Varying Lighting Conditions Research Articles

Image patch-based deep learning approach for crop and weed recognition

Accurate classification of weed species in crop plants plays a crucial role in precision agriculture by enabling targeted treatment. Recent studies show that artificial intelligence deep learning (DL) models achieve promising solutions. However, several challenging issues, such as lack of adequate training data, inter-class similarity between weed species and intra-class dissimilarity between the images of the same weed species at different growth stages or for other reasons (e.g., variations in lighting conditions, image capturing mechanism, agricultural field environments) limit their performance. In this research, we propose an image based weed classification pipeline where a patch of the image is considered at a time to improve the performance. We first enhance the images using generative adversarial networks. The enhanced images are divided into overlapping patches, a subset of which are used for training the DL models. For selecting the most informative patches, we use the variance of Laplacian and the mean frequency of Fast Fourier Transforms. At test time, the model's outputs are fused using a weighted majority voting technique to infer the class label of an image. The proposed pipeline was evaluated using 10 state-of-the-art DL models on four publicly available crop weed datasets: DeepWeeds, Cotton weed, Corn weed, and Cotton Tomato weed. Our pipeline achieved significant performance improvements on all four datasets. DenseNet201 achieved the top performance with F1 scores of 98.49%, 99.83% and 100% on Deepweeds, Corn weed and Cotton Tomato weed datasets, respectively. The highest F1 score on the Cotton weed dataset was 98.96%, obtained by InceptionResNetV2. Moreover, the proposed pipeline addressed the issues of intra-class dissimilarity and inter-class similarity in the DeepWeeds dataset and more accurately classified the minority weed classes in the Cotton weed dataset. This performance indicates that the proposed pipeline can be used in farming applications.

Unleashing the Power of VGG16: Advancements in Facial Emotion Recognization

In facial emotion detection, researchers are actively exploring effective methods to identify and understand facial expressions. This study introduces a novel mechanism for emotion identification using diverse facial photos captured under varying lighting conditions. A meticulously pre-processed dataset ensures data consistency and quality. Leveraging deep learning architectures, the study utilizes feature extraction techniques to capture subtle emotive cues and build an emotion classification model using convolutional neural networks (CNNs). The proposed methodology achieves an impressive 97% accuracy on the validation set, outperforming previous methods in terms of accuracy and robustness. Challenges such as lighting variations, head posture, and occlusions are acknowledged, and multimodal approaches incorporating additional modalities like auditory or physiological data are suggested for further improvement. The outcomes of this research have wide-ranging implications for affective computing, human-computer interaction, and mental health diagnosis, advancing the field of facial emotion identification and paving the way for sophisticated technology capable of understanding and responding to human emotions across diverse domains.

MACHINE LEARNING-BASED FACIAL RECOGNITION FOR VIDEO SURVEILLANCE SYSTEMS

Video surveillance systems play a crucial role in ensuring public safety and security. However, the traditional methods of surveillance often fall short in effectively identifying individuals, particularly in crowded or dynamic environments. This research addresses the limitations of conventional video surveillance by proposing a machine learning-based facial recognition system. The increasing demand for robust security measures necessitates the development of advanced technologies in video surveillance. Facial recognition has emerged as a promising solution, but existing systems struggle with accuracy and efficiency. This research aims to bridge these gaps by leveraging machine learning techniques for facial recognition in video surveillance. Conventional video surveillance struggles with accurate and rapid identification of individuals, leading to potential security lapses. This research addresses the challenge of enhancing facial recognition accuracy in real-time video feeds, especially in scenarios with varying lighting conditions and occlusions. While facial recognition has gained traction, there is a significant research gap in the implementation of machine learning algorithms tailored for video surveillance. This study aims to fill this void by proposing a novel methodology that combines deep learning and computer vision techniques for robust facial recognition in dynamic environments. The proposed methodology involves training a deep neural network on a diverse dataset of facial images to enable the model to learn intricate facial features. Additionally, computer vision algorithms will be employed to handle challenges such as occlusions and varying lighting conditions. The model''''s performance will be evaluated using real-world video surveillance data. Preliminary results demonstrate a significant improvement in facial recognition accuracy compared to traditional methods. The machine learning-based system exhibits enhanced performance in challenging scenarios, showcasing its potential for practical implementation in video surveillance systems.

