionado con la lluvia, mostrando que la producción de raíces finas disminuyó sustancialmente en el período seco del año 2005. Estos resultados sugieren que los recursos del suelo pueden desempeñar un papel importante en los patrones de la asignación del carbono entre los componentes aéreo y subterráneo de los bosques tropicales; y que el suelo no sólo influye en las diferencias en la masa de raíces finas y su producción, sino que también, en conjunto con la lluvia, sobre la estacionalidad de la producción. En el Capítulo 3 estimé y analicé los tres componentes de la asignación del carbono a nivel del ecosistema: la biomasa, la productividad primaria neta PPN, y su fraccionamiento, en los mismos bosques del Capítulo 2 (el bosque sobre arcillas y el bosque sobre arena-francosa). Encontré diferencias significativas en los patrones de la asignación del carbono entre los bosques; el bosque sobre arcillas presentó una mayor biomasa total y aérea, así como una PPN, que el bosque sobre arena-francosa. Sin embargo, la diferencia entre los dos bosques en términos de la productividad primaria neta total fue menor en comparación con las diferencias entre la biomasa total de los bosques, como consecuencia de las diferentes estrategias en la asignación del carbono a los componentes aéreo y subterráneo del bosque. La proporción o fracción de la PPN asignada a la nueva producción de follaje fue relativamente similar entre los dos bosques. Nuestros resultados de los incrementos de la biomasa aérea sugieren una posible compensación entre la asignación del carbono al crecimiento de las raíces finas versus el de la madera, a diferencia de la compensación comúnmente asumida entre la parte aérea y la subterránea en general. A pesar de estas diferencias entre los bosques en términos de los componentes de la asignación del carbono, el índice de área foliar fue relativamente similar entre ellos, lo que sugiere que el índice de área foliar es más un indicador de la PPN total que de la asignación de carbono entre componentes. En el Capítulo 4 evalué la variación espacial y temporal de los componentes de la asignación del carbono y la dinámica forestal de cinco tipos e bosques amazónicos y sus respuestas a fluctuaciones en la precipitación, lo cual es completamente relevante en el ciclo global del carbono y los procesos biogeoquímicos en general. Estas variaciones son así mismo importantes para evaluar los efectos de la sequía o eventos extremos sobre la dinámica natural de los bosques amazónicos. Evalué la variación interanual y la estacionalidad de los componentes de la asignación del carbono y la dinámica forestal durante el periodo 2004−2012, en cinco bosques maduros sobre diferentes suelos (arcilloso, franco-arcilloso, franco-arcilloso-arenoso, franco-arenoso y arena-francosa), todos bajo el mismo régimen local de precipitación en la Amazonia noroccidental (Colombia). Quería examinar sí estos bosques responden de forma similar a las fluctuaciones en la precipitación, tal y como pronostican muchos modelos. Consideré las siguientes preguntas: (i) ¿Existe una correlación entre los componentes de la asignación del carbono y la dinámica forestal con la precipitación? (ii) ¿Existe correlación entre los bosques? (iii) ¿Es el índice de área foliar (LAI) un indicador de las variaciones en la producción aérea o es un reflejo de los cambios en los patrones de la asignación del carbono entre bosques?. En general, la correlación entre los componentes aéreo y subterráneo de la asignación del carbono con la precipitación sugiere que los suelos juegan un papel importante en las diferencias espaciales y temporales de las respuestas de estos bosques a las variaciones en la precipitación. Por un lado, la mayoría de los bosques mostraron que los componentes aéreos de la asignación del carbono son susceptibles a las fluctuaciones en la precipitación; sin embargo, el bosque sobre arena-francosa solamente presentó correlación con la lluvia con el componente subterráneo (raíces finas). Por otra parte, a pesar de que el noroeste Amazónico es considerado sin una estación seca propiamente (definida como <100 mm meses −1), la hojarasca y la masa de raíces finas mostraron una alta variabilidad y estacionalidad, especialmente marcada durante la sequía del 2005. Además, los bosques del grupo de suelos francos mostraron que la hojarasca responde a retrasos en la precipitación, al igual que la masa de raíces finas del bosque sobre arena-francosa. En cuanto a la dinámica forestal, sólo la tasa de mortalidad del bosque sobre arena-francosa estuvo correlacionada con la precipitación (ρ = 0.77, P <0.1). La variabilidad interanual en los incrementos en el tallo y la biomasa de los individuos resalta la importancia de la mortalidad en la variación de los incrementos en la biomasa aérea. Sin embargo, las tasas de mortalidad y las proporciones de individuos muertos por categoría de muerte (en pie, caído de raíz, partido y desaparecido), no mostraron tendencias claras relacionadas con la sequía. Curiosamente, la hojarasca, el incremento en la biomasa aérea y las tasas de reclutamiento mostraron una alta correlación entre los bosques, en particular dentro del grupo de los bosques con suelos francos. Sin embargo, el índice de área foliar estimado para los bosques con suelos más contrastantes (arcilla y arena-francosa), no presentó correlación significativa con la lluvia; no obstante, estuvo muy correlacionado entre bosques; índice de área foliar no reflejó las diferencias en la asignación de los componentes del carbono, y su respuesta a la precipitación en estos bosques. Por último, los bosques estudiados muestran que el noroeste amazónico es susceptible a fenómenos climáticos, contrario a lo propuesto anteriormente debido a la ausencia de una estación seca propiamente dicha. ABSTRACT Studies of carbon allocation in forests provide essential information for understanding spatial and temporal differences in carbon cycling that can inform models and predict possible responses to changes in climate. Amazon forests play a particularly significant role in the global carbon balance, but there are still large uncertainties regarding abiotic controls on the rates of net primary production (NPP) and the allocation of photosynthetic products to different ecosystem components; and how the carbon allocation components of Amazon forests respond to extreme climate events. The overall objective of this thesis is to examine the carbon