Harnessing Large-Scale Herbarium Image Datasets Through Representation Learning

Long-term trends in global flowering phenology

Flowering phenology is an indicator of the impact of climate change on natural systems. Anthropogenic climate change has progressed over more than two centuries, but ecological studies are mostly short in comparison. Here we harness the large-scale digitization of herbaria specimens to investigate temporal trends in flowering phenology at a global scale. We trained a convolutional neural network model to classify images of angiosperm herbarium specimens as being in flower or not in flower. This model was used to infer flowering across 8 million specimens spanning a century and global scales. We investigated temporal trends in mean flowering date and flowering season duration within ecoregions. We found high diversity of temporal trends in flowering seasonality across ecoregions with a median absolute shift of 2.5 d per decade in flowering date and 1.4 d per decade in flowering season duration. Variability in temporal trends in phenology was higher at low latitudes than at high latitudes. Our study demonstrates the value of digitized herbarium specimens for understanding natural dynamics in a time of change. The higher variability in phenological trends at low latitudes likely reflects the effects of a combination of shifts in temperature and precipitation seasonality, together with lower photoperiodic constraints to flowering.

Opportunities and Challenges in Applying AI to Evolutionary Morphology

Artificial intelligence (AI) is poised to revolutionize many aspects of science, including the study of evolutionary morphology. While classical AI methods such as principal component analysis and cluster analysis have been commonplace in the study of evolutionary morphology for decades, recent years have seen increasing application of deep learning to ecology and evolutionary biology. As digitized specimen databases become increasingly prevalent and openly available, AI is offering vast new potential to circumvent long-standing barriers to rapid, big data analysis of phenotypes. Here, we review the current state of AI methods available for the study of evolutionary morphology, which are most developed in the area of data acquisition and processing. We introduce the main available AI techniques, categorizing them into 3 stages based on their order of appearance: (1) machine learning, (2) deep learning, and (3) the most recent advancements in large-scale models and multimodal learning. Next, we present case studies of existing approaches using AI for evolutionary morphology, including image capture and segmentation, feature recognition, morphometrics, and phylogenetics. We then discuss the prospectus for near-term advances in specific areas of inquiry within this field, including the potential of new AI methods that have not yet been applied to the study of morphological evolution. In particular, we note key areas where AI remains underutilized and could be used to enhance studies of evolutionary morphology. This combination of current methods and potential developments has the capacity to transform the evolutionary analysis of the organismal phenotype into evolutionary phenomics, leading to an era of "big data" that aligns the study of phenotypes with genomics and other areas of bioinformatics.

Botanical Collection Patterns and Conservation Categories of the Most Traded Timber Species from the Ecuadorian Amazon: The Role of Protected Areas

The Ecuadorian Amazon is home to a rich biodiversity of woody plant species. Nonetheless, their conservation remains difficult, as some areas remain poorly explored and lack georeferenced records. Therefore, the current study aims predominantly to analyze the collection patterns of timber species in the Amazon lowlands of Ecuador and to evaluate the conservation coverage of these species in protected areas. Furthermore, we try to determine the conservation category of the species according to the criteria of the IUCN Red List. We identified that one third of the timber species in the study area was concentrated in three provinces due to historical botanical expeditions. However, a worrying 22.0% of the species had less than five records of presence, and 29.9% had less than ten records, indicating a possible underestimation of their presence. In addition, almost half of the species evaluated were unprotected, exposing them to deforestation risks and threats. To improve knowledge and conservation of forest biodiversity in the Ecuadorian Amazon, it is recommended to perform new botanical samplings in little-explored areas and digitize data in national herbaria. It is critical to implement automated assessments of the conservation status of species with insufficient data. In addition, it is suggested to use species distribution models to identify optimal areas for forest restoration initiatives. Effective communication of results and collaboration between scientists, governments, and local communities are key to the protection and sustainable management of forest biodiversity in the Amazon region.

An Industrial-Grade Solution for Crop Disease Image Detection Tasks

Crop leaf diseases can reflect the current health status of the crop, and the rapid and automatic detection of field diseases has become one of the difficulties in the process of industrialization of agriculture. In the widespread application of various machine learning techniques, recognition time consumption and accuracy remain the main challenges in moving agriculture toward industrialization. This article proposes a novel network architecture called YOLO V5-CAcT to identify crop diseases. The fast and efficient lightweight YOLO V5 is chosen as the base network. Repeated Augmentation, FocalLoss, and SmoothBCE strategies improve the model robustness and combat the positive and negative sample ratio imbalance problem. Early Stopping is used to improve the convergence of the model. We use two technical routes of model pruning, knowledge distillation and memory activation parameter compression ActNN for model training and identification under different hardware conditions. Finally, we use simplified operators with INT8 quantization for further optimization and deployment in the deep learning inference platform NCNN to form an industrial-grade solution. In addition, some samples from the Plant Village and AI Challenger datasets were applied to build our dataset. The average recognition accuracy of 94.24% was achieved in images of 59 crop disease categories for 10 crop species, with an average inference time of 1.563 ms per sample and model size of only 2 MB, reducing the model size by 88% and the inference time by 72% compared with the original model, with significant performance advantages. Therefore, this study can provide a solid theoretical basis for solving the common problems in current agricultural disease image detection. At the same time, the advantages in terms of accuracy and computational cost can meet the needs of agricultural industrialization.