Effect of Blue LED Light on Bioemulsifier Production in Bioreactor by Aureobasidium pullulans LB83 in Solid State Fermentation

This study analyzed the impact of LED light on bioemulsifier production by Aureobasidium pullulans LB83 in solid-state fermentation (SSF) using pre-treated sugarcane bagasse (PSB). The biomass was subjected to alkaline pre-treatment and conducted fermentations in Erlenmeyer flasks containing 2 g of PSB that were immersed in a humectant solution with a cell concentration of 108 cells/mL. The screening involved varying LED light wavelengths (green, red, orange, and blue) over a 7-day period at 28 °C. Notably, under the influence of blue light, the process achieved maximum production, yielding an EI24% of 63.9% and 45.1% for soybean oil and kerosene, respectively. Prolonged exposure to blue light for 11 days at 28 °C resulted in maximum bioemulsifier production (75%) and cellulolytic enzyme activity (3.67 IU g−1 for endoglucanase and 0.41 IU g−1 for exoglucanase) with soybean oil and kerosene. Experiments in a bioreactor, with varying light conditions (dark, white light, and blue LED light), demonstrated that the blue LED bioreactor outperformed others, achieving EI24% values of 55.0% and 45.7% for soybean oil and kerosene, respectively. The scanning electron microscopy (SEM) confirmed yeast growth under these conditions after 9 days. Our findings highlight the significant potential of LED light to enhance bioemulsifier production by A. pullulans LB83 from PSB.

MSER-Vertical Sobel for Vehicle Logo Detection

Detecting a vehicle logo is the first step before realizing the identity of the logo. However, the detection of logos can pose difficulties due to various factors, including logo variations, differing scales and orientations, background interference, varying lighting conditions, and partial obstruction. This paper presents a vehicle logo detection method using hand-crafted features. We used a combination of Maximally Stable Extremal Region (MSER) and Vertical Sobel. We combine vertical Sobel with MSER to overcome MSER's limitation in recognizing objects of different sizes. These two features are merged using a closing morphology operation to form blobs selected as logo candidate areas. Moreover, a Support Vector Machine (SVM) is implemented to choose a logo area by analyzing each candidate's Histogram of Oriented Gradient (HOG). The proposed method was compared with other methods by implementing them on the same dataset. The significant advantage of using MSER-Vertical Sobel is its fast computation time. It is faster than other approaches that use non-handcrafted features. The test results show that the MSER-Vertical Sobel can achieve high accuracy and the fastest computation time.

Addressing the sample volume dependency of the colorimetric glucose measurement on microfluidic paper-based and thread/paper-based analytical devices using a novel low-cost analytical viewpoint

The colorimetric method is widely exploited for glucose measurement on microfluidic paper-based and microfluidic thread/paper-based analytical devices, μPADs and μTPADs, respectively. However, two significant challenges still hinder the real-world applications: the variation of the generated color based on sample volume variation, and the consistent dependency on an imaging camera - usually a smartphone camera - for color readout. The latter also suffers from multiple limitations, such as the variable ambient light conditions and imaging variation among different smartphone brands and models. This manuscript suggests a new device demonstrating a novel viewpoint to the colorimetric method to address both aforementioned challenges. The presented device, Volume-Independent Autonomous box (VIA box), continuously monitors the transient color of the µPAD upon the introduction of the sample till color stabilization, in contrast to similar studies where merely the final stabilized color is read. The interpretation of the transient color profile with respect to time results in the measurement of the glucose level independent of sample volume. In addition, the VIA box monitors µPAD’s color using an ordinary RGB sensor instead of an imaging camera in similar studies. Since merely RGB values are recorded in any measurement instance, the computational cost is extremely low compared to the relevant literature where image processing techniques are used. VIA box exploits a simple low-end microcontroller to continuously monitor the transient color, interpret the color profile, and show the glucose level. The samples used in this manuscript are made of artificial sweat samples, including the known glucose range in real sweat samples.