allocation components in old-growth tropical forests on contrasting soils, and under similar climatic conditions in two sites at the Amacayacu National Natural Park and the Zafire Biological Station, located in the north-western Amazon (Colombia). Measurements of above- and below-ground carbon allocation components (biomass, net primary production, and its partitioning) at the ecosystem level, and dynamics of tree mortality and recruitment were done along eight years (20042012) in six 1-ha plots established in five Amazon forest types on different soils (clay, clay-loam, sandy-clay-loam, sandy-loam and loamy-sand) to address specific questions detailed in the next paragraphs. In Chapter 2, I evaluated the hypothesis that as soil fertility increases the amount of carbon allocated to below-ground production (fine-roots) should decrease. To address this hypothesis the standing crop mass and production of fine-roots (<2 mm) were estimated by two methods: (1) ingrowth cores and, (2) sequential soil coring, during 2.2 years in the most contrasting forests: the clay-soil forest and the loamy-sand forest. We found that the standing crop fine-root mass and its production were significantly different between forests and also between soil depths (0–10 and 10–20 cm). The loamysand forest allocated more carbon to fine-roots than the clay-soil forest, with fine-root production in the loamy-sand forest twice (mean ± standard error = 2.98 ± 0.36 and 3.33 ± 0.69 Mg C ha −1 yr −1, method 1 and 2, respectively) as much as for the more fertile claysoil forest (1.51 ± 0.14, method 1, and from 1.03 ± 0.31 to 1.36 ± 0.23 Mg C ha −1 yr −1, method 2). Similarly, the average of standing crop fine-root mass was three times higher in the loamy-sand forest (5.47 ± 0.17 Mg C ha1) than in the more fertile soil (from 1.52 ± 0.08 a 1.82 ± 0.09 Mg C ha1). The standing crop fine-root mass also showed a temporal pattern related to rainfall, with the production of fine-roots decreasing substantially in the dry period of the year 2005. These results suggest that soil resources may play an important role in patterns of carbon allocation of below-ground components, not only driven the differences in the biomass and its production, but also in the time when it is produced. In Chapter 3, I assessed the three components of stand-level carbon allocation (biomass, NPP, and its partitioning) for the same forests evaluated in Chapter 2 (clay-soil forest and loamy-sand forest). We found differences in carbon allocation patterns between these two forests, showing that the forest on clay-soil had a higher aboveground and total biomass as well as a higher above-ground NPP than the loamy-sand forest. However, differences between the two types of forests in terms of stand-level NPP were smaller, as a consequence of different strategies in the carbon allocation of above- and below-ground components. The proportional allocation of NPP to new foliage production was relatively similar between the two forests. Our results of aboveground biomass increments and fine-root production suggest a possible trade-off between carbon allocation to fine-roots versus wood growth (as it has been reported by other authors), as opposed to the most commonly assumed trade-off between total above- and below-ground production. Despite these differences among forests in terms of carbon allocation components, the leaf area index showed differences between forests like total NPP, suggesting that the leaf area index is more indicative of total NPP than carbon allocation. In Chapter 4, I evaluated the spatial and temporal variation of carbon allocation components and forest dynamics of Amazon forests as well as their responses to climatic fluctuations. I evaluated the intra- and inter-annual variation of carbon allocation components and forest dynamics during the period 2004−2012 in five forests on different soils (clay, clay-loam, sandy-clay-loam, sandy-loam and loamy-sand), but growing under the same local precipitation regime in north-western Amazonia (Colombia). We were interested in examining if these forests respond similarly to rainfall fluctuations as many models predict, considering the following questions: (i) Is there a correlation in carbon allocation components and forest dynamics with precipitation? (ii) Is there a correlation among forests? (iii) Are temporal responses in leaf area index (LAI) indicative of variations of above-ground production or a reflection of changes in carbon allocation patterns among forests?. Overall, the correlation of above- and below-ground carbon allocation components with rainfall suggests that soils play an important role in the spatial and temporal differences of responses of these forests to rainfall fluctuations. On the one hand, most forests showed that the above-ground components are susceptible to rainfall fluctuations; however, there was a forest on loamy-sand that only showed a correlation with the below-ground component (fine-roots). On the other hand, despite the fact that north-western Amazonia is considered without a conspicuous dry season (defined as <100 mm month−1), litterfall and fine-root mass showed high seasonality and variability, particularly marked during the drought of 2005. Additionally, forests of the loam-soil group showed that litterfall respond to time-lags in rainfall as well as and the fine-root mass of the loamy-sand forest. With regard to forest dynamics, only the mortality rate of the loamy-sand forest was significantly correlated with rainfall (77%). The observed inter-annual variability of stem and biomass increments of individuals highlighted the importance of the mortality in the above-ground biomass increment. However, mortality rates and death type proportion did not show clear trends related to droughts. Interestingly, litterfall, above-ground biomass increment and recruitment rates of forests showed high correlation among forests, particularly within the loam-soil forests group. Nonetheless, LAI measured in the most contrasting forests (clay-soil and loamysand) was poorly correlated with rainfall but highly correlated between forests; LAI did not reflect the differences in the carbon allocation components, and their response to rainfall on these forests. Finally, the forests studied highlight that north-western Amazon forests are also susceptible to climate fluctuations, contrary to what has been proposed previously due to their lack of a pronounced dry season.