SolarKey: Battery-free Key Generation Using Solar Cells

Solar cells have been widely used for offering energy for Internet of Things (IoT) devices. Recently, solar cells have also been used as sensors for context awareness sensing due to their sensitivity to varying lighting conditions. In this article, we are the first to use solar cells for symmetric key generation. To generate symmetric keys, we take advantage of photovoltage measurements generated from solar cells equipped with a pair of IoT devices. Symmetric keys are essential for pairing IoT devices and further securing wireless communication. Despite the sensitivity to varying lighting conditions, challenges still remain for the use of solar cells for key generation, such as time unsynchronisation and noisy measurements. To solve these challenges, we design a novel key generation framework, SolarKey, which includes the starting point detection and a compressed sensing-based two-tier key reconciliation method. Extensive experiments have been conducted to evaluate the performance of our proposed key generation method in various environments, which shows the proposed method can improve the key matching rate by up to 25%. We also conduct security analysis and the randomness test, which shows that SolarKey is resilient to common attacks such as the eavesdropping attack and the imitating attack and sufficiently random.

A new deep CNN for 3D text localization in the wild through shadow removal

Text localization in the wild is challenging due to the presence of 2D and 3D texts, the presence of shadows, arbitrary orientated text with non-linear arrangements, varying lighting conditions as well as complex background. This paper proposes the first approach for 3D text localization in natural scene images through shadow removal and a new deep CNN model. In a first step, exploiting the observation that 3D text generates shadow information in natural scenes, the proposed model detects and removes the shadow pixels of 3D text based on the Generalized Gradient Vector Flow concept and a new clustering approach. The performance of the classification of 2D and 3D texts in the scene images is strengthened by using key features, including pixel strength, sharpness and edge potential, which are extracted to eliminate false text and shadow pixels. For text localization after removing shadow information, EfficientNet is used as an encoder (backbone) and UNet as a decoder in a novel way employing differential binarization. Experimental validation and comparative analysis with state-of-the-art approaches on both a new purpose-built dataset as well as on the benchmark datasets of ICDAR MLT 2019, ICDAR ArT 2019, CTW1500, DAST1500, Total-Text, and MSRATD500 for each of the different steps of the method, show that the proposed approach outperforms the existing methods.

Sooty Mold Detection on Citrus Tree Canopy Using Deep Learning Algorithms.

Sooty mold is a common disease found in citrus plants and is characterized by black fungi growth on fruits, leaves, and branches. This mold reduces the plant's ability to carry out photosynthesis. In small leaves, it is very difficult to detect sooty mold at the early stages. Deep learning-based image recognition techniques have the potential to identify and diagnose pest damage and diseases such as sooty mold. Recent studies used advanced and expensive hyperspectral or multispectral cameras attached to UAVs to examine the canopy of the plants and mid-range cameras to capture close-up infected leaf images. To bridge the gap on capturing canopy level images using affordable camera sensors, this study used a low-cost home surveillance camera to monitor and detect sooty mold infection on citrus canopy combined with deep learning algorithms. To overcome the challenges posed by varying light conditions, the main reason for using specialized cameras, images were collected at night, utilizing the camera's built-in night vision feature. A total of 4200 sliced night-captured images were used for training, 200 for validation, and 100 for testing, employed on the YOLOv5m, YOLOv7, and CenterNet models for comparison. The results showed that YOLOv7 was the most accurate in detecting sooty molds at night, with 74.4% mAP compared to YOLOv5m (72%) and CenterNet (70.3%). The models were also tested using preprocessed (unsliced) night images and day-captured sliced and unsliced images. The testing on preprocessed (unsliced) night images demonstrated the same trend as the training results, with YOLOv7 performing best compared to YOLOv5m and CenterNet. In contrast, testing on the day-captured images had underwhelming outcomes for both sliced and unsliced images. In general, YOLOv7 performed best in detecting sooty mold infections at night on citrus canopy and showed promising potential in real-time orchard disease monitoring and detection. Moreover, this study demonstrated that utilizing a cost-effective surveillance camera and deep learning algorithms can accurately detect sooty molds at night, enabling growers to effectively monitor and identify occurrences of the disease at the canopy level.

Cross-domain car detection model with integrated convolutional block attention mechanism

Car detection, especially through camera vision, has become a major focus in the field of computer vision and has gained widespread adoption. While current car detection systems are capable of achieving good detection performance, reliable detection can still be challenging due to factors such as car proximity, varying light conditions, and environmental visibility. To address these issues, we propose Cross-Domain Car Detection Model with integrated convolutional block Attention mechanism(CDCDMA) that is specifically designed for car recognition in autonomous driving and related domains. CDCDMA includes several novelties: 1)Building a complete cross-domain target detection framework. 2)Developing an unpaired target domain picture generation module with an integrated convolutional attention mechanism which specifically emphasizes the car headlights feature. 3)Adopting Generalized Intersection over Union (GIOU) as the loss function of the target detection framework. 4)Designing an object detection model integrated with two-headed Convolutional Block Attention Module(CBAM). To evaluate the model's effectiveness, we performed experiments on the SODA 10 M and BDD100K datasets by applying a reduced resolution process to the data, which served as our benchmark dataset for the task. The experimental results demonstrate that the performance of the cross-domain car target detection model improves by 40% compared to the model without our CDCDMA framework. Moreover, our improvements have a significant impact on cross-domain car recognition, surpassing the performance of most advanced cross-domain models.

Validating and Developing Hyperspectral Indices for Tracing Leaf Chlorophyll Fluorescence Parameters under Varying Light Conditions

Chlorophyll a fluorescence (ChlFa) parameters provide insight into the physiological and biochemical processes of plants and have been widely applied to monitor and evaluate the photochemical process and photosynthetic capacity of plants in a variety of environments. Recent advances in remote sensing provide new opportunities for the detection of ChlFa at large scales but demand further tremendous efforts. Among such efforts, application of the hyperspectral index is always possible, but the performance of hyperspectral indices in detecting ChlFa parameters under varying light conditions is much less investigated. The objective of this study is to investigate the performance of reported hyperspectral indices for tracking ChlFa parameters under different light conditions and to develop and evaluate novel spectral indices. Therefore, an experiment was conducted to simultaneously measure ChlFa parameters and spectral reflectance of sunlit and shaded leaves under varying light conditions, and 28 reported hyperspectral indices were examined for their performance in tracking the ChlFa parameters. Furthermore, we developed novel hyperspectral indices based on various spectral transformations. The results indicated that the maximum quantum efficiency of photosystem II (PSIImax), the cumulative quantum yield of photochemistry (ΦP), and the fraction of open reaction centers in photosystem II (qL) of sunlit leaves were significantly higher than those of shaded leaves, while the cumulative quantum yield of regulated thermal dissipation (ΦN) and fluorescence (ΦF) of shaded leaves was higher than that of sunlit leaves. Efficient tracing of ChlFa parameters could not be achieved from previously published spectral indices. In comparison, all ChlFa parameters were well quantified in shaded leaves when using novel hyperspectral indices, although the hyperspectral indices for tracing the non-photochemical quenching (NPQ) and ΦF were not stable, especially for sunlit leaves. Our findings justify the use of hyperspectral indices as a practical approach to estimating ChlFa parameters. However, caution should be used when using spectral indices to track ChlFa parameters based on the differences in sunlit and shaded leaves.

Misaligned RGB-Infrared Object Detection via Adaptive Dual-Discrepancy Calibration

Object detection based on RGB and infrared images has emerged as a crucial research area in computer vision, and the synergy of RGB-Infrared ensures the robustness of object-detection algorithms under varying lighting conditions. However, the RGB-IR image pairs captured typically exhibit spatial misalignment due to sensor discrepancies, leading to compromised localization performance. Furthermore, since the inconsistent distribution of deep features from the two modalities, directly fusing multi-modal features will weaken the feature difference between the object and the background, therefore interfering with the RGB-Infrared object-detection performance. To address these issues, we propose an adaptive dual-discrepancy calibration network (ADCNet) for misaligned RGB-Infrared object detection, including spatial discrepancy and domain-discrepancy calibration. Specifically, the spatial discrepancy calibration module conducts an adaptive affine transformation to achieve spatial alignment of features. Then, the domain-discrepancy calibration module separately aligns object and background features from different modalities, making the distribution of the object and background of the fusion feature easier to distinguish, therefore enhancing the effectiveness of RGB-Infrared object detection. Our ADCNet outperforms the baseline by 3.3% and 2.5% in mAP50 on the FLIR and misaligned M3FD datasets, respectively. Experimental results demonstrate the superiorities of our proposed method over the state-of-the-art approaches.

Identification and evaluation of suitable reference genes for RT-qPCR analyses in Trichoderma atroviride under varying light conditions

BackgroundTrichoderma atroviride is a competitive soil-borne mycoparasitic fungus with extensive applications as a biocontrol agent in plant protection. Despite its importance and application potential, reference genes for RT-qPCR analysis in T. atroviride have not been evaluated. Light exerts profound effects on physiology, such as growth, conidiation, secondary metabolism, and stress response in T. atroviride, as well as in other fungi. In this study, we aimed to address this gap by identifying stable reference genes for RT-qPCR experiments in T. atroviride under different light conditions, thereby enhancing accurate and reliable gene expression analysis in this model mycoparasite. We measured and compared candidate reference genes using commonly applied statistical algorithms.ResultsUnder cyclic light–dark cultivation conditions, tbp and rho were identified as the most stably expressed genes, while act1, fis1, btl, and sar1 were found to be the least stable. Similar stability rankings were obtained for cultures grown under complete darkness, with tef1 and vma1 emerging as the most stable genes and act1, rho, fis1, and btl as the least stable genes. Combining the data from both cultivation conditions, gapdh and vma1 were identified as the most stable reference genes, while sar1 and fis1 were the least stable. The selection of different reference genes had a significant impact on the calculation of relative gene expression, as demonstrated by the expression patterns of target genes pks4 and lox1.ConclusionThe data emphasize the importance of validating reference genes for different cultivation conditions in fungi to ensure accurate interpretation of gene expression data.

The Design of a Video Reflection Removal Method Based on Illumination Compensation and Image Completion Fusion

Our objective is to develop a video reflection removal algorithm that is both easy to compute and effective. Unlike previous methods that depend on machine learning, our approach proposes a local image reflection removal technique that combines image completion and lighting compensation. To achieve this, we utilized the MSER image region feature point matching method to reduce image processing time and the spatial area of layer separation regions. In order to improve the adaptability of our method, we implemented a local image reflection removal technique that utilizes image completion and lighting compensation to interpolate layers and update motion field data in real-time. Our approach is both simple and efficient, allowing us to quickly obtain reflection-free video sequences under a variety of lighting conditions. This enabled us to achieve real-time detection effects through video restoration. This experiment has confirmed the efficacy of our method and demonstrated its comparable performance to advanced methods.

A Self-Occlusion Aware Lighting Model for Real-Time Dynamic Reconstruction.

In real-time dynamic reconstruction, geometry and motion are the major focuses while appearance is not fully explored, leading to the low-quality appearance of the reconstructed surfaces. In this article, we propose a lightweight lighting model that considers spatially varying lighting conditions caused by self-occlusion. This model estimates per-vertex masks on top of a single Spherical Harmonic (SH) lighting to represent spatially varying lighting conditions without adding too much computation cost. The mask is estimated based on the local geometry of a vertex to model the self-occlusion effect, which is the major reason leading to the spatial variation of lighting. Furthermore, to use this model in dynamic reconstruction, we also improve the motion estimation quality by adding a real-time per-vertex displacement estimation step. Experiments demonstrate that both the reconstructed appearance and the motion are largely improved compared with the current state-of-the-art techniques.

Improving the False Alarm Capability of the Extended Maximum Average Correlation Height Filter

The extended maximum average correlation height (EMACH) filter is a potent pattern-detection tool used in image processing and computer vision applications. This filter enhances the effectiveness of the maximum average correlation height (MACH) filter by adding more features and flexibility. Incorporating the benefits of wavelet decomposition, we updated the EMACH filter to enhance its performance. The updated filter offered improved accuracy, robustness, and flexibility in recognizing complex patterns and objects in images with varying lighting conditions, noise levels, and occlusions. To verify the results’ consistency and compare their performance with that of the MACH filter and EMACH filter, performance metrics like peak-to-correlation energy, peak-to-sidelobe ratio, signal-to-noise ratio, and discrimination ratio were computed. Through numerical and experimental studies, we found that the proposed filter enhances the identification rate and decreases the number of false alarms.

Neural Radiation Fields in a Tidal Flat Environment

Tidal flats are critical ecosystems, playing a vital role in biodiversity conservation and ecological balance. Collecting tidal flat environmental information using unmanned aerial vehicles (UAVs) and subsequently utilizing 3D reconstruction techniques for their detection and protection holds significance in providing comprehensive and detailed tidal flat information, including terrain, slope, and other parameters. It also enables scientific decision-making for the preservation of tidal flat ecosystems and the monitoring of factors such as rising sea levels. Moreover, the latest advancements in neural radiance fields (Nerf) have provided valuable insights and novel perspectives for our work. We face the following challenges: (1) the performance of a single network is limited due to the vast area to cover; (2) regions far from the camera center may exhibit suboptimal rendering results; and (3) changes in lighting conditions present challenges for the achievement of precise reconstruction. To tackle these challenges, we partitioned the tidal flat scene into distinct submodules, carefully preserving overlapping regions between each submodule for collaborative optimization. The luminance of each image is quantified by the appearance embedding vector produced by every captured image. Subsequently, this corresponding vector serves as an input to the model, enhancing its performance across varying lighting conditions. We also introduce an ellipsoidal sphere transformation that brings distant image elements into the sphere’s interior, enhancing the algorithm’s capacity to represent remote image information. Our algorithm is validated using tidal plane images collected from UAVs and compared with traditional Nerf based on two metrics: peak signal-to-noise ratio (PSNR) and learned perceptual image patch similarity (LPIPS). Our method enhances the PSNR value by 2.28 and reduces the LPIPS value by 0.11. The results further demonstrate that our approach significantly enhances Nerf’s performance in tidal flat environments. Utilizing Nerf for the 3D reconstruction of tidal flats, we bypass the need for explicit representation and geometric priors. This innovative approach yields superior novel view synthesis and enhances geometric perception, resulting in high-quality reconstructions. Our method not only provides valuable data but also offers profound insights for environmental monitoring and management.

6IMPOSE: bridging the reality gap in 6D pose estimation for robotic grasping.

6D pose recognition has been a crucial factor in the success of robotic grasping, and recent deep learning based approaches have achieved remarkable results on benchmarks. However, their generalization capabilities in real-world applications remain unclear. To overcome this gap, we introduce 6IMPOSE, a novel framework for sim-to-real data generation and 6D pose estimation. 6IMPOSE consists of four modules: First, a data generation pipeline that employs the 3D software suite Blender to create synthetic RGBD image datasets with 6D pose annotations. Second, an annotated RGBD dataset of five household objects was generated using the proposed pipeline. Third, a real-time two-stage 6D pose estimation approach that integrates the object detector YOLO-V4 and a streamlined, real-time version of the 6D pose estimation algorithm PVN3D optimized for time-sensitive robotics applications. Fourth, a codebase designed to facilitate the integration of the vision system into a robotic grasping experiment. Our approach demonstrates the efficient generation of large amounts of photo-realistic RGBD images and the successful transfer of the trained inference model to robotic grasping experiments, achieving an overall success rate of 87% in grasping five different household objects from cluttered backgrounds under varying lighting conditions. This is made possible by fine-tuning data generation and domain randomization techniques and optimizing the inference pipeline, overcoming the generalization and performance shortcomings of the original PVN3D algorithm. Finally, we make the code, synthetic dataset, and all the pre-trained models available on GitHub.

Functional trait divergence associated with heteromorphic leaves in a climbing fig.

Plants that display heteroblasty possess conspicuous variations in leaf morphology between their juvenile and adult phases, with certain species retaining juvenile-like leaves even in adulthood. Nevertheless, the ecological advantages of maintaining two or more distinct leaf types in heteroblastic plants at the adult stage remain unclear. The aim of this study is to examine the adaptive significance of heteroblastic leaves sampled from branches with divergent functions (sterile and fertile branches) of mature Ficus pumila individuals by comparing their morphological, anatomical, and physiological characteristics. Leaves on sterile branches (LSs) exhibited a significantly larger specific leaf area, thinner palisade and spongy tissues, lower chlorophyll contents, and lower light saturation points than leaves on fertile branches (LFs). These results demonstrate that LSs are better adapted to low light environments, while LFs are well equipped to take advantages of high light conditions. However, both LFs and LSs have a low light compensation point with no significant difference between them, indicating that they start to accumulate photosynthetic products under similar light conditions. Interestingly, significant higher net photosynthetic rate was detected in LFs, showing they have higher photosynthetic capacity. Furthermore, LFs produced significant more nutrients compared to LSs, which may associate to their ability of accumulating more photosynthetic products under full light conditions and higher photosynthetic capacity. Overall, we observed a pattern of divergence in morphological features of leaves on two functional branches. Anatomical and physiological features indicate that LFs have an advantage in varied light conditions, providing amounts of photosynthetic products to support the sexual reproduction, while LSs adapt to low light environments. Our findings provide evidence that heteroblasty facilitates F. pumila to utilize varying light environments, likely associated with its growth form as a climbing plant. This strategy allows the plant to allocate resources more effectively and optimize its overall fitness.

Canopy tree mortality depends on the proportion of crown exposed to sunlight, but this effect varies with species' wood density

AbstractUnderstanding what drives changes in tree mortality as well as the covariates influencing trees' response is a research priority to predict forest responses to global change. Here, we combined drone photogrammetry and ground‐based data to assess the influence of crown exposure to light (relative to total crown area), growth deviations (relative to conspecifics), tree size, and species' wood density (as a surrogate for light‐demanding and shade‐tolerant life‐history strategies) on the mortality of 984 canopy trees in an Amazon terra firme forest. Trees with lower wood density were less prone to die when their proportion of crown was more exposed to sunlight, but this relationship with relative crown exposure weakened and slightly reversed as wood density increased. Trees growing less than their species average had higher mortality, especially when the species' wood density decreased. The role of wood density in determining the survival of canopy trees under varying light conditions indicates differential responses of light‐demanding versus shade‐tolerant species. Our results highlight the importance of accounting for life‐history strategies, via plant functional types, in vegetation dynamic models aiming to predict forest demography under a rapidly changing climate.Abstract in Spanish is available with online